--- /dev/null
+# Altivec instruction set, for PSIM, the PowerPC simulator.
+
+# Copyright 2003 Free Software Foundation, Inc.
+
+# Contributed by Red Hat Inc; developed under contract from Motorola.
+# Written by matthew green <mrg@redhat.com>.
+
+# This file is part of GDB.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 59 Temple Place - Suite 330,
+# Boston, MA 02111-1307, USA. */
+
+
+#
+# Motorola AltiVec instructions.
+#
+
+:cache:av:::VS:VS:
+:cache:av::vreg *:vS:VS:(cpu_registers(processor)->altivec.vr + VS)
+:cache:av::unsigned32:VS_BITMASK:VS:(1 << VS)
+:cache:av:::VA:VA:
+:cache:av::vreg *:vA:VA:(cpu_registers(processor)->altivec.vr + VA)
+:cache:av::unsigned32:VA_BITMASK:VA:(1 << VA)
+:cache:av:::VB:VB:
+:cache:av::vreg *:vB:VB:(cpu_registers(processor)->altivec.vr + VB)
+:cache:av::unsigned32:VB_BITMASK:VB:(1 << VB)
+:cache:av:::VC:VC:
+:cache:av::vreg *:vC:VC:(cpu_registers(processor)->altivec.vr + VC)
+:cache:av::unsigned32:VC_BITMASK:VC:(1 << VC)
+
+# Flags for model.h
+::model-macro:::
+ #define PPC_INSN_INT_VR(OUT_MASK, IN_MASK, OUT_VMASK, IN_VMASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_int_vr(MY_INDEX, cpu_model(processor), OUT_MASK, IN_MASK, OUT_VMASK, IN_VMASK); \
+ } while (0)
+
+ #define PPC_INSN_VR(OUT_VMASK, IN_VMASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_vr(MY_INDEX, cpu_model(processor), OUT_VMASK, IN_VMASK); \
+ } while (0)
+
+ #define PPC_INSN_VR_CR(OUT_VMASK, IN_VMASK, CR_MASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_vr_cr(MY_INDEX, cpu_model(processor), OUT_VMASK, IN_VMASK, CR_MASK); \
+ } while (0)
+
+ #define PPC_INSN_VR_VSCR(OUT_VMASK, IN_VMASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_vr_vscr(MY_INDEX, cpu_model(processor), OUT_VMASK, IN_VMASK); \
+ } while (0)
+
+ #define PPC_INSN_FROM_VSCR(VR_MASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_from_vscr(MY_INDEX, cpu_model(processor), VR_MASK); \
+ } while (0)
+
+ #define PPC_INSN_TO_VSCR(VR_MASK) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_to_vscr(MY_INDEX, cpu_model(processor), VR_MASK); \
+ } while (0)
+
+# Trace waiting for AltiVec registers to become available
+void::model-static::model_trace_altivec_busy_p:model_data *model_ptr, unsigned32 vr_busy
+ int i;
+ if (vr_busy) {
+ vr_busy &= model_ptr->vr_busy;
+ for(i = 0; i < 32; i++) {
+ if (((1 << i) & vr_busy) != 0) {
+ TRACE(trace_model, ("Waiting for register v%d.\n", i));
+ }
+ }
+ }
+ if (model_ptr->vscr_busy)
+ TRACE(trace_model, ("Waiting for VSCR\n"));
+
+# Trace making AltiVec registers busy
+void::model-static::model_trace_altivec_make_busy:model_data *model_ptr, unsigned32 vr_mask, unsigned32 cr_mask
+ int i;
+ if (vr_mask) {
+ for(i = 0; i < 32; i++) {
+ if (((1 << i) & vr_mask) != 0) {
+ TRACE(trace_model, ("Register v%d is now busy.\n", i));
+ }
+ }
+ }
+ if (cr_mask) {
+ for(i = 0; i < 8; i++) {
+ if (((1 << i) & cr_mask) != 0) {
+ TRACE(trace_model, ("Register cr%d is now busy.\n", i));
+ }
+ }
+ }
+
+# Schedule an AltiVec instruction that takes integer input registers and produces output registers
+void::model-function::ppc_insn_int_vr:itable_index index, model_data *model_ptr, const unsigned32 out_mask, const unsigned32 in_mask, const unsigned32 out_vmask, const unsigned32 in_vmask
+ const unsigned32 int_mask = out_mask | in_mask;
+ const unsigned32 vr_mask = out_vmask | in_vmask;
+ model_busy *busy_ptr;
+
+ if ((model_ptr->int_busy & int_mask) != 0 || (model_ptr->vr_busy & vr_mask)) {
+ model_new_cycle(model_ptr); /* don't count first dependency as a stall */
+
+ while ((model_ptr->int_busy & int_mask) != 0 || (model_ptr->vr_busy & vr_mask)) {
+ if (WITH_TRACE && ppc_trace[trace_model]) {
+ model_trace_busy_p(model_ptr, int_mask, 0, 0, PPC_NO_SPR);
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+ }
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ }
+
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ model_ptr->int_busy |= out_mask;
+ busy_ptr->int_busy |= out_mask;
+ model_ptr->vr_busy |= out_vmask;
+ busy_ptr->vr_busy |= out_vmask;
+
+ if (out_mask)
+ busy_ptr->nr_writebacks = (PPC_ONE_BIT_SET_P(out_vmask)) ? 1 : 2;
+
+ if (out_vmask)
+ busy_ptr->nr_writebacks += (PPC_ONE_BIT_SET_P(out_vmask)) ? 1 : 2;
+
+ if (WITH_TRACE && ppc_trace[trace_model]) {
+ model_trace_make_busy(model_ptr, out_mask, 0, 0);
+ model_trace_altivec_make_busy(model_ptr, vr_mask, 0);
+ }
+
+# Schedule an AltiVec instruction that takes vector input registers and produces vector output registers
+void::model-function::ppc_insn_vr:itable_index index, model_data *model_ptr, const unsigned32 out_vmask, const unsigned32 in_vmask
+ const unsigned32 vr_mask = out_vmask | in_vmask;
+ model_busy *busy_ptr;
+
+ if (model_ptr->vr_busy & vr_mask) {
+ model_new_cycle(model_ptr); /* don't count first dependency as a stall */
+
+ while (model_ptr->vr_busy & vr_mask) {
+ if (WITH_TRACE && ppc_trace[trace_model]) {
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+ }
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ }
+
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ model_ptr->vr_busy |= out_vmask;
+ busy_ptr->vr_busy |= out_vmask;
+ if (out_vmask)
+ busy_ptr->nr_writebacks = (PPC_ONE_BIT_SET_P(out_vmask)) ? 1 : 2;
+
+ if (WITH_TRACE && ppc_trace[trace_model]) {
+ model_trace_altivec_make_busy(model_ptr, vr_mask, 0);
+ }
+
+# Schedule an AltiVec instruction that takes vector input registers and produces vector output registers, touches CR
+void::model-function::ppc_insn_vr_cr:itable_index index, model_data *model_ptr, const unsigned32 out_vmask, const unsigned32 in_vmask, const unsigned32 cr_mask
+ const unsigned32 vr_mask = out_vmask | in_vmask;
+ model_busy *busy_ptr;
+
+ if ((model_ptr->vr_busy & vr_mask) || (model_ptr->cr_fpscr_busy & cr_mask)) {
+ model_new_cycle(model_ptr); /* don't count first dependency as a stall */
+
+ while ((model_ptr->vr_busy & vr_mask) || (model_ptr->cr_fpscr_busy & cr_mask)) {
+ if (WITH_TRACE && ppc_trace[trace_model]) {
+ model_trace_busy_p(model_ptr, 0, 0, cr_mask, PPC_NO_SPR);
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+ }
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ }
+
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ model_ptr->cr_fpscr_busy |= cr_mask;
+ busy_ptr->cr_fpscr_busy |= cr_mask;
+ model_ptr->vr_busy |= out_vmask;
+ busy_ptr->vr_busy |= out_vmask;
+
+ if (out_vmask)
+ busy_ptr->nr_writebacks = (PPC_ONE_BIT_SET_P(out_vmask)) ? 1 : 2;
+
+ if (cr_mask)
+ busy_ptr->nr_writebacks++;
+
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_make_busy(model_ptr, vr_mask, cr_mask);
+
+# Schedule an AltiVec instruction that takes vector input registers and produces vector output registers, touches VSCR
+void::model-function::ppc_insn_vr_vscr:itable_index index, model_data *model_ptr, const unsigned32 out_vmask, const unsigned32 in_vmask
+ const unsigned32 vr_mask = out_vmask | in_vmask;
+ model_busy *busy_ptr;
+
+ if ((model_ptr->vr_busy & vr_mask) != 0 || model_ptr->vscr_busy != 0) {
+ model_new_cycle(model_ptr); /* don't count first dependency as a stall */
+
+ while ((model_ptr->vr_busy & vr_mask) != 0 || model_ptr->vscr_busy != 0) {
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ }
+
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ model_ptr->vr_busy |= out_vmask;
+ busy_ptr->vr_busy |= out_vmask;
+ model_ptr->vscr_busy = 1;
+ busy_ptr->vscr_busy = 1;
+
+ if (out_vmask)
+ busy_ptr->nr_writebacks = 1 + (PPC_ONE_BIT_SET_P(out_vmask)) ? 1 : 2;
+
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_make_busy(model_ptr, vr_mask, 0);
+
+# Schedule an MFVSCR instruction that VSCR input register and produces an AltiVec output register
+void::model-function::ppc_insn_from_vscr:itable_index index, model_data *model_ptr, const unsigned32 vr_mask
+ model_busy *busy_ptr;
+
+ while ((model_ptr->vr_busy & vr_mask) != 0 || model_ptr->vscr_busy != 0) {
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ model_ptr->cr_fpscr_busy |= vr_mask;
+ busy_ptr->cr_fpscr_busy |= vr_mask;
+
+ if (vr_mask)
+ busy_ptr->nr_writebacks = 1;
+
+ model_ptr->vr_busy |= vr_mask;
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_make_busy(model_ptr, vr_mask, 0);
+
+# Schedule an MTVSCR instruction that one AltiVec input register and produces a vscr output register
+void::model-function::ppc_insn_to_vscr:itable_index index, model_data *model_ptr, const unsigned32 vr_mask
+ model_busy *busy_ptr;
+
+ while ((model_ptr->vr_busy & vr_mask) != 0 || model_ptr->vscr_busy != 0) {
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_altivec_busy_p(model_ptr, vr_mask);
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ busy_ptr ->vscr_busy = 1;
+ model_ptr->vscr_busy = 1;
+ busy_ptr->nr_writebacks = 1;
+
+ TRACE(trace_model,("Making VSCR busy.\n"));
+
+# The follow are AltiVec saturate operations
+
+signed8::model-function::altivec_signed_saturate_8:signed16 val, int *sat
+ signed8 rv;
+ if (val > 127) {
+ rv = 127;
+ *sat = 1;
+ } else if (val < -128) {
+ rv = -128;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+signed16::model-function::altivec_signed_saturate_16:signed32 val, int *sat
+ signed16 rv;
+ if (val > 32767) {
+ rv = 32767;
+ *sat = 1;
+ } else if (val < -32768) {
+ rv = -32768;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+signed32::model-function::altivec_signed_saturate_32:signed64 val, int *sat
+ signed32 rv;
+ if (val > 2147483647) {
+ rv = 2147483647;
+ *sat = 1;
+ } else if (val < -2147483648LL) {
+ rv = -2147483648LL;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+unsigned8::model-function::altivec_unsigned_saturate_8:signed16 val, int *sat
+ unsigned8 rv;
+ if (val > 255) {
+ rv = 255;
+ *sat = 1;
+ } else if (val < 0) {
+ rv = 0;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+unsigned16::model-function::altivec_unsigned_saturate_16:signed32 val, int *sat
+ unsigned16 rv;
+ if (val > 65535) {
+ rv = 65535;
+ *sat = 1;
+ } else if (val < 0) {
+ rv = 0;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+unsigned32::model-function::altivec_unsigned_saturate_32:signed64 val, int *sat
+ unsigned32 rv;
+ if (val > 4294967295LL) {
+ rv = 4294967295LL;
+ *sat = 1;
+ } else if (val < 0) {
+ rv = 0;
+ *sat = 1;
+ } else {
+ rv = val;
+ *sat = 0;
+ }
+ return rv;
+
+#
+# Load instructions, 6-14 ... 6-22.
+#
+
+0.31,6.VS,11.RA,16.RB,21.7,31.0:X:av:lvebx %VD, %RA, %RB:Load Vector Element Byte Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ eb = EA & 0xf;
+ (*vS).b[AV_BINDEX(eb)] = MEM(unsigned, EA, 1);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.39,31.0:X:av:lvehx %VD, %RA, %RB:Load Vector Element Half Word Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~1;
+ eb = EA & 0xf;
+ (*vS).h[AV_HINDEX(eb/2)] = MEM(unsigned, EA, 2);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.71,31.0:X:av:lvewx %VD, %RA, %RB:Load Vector Element Word Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~3;
+ eb = EA & 0xf;
+ (*vS).w[eb/4] = MEM(unsigned, EA, 4);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+
+0.31,6.VS,11.RA,16.RB,21.6,31.0:X:av:lvsl %VD, %RA, %RB:Load Vector for Shift Left
+ unsigned_word b;
+ unsigned_word addr;
+ int i, j;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ addr = b + *rB;
+ j = addr & 0xf;
+ for (i = 0; i < 16; i++)
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ (*vS).b[AV_BINDEX(i)] = j++;
+ else
+ (*vS).b[AV_BINDEX(15 - i)] = j++;
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.38,31.0:X:av:lvsr %VD, %RA, %RB:Load Vector for Shift Right
+ unsigned_word b;
+ unsigned_word addr;
+ int i, j;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ addr = b + *rB;
+ j = 0x10 - (addr & 0xf);
+ for (i = 0; i < 16; i++)
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ (*vS).b[AV_BINDEX(i)] = j++;
+ else
+ (*vS).b[AV_BINDEX(15 - i)] = j++;
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+
+0.31,6.VS,11.RA,16.RB,21.103,31.0:X:av:lvx %VD, %RA, %RB:Load Vector Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN) {
+ (*vS).w[0] = MEM(unsigned, EA + 0, 4);
+ (*vS).w[1] = MEM(unsigned, EA + 4, 4);
+ (*vS).w[2] = MEM(unsigned, EA + 8, 4);
+ (*vS).w[3] = MEM(unsigned, EA + 12, 4);
+ } else {
+ (*vS).w[0] = MEM(unsigned, EA + 12, 4);
+ (*vS).w[1] = MEM(unsigned, EA + 8, 4);
+ (*vS).w[2] = MEM(unsigned, EA + 4, 4);
+ (*vS).w[3] = MEM(unsigned, EA + 0, 4);
+ }
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.359,31.0:X:av:lvxl %VD, %RA, %RB:Load Vector Indexed LRU
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN) {
+ (*vS).w[0] = MEM(unsigned, EA + 0, 4);
+ (*vS).w[1] = MEM(unsigned, EA + 4, 4);
+ (*vS).w[2] = MEM(unsigned, EA + 8, 4);
+ (*vS).w[3] = MEM(unsigned, EA + 12, 4);
+ } else {
+ (*vS).w[0] = MEM(unsigned, EA + 12, 4);
+ (*vS).w[1] = MEM(unsigned, EA + 8, 4);
+ (*vS).w[2] = MEM(unsigned, EA + 4, 4);
+ (*vS).w[3] = MEM(unsigned, EA + 0, 4);
+ }
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+#
+# Move to/from VSCR instructions, 6-23 & 6-24.
+#
+
+0.4,6.VS,11.0,16.0,21.1540:VX:av:mfvscr %VS:Move from Vector Status and Control Register
+ (*vS).w[0] = 0;
+ (*vS).w[1] = 0;
+ (*vS).w[2] = 0;
+ (*vS).w[3] = VSCR;
+ PPC_INSN_FROM_VSCR(VS_BITMASK);
+
+0.4,6.0,11.0,16.VB,21.1604:VX:av:mtvscr %VB:Move to Vector Status and Control Register
+ VSCR = (*vB).w[3];
+ PPC_INSN_TO_VSCR(VB_BITMASK);
+
+#
+# Store instructions, 6-25 ... 6-29.
+#
+
+0.31,6.VS,11.RA,16.RB,21.135,31.0:X:av:stvebx %VD, %RA, %RB:Store Vector Element Byte Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ eb = EA & 0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ STORE(EA, 1, (*vS).b[eb]);
+ else
+ STORE(EA, 1, (*vS).b[15-eb]);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.167,31.0:X:av:stvehx %VD, %RA, %RB:Store Vector Element Half Word Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~1;
+ eb = EA & 0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ STORE(EA, 2, (*vS).h[eb/2]);
+ else
+ STORE(EA, 2, (*vS).h[7-eb]);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.199,31.0:X:av:stvewx %VD, %RA, %RB:Store Vector Element Word Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned_word eb;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~3;
+ eb = EA & 0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ STORE(EA, 4, (*vS).w[eb/4]);
+ else
+ STORE(EA, 4, (*vS).w[3-(eb/4)]);
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.231,31.0:X:av:stvx %VD, %RA, %RB:Store Vector Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN) {
+ STORE(EA + 0, 4, (*vS).w[0]);
+ STORE(EA + 4, 4, (*vS).w[1]);
+ STORE(EA + 8, 4, (*vS).w[2]);
+ STORE(EA + 12, 4, (*vS).w[3]);
+ } else {
+ STORE(EA + 12, 4, (*vS).w[0]);
+ STORE(EA + 8, 4, (*vS).w[1]);
+ STORE(EA + 4, 4, (*vS).w[2]);
+ STORE(EA + 0, 4, (*vS).w[3]);
+ }
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+0.31,6.VS,11.RA,16.RB,21.487,31.0:X:av:stvxl %VD, %RA, %RB:Store Vector Indexed LRU
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = (b + *rB) & ~0xf;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN) {
+ STORE(EA + 0, 4, (*vS).w[0]);
+ STORE(EA + 4, 4, (*vS).w[1]);
+ STORE(EA + 8, 4, (*vS).w[2]);
+ STORE(EA + 12, 4, (*vS).w[3]);
+ } else {
+ STORE(EA + 12, 4, (*vS).w[0]);
+ STORE(EA + 8, 4, (*vS).w[1]);
+ STORE(EA + 4, 4, (*vS).w[2]);
+ STORE(EA + 0, 4, (*vS).w[3]);
+ }
+ PPC_INSN_INT_VR(0, RA_BITMASK | RB_BITMASK, VS_BITMASK, 0);
+
+#
+# Vector Add instructions, 6-30 ... 6-40.
+#
+
+0.4,6.VS,11.VA,16.VB,21.384:VX:av:vaddcuw %VD, %VA, %VB:Vector Add Carryout Unsigned Word
+ unsigned64 temp;
+ int i;
+ for (i = 0; i < 4; i++) {
+ temp = (unsigned64)(*vA).w[i] + (unsigned64)(*vB).w[i];
+ (*vS).w[i] = temp >> 32;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.10:VX:av:vaddfp %VD, %VA, %VB:Vector Add Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_add (&d, &a, &b);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.768:VX:av:vaddsbs %VD, %VA, %VB:Vector Add Signed Byte Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ for (i = 0; i < 16; i++) {
+ temp = (signed16)(signed8)(*vA).b[i] + (signed16)(signed8)(*vB).b[i];
+ (*vS).b[i] = altivec_signed_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.832:VX:av:vaddshs %VD, %VA, %VB:Vector Add Signed Half Word Saturate
+ int i, sat, tempsat;
+ signed32 temp, a, b;
+ for (i = 0; i < 8; i++) {
+ a = (signed32)(signed16)(*vA).h[i];
+ b = (signed32)(signed16)(*vB).h[i];
+ temp = a + b;
+ (*vS).h[i] = altivec_signed_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.896:VX:av:vaddsws %VD, %VA, %VB:Vector Add Signed Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ for (i = 0; i < 4; i++) {
+ temp = (signed64)(signed32)(*vA).w[i] + (signed64)(signed32)(*vB).w[i];
+ (*vS).w[i] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.0:VX:av:vaddubm %VD, %VA, %VB:Vector Add Unsigned Byte Modulo
+ int i;
+ for (i = 0; i < 16; i++)
+ (*vS).b[i] = ((*vA).b[i] + (*vB).b[i]) & 0xff;
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.512:VX:av:vaddubs %VD, %VA, %VB:Vector Add Unsigned Byte Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ temp = (signed16)(unsigned8)(*vA).b[i] + (signed16)(unsigned8)(*vB).b[i];
+ (*vS).b[i] = altivec_unsigned_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.64:VX:av:vadduhm %VD, %VA, %VB:Vector Add Unsigned Half Word Modulo
+ int i;
+ for (i = 0; i < 8; i++)
+ (*vS).h[i] = ((*vA).h[i] + (*vB).h[i]) & 0xffff;
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.576:VX:av:vadduhs %VD, %VA, %VB:Vector Add Unsigned Half Word Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ for (i = 0; i < 8; i++) {
+ temp = (signed32)(unsigned16)(*vA).h[i] + (signed32)(unsigned16)(*vB).h[i];
+ (*vS).h[i] = altivec_unsigned_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.128:VX:av:vadduwm %VD, %VA, %VB:Vector Add Unsigned Word Modulo
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] + (*vB).w[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.640:VX:av:vadduws %VD, %VA, %VB:Vector Add Unsigned Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ for (i = 0; i < 4; i++) {
+ temp = (signed64)(unsigned32)(*vA).w[i] + (signed64)(unsigned32)(*vB).w[i];
+ (*vS).w[i] = altivec_unsigned_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+#
+# Vector AND instructions, 6-41, 6-42
+#
+
+0.4,6.VS,11.VA,16.VB,21.1028:VX:av:vand %VD, %VA, %VB:Vector Logical AND
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] & (*vB).w[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1092:VX:av:vandc %VD, %VA, %VB:Vector Logical AND with Compliment
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] & ~((*vB).w[i]);
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Average instructions, 6-43, 6-48
+#
+
+0.4,6.VS,11.VA,16.VB,21.1282:VX:av:vavgsb %VD, %VA, %VB:Vector Average Signed Byte
+ int i;
+ signed16 temp, a, b;
+ for (i = 0; i < 16; i++) {
+ a = (signed16)(signed8)(*vA).b[i];
+ b = (signed16)(signed8)(*vB).b[i];
+ temp = a + b + 1;
+ (*vS).b[i] = (temp >> 1) & 0xff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1346:VX:av:vavgsh %VD, %VA, %VB:Vector Average Signed Half Word
+ int i;
+ signed32 temp, a, b;
+ for (i = 0; i < 8; i++) {
+ a = (signed32)(signed16)(*vA).h[i];
+ b = (signed32)(signed16)(*vB).h[i];
+ temp = a + b + 1;
+ (*vS).h[i] = (temp >> 1) & 0xffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1410:VX:av:vavgsw %VD, %VA, %VB:Vector Average Signed Word
+ int i;
+ signed64 temp, a, b;
+ for (i = 0; i < 4; i++) {
+ a = (signed64)(signed32)(*vA).w[i];
+ b = (signed64)(signed32)(*vB).w[i];
+ temp = a + b + 1;
+ (*vS).w[i] = (temp >> 1) & 0xffffffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1026:VX:av:vavgub %VD, %VA, %VB:Vector Average Unsigned Byte
+ int i;
+ unsigned16 temp, a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ temp = a + b + 1;
+ (*vS).b[i] = (temp >> 1) & 0xff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1090:VX:av:vavguh %VD, %VA, %VB:Vector Average Unsigned Half Word
+ int i;
+ unsigned32 temp, a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ temp = a + b + 1;
+ (*vS).h[i] = (temp >> 1) & 0xffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1154:VX:av:vavguw %VD, %VA, %VB:Vector Average Unsigned Word
+ int i;
+ unsigned64 temp, a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ temp = a + b + 1;
+ (*vS).w[i] = (temp >> 1) & 0xffffffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+#
+# Vector Fixed Point Convert instructions, 6-49, 6-50
+#
+
+0.4,6.VS,11.UIMM,16.VB,21.842:VX:av:vcfsx %VD, %VB, %UIMM:Vector Convert From Signed Fixed-Point Word
+ int i;
+ unsigned32 f;
+ sim_fpu b, div, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_u32to (&div, 2 << UIMM, sim_fpu_round_default);
+ sim_fpu_div (&d, &b, &div);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.UIMM,16.VB,21.778:VX:av:vcfux %VD, %VA, %UIMM:Vector Convert From Unsigned Fixed-Point Word
+ int i;
+ unsigned32 f;
+ sim_fpu b, d, div;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_u32to (&div, 2 << UIMM, sim_fpu_round_default);
+ sim_fpu_div (&d, &b, &div);
+ sim_fpu_to32u (&f, &d, sim_fpu_round_default);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+#
+# Vector Compare instructions, 6-51 ... 6-64
+#
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.966:VXR:av:vcmpbpfpx %VD, %VA, %VB:Vector Compare Bounds Floating Point
+ int i, le, ge;
+ sim_fpu a, b, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ le = sim_fpu_is_le(&a, &b);
+ ge = sim_fpu_is_ge(&a, &b);
+ (*vS).w[i] = (le ? 0 : 1 << 31) | (ge ? 0 : 1 << 30);
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 0);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.198:VXR:av:vcmpeqfpx %VD, %VA, %VB:Vector Compare Equal-to-Floating Point
+ int i;
+ sim_fpu a, b;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ if (sim_fpu_is_eq(&a, &b))
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.6:VXR:av:vcmpequbx %VD, %VA, %VB:Vector Compare Equal-to Unsigned Byte
+ int i;
+ for (i = 0; i < 16; i++)
+ if ((*vA).b[i] == (*vB).b[i])
+ (*vS).b[i] = 0xff;
+ else
+ (*vS).b[i] = 0;
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.70:VXR:av:vcmpequhx %VD, %VA, %VB:Vector Compare Equal-to Unsigned Half Word
+ int i;
+ for (i = 0; i < 8; i++)
+ if ((*vA).h[i] == (*vB).h[i])
+ (*vS).h[i] = 0xffff;
+ else
+ (*vS).h[i] = 0;
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.134:VXR:av:vcmpequwx %VD, %VA, %VB:Vector Compare Equal-to Unsigned Word
+ int i;
+ for (i = 0; i < 4; i++)
+ if ((*vA).w[i] == (*vB).w[i])
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.454:VXR:av:vcmpgefpx %VD, %VA, %VB:Vector Compare Greater-Than-or-Equal-to Floating Point
+ int i;
+ sim_fpu a, b;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ if (sim_fpu_is_ge(&a, &b))
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.710:VXR:av:vcmpgtfpx %VD, %VA, %VB:Vector Compare Greater-Than Floating Point
+ int i;
+ sim_fpu a, b;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ if (sim_fpu_is_gt(&a, &b))
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.774:VXR:av:vcmpgtsbx %VD, %VA, %VB:Vector Compare Greater-Than Signed Byte
+ int i;
+ signed8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a > b)
+ (*vS).b[i] = 0xff;
+ else
+ (*vS).b[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.838:VXR:av:vcmpgtshx %VD, %VA, %VB:Vector Compare Greater-Than Signed Half Word
+ int i;
+ signed16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a > b)
+ (*vS).h[i] = 0xffff;
+ else
+ (*vS).h[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.902:VXR:av:vcmpgtswx %VD, %VA, %VB:Vector Compare Greater-Than Signed Word
+ int i;
+ signed32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a > b)
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.518:VXR:av:vcmpgtubx %VD, %VA, %VB:Vector Compare Greater-Than Unsigned Byte
+ int i;
+ unsigned8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a > b)
+ (*vS).b[i] = 0xff;
+ else
+ (*vS).b[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.582:VXR:av:vcmpgtuhx %VD, %VA, %VB:Vector Compare Greater-Than Unsigned Half Word
+ int i;
+ unsigned16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a > b)
+ (*vS).h[i] = 0xffff;
+ else
+ (*vS).h[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+0.4,6.VS,11.VA,16.VB,21.RC,22.646:VXR:av:vcmpgtuwx %VD, %VA, %VB:Vector Compare Greater-Than Unsigned Word
+ int i;
+ unsigned32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a > b)
+ (*vS).w[i] = 0xffffffff;
+ else
+ (*vS).w[i] = 0;
+ }
+ if (RC)
+ ALTIVEC_SET_CR6(vS, 1);
+ PPC_INSN_VR_CR(VS_BITMASK, VA_BITMASK | VB_BITMASK, RC ? 0x000000f0 : 0);
+
+#
+# Vector Convert instructions, 6-65, 6-66.
+#
+
+0.4,6.VS,11.UIMM,16.VB,21.970:VX:av:vctsxs %VD, %VB, %UIMM:Vector Convert to Signed Fixed-Point Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ sim_fpu a, b, m;
+ sat = 0;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_u32to (&m, 2 << UIMM, sim_fpu_round_default);
+ sim_fpu_mul (&a, &b, &m);
+ sim_fpu_to64i (&temp, &a, sim_fpu_round_default);
+ (*vS).w[i] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.UIMM,16.VB,21.906:VX:av:vctuxs %VD, %VB, %UIMM:Vector Convert to Unsigned Fixed-Point Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ sim_fpu a, b, m;
+ sat = 0;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_u32to (&m, 2 << UIMM, sim_fpu_round_default);
+ sim_fpu_mul (&a, &b, &m);
+ sim_fpu_to64u (&temp, &a, sim_fpu_round_default);
+ (*vS).w[i] = altivec_unsigned_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VB_BITMASK);
+
+#
+# Vector Estimate instructions, 6-67 ... 6-70.
+#
+
+0.4,6.VS,11.0,16.VB,21.394:VX:av:vexptefp %VD, %VB:Vector 2 Raised to the Exponent Estimate Floating Point
+ int i;
+ unsigned32 f;
+ signed32 bi;
+ sim_fpu b, d;
+ for (i = 0; i < 4; i++) {
+ /*HACK!*/
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_to32i (&bi, &b, sim_fpu_round_default);
+ bi = 2 ^ bi;
+ sim_fpu_32to (&d, bi);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR_VSCR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.458:VX:av:vlogefp %VD, %VB:Vector Log2 Estimate Floating Point
+ int i;
+ unsigned32 c, u, f;
+ sim_fpu b, cfpu, d;
+ for (i = 0; i < 4; i++) {
+ /*HACK!*/
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_to32u (&u, &b, sim_fpu_round_default);
+ for (c = 0; (u /= 2) > 1; c++)
+ ;
+ sim_fpu_32to (&cfpu, c);
+ sim_fpu_add (&d, &b, &cfpu);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR_VSCR(VS_BITMASK, VB_BITMASK);
+
+#
+# Vector Multiply Add instruction, 6-71
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.46:VAX:av:vmaddfp %VD, %VA, %VB, %VC:Vector Multiply Add Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, c, d, e;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_32to (&c, (*vC).w[i]);
+ sim_fpu_mul (&e, &a, &c);
+ sim_fpu_add (&d, &e, &b);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Maximum instructions, 6-72 ... 6-78.
+#
+
+0.4,6.VS,11.VA,16.VB,21.1034:VX:av:vmaxfp %VD, %VA, %VB:Vector Maximum Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_max (&d, &a, &b);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.258:VX:av:vmaxsb %VD, %VA, %VB:Vector Maximum Signed Byte
+ int i;
+ signed8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a > b)
+ (*vS).b[i] = a;
+ else
+ (*vS).b[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.322:VX:av:vmaxsh %VD, %VA, %VB:Vector Maximum Signed Half Word
+ int i;
+ signed16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a > b)
+ (*vS).h[i] = a;
+ else
+ (*vS).h[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.386:VX:av:vmaxsw %VD, %VA, %VB:Vector Maximum Signed Word
+ int i;
+ signed32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a > b)
+ (*vS).w[i] = a;
+ else
+ (*vS).w[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.2:VX:av:vmaxub %VD, %VA, %VB:Vector Maximum Unsigned Byte
+ int i;
+ unsigned8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a > b)
+ (*vS).b[i] = a;
+ else
+ (*vS).b[i] = b;
+ };
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.66:VX:av:vmaxus %VD, %VA, %VB:Vector Maximum Unsigned Half Word
+ int i;
+ unsigned16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a > b)
+ (*vS).h[i] = a;
+ else
+ (*vS).h[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.130:VX:av:vmaxuw %VD, %VA, %VB:Vector Maximum Unsigned Word
+ int i;
+ unsigned32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a > b)
+ (*vS).w[i] = a;
+ else
+ (*vS).w[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Multiple High instructions, 6-79, 6-80.
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.32:VAX:av:vmhaddshs %VD, %VA, %VB, %VC:Vector Multiple High and Add Signed Half Word Saturate
+ int i, sat, tempsat;
+ signed16 a, b;
+ signed32 prod, temp, c;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ c = (signed32)(signed16)(*vC).h[i];
+ prod = (signed32)a * (signed32)b;
+ temp = (prod >> 15) + c;
+ (*vS).h[i] = altivec_signed_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.33:VAX:av:vmhraddshs %VD, %VA, %VB, %VC:Vector Multiple High Round and Add Signed Half Word Saturate
+ int i, sat, tempsat;
+ signed16 a, b;
+ signed32 prod, temp, c;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ c = (signed32)(signed16)(*vC).h[i];
+ prod = (signed32)a * (signed32)b;
+ prod += 0x4000;
+ temp = (prod >> 15) + c;
+ (*vS).h[i] = altivec_signed_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Minimum instructions, 6-81 ... 6-87
+#
+
+0.4,6.VS,11.VA,16.VB,21.1098:VX:av:vminfp %VD, %VA, %VB:Vector Minimum Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_min (&d, &a, &b);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.770:VX:av:vminsb %VD, %VA, %VB:Vector Minimum Signed Byte
+ int i;
+ signed8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a < b)
+ (*vS).b[i] = a;
+ else
+ (*vS).b[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.834:VX:av:vminsh %VD, %VA, %VB:Vector Minimum Signed Half Word
+ int i;
+ signed16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a < b)
+ (*vS).h[i] = a;
+ else
+ (*vS).h[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.898:VX:av:vminsw %VD, %VA, %VB:Vector Minimum Signed Word
+ int i;
+ signed32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a < b)
+ (*vS).w[i] = a;
+ else
+ (*vS).w[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.514:VX:av:vminub %VD, %VA, %VB:Vector Minimum Unsigned Byte
+ int i;
+ unsigned8 a, b;
+ for (i = 0; i < 16; i++) {
+ a = (*vA).b[i];
+ b = (*vB).b[i];
+ if (a < b)
+ (*vS).b[i] = a;
+ else
+ (*vS).b[i] = b;
+ };
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.578:VX:av:vminuh %VD, %VA, %VB:Vector Minimum Unsigned Half Word
+ int i;
+ unsigned16 a, b;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ if (a < b)
+ (*vS).h[i] = a;
+ else
+ (*vS).h[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.642:VX:av:vminuw %VD, %VA, %VB:Vector Minimum Unsigned Word
+ int i;
+ unsigned32 a, b;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).w[i];
+ b = (*vB).w[i];
+ if (a < b)
+ (*vS).w[i] = a;
+ else
+ (*vS).w[i] = b;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Multiply Low instruction, 6-88
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.34:VAX:av:vmladduhm %VD, %VA, %VB, %VC:Vector Multiply Low and Add Unsigned Half Word Modulo
+ int i;
+ unsigned16 a, b, c;
+ unsigned32 prod;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).h[i];
+ b = (*vB).h[i];
+ c = (*vC).h[i];
+ prod = (unsigned32)a * (unsigned32)b;
+ (*vS).h[i] = (prod + c) & 0xffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Merge instructions, 6-89 ... 6-94
+#
+
+0.4,6.VS,11.VA,16.VB,21.12:VX:av:vmrghb %VD, %VA, %VB:Vector Merge High Byte
+ int i;
+ for (i = 0; i < 16; i += 2) {
+ (*vS).b[AV_BINDEX(i)] = (*vA).b[AV_BINDEX(i/2)];
+ (*vS).b[AV_BINDEX(i+1)] = (*vB).b[AV_BINDEX(i/2)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.76:VX:av:vmrghh %VD, %VA, %VB:Vector Merge High Half Word
+ int i;
+ for (i = 0; i < 8; i += 2) {
+ (*vS).h[AV_HINDEX(i)] = (*vA).h[AV_HINDEX(i/2)];
+ (*vS).h[AV_HINDEX(i+1)] = (*vB).h[AV_HINDEX(i/2)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.140:VX:av:vmrghw %VD, %VA, %VB:Vector Merge High Word
+ int i;
+ for (i = 0; i < 4; i += 2) {
+ (*vS).w[i] = (*vA).w[i/2];
+ (*vS).w[i+1] = (*vB).w[i/2];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.268:VX:av:vmrglb %VD, %VA, %VB:Vector Merge Low Byte
+ int i;
+ for (i = 0; i < 16; i += 2) {
+ (*vS).b[AV_BINDEX(i)] = (*vA).b[AV_BINDEX((i/2) + 8)];
+ (*vS).b[AV_BINDEX(i+1)] = (*vB).b[AV_BINDEX((i/2) + 8)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.332:VX:av:vmrglh %VD, %VA, %VB:Vector Merge Low Half Word
+ int i;
+ for (i = 0; i < 8; i += 2) {
+ (*vS).h[AV_HINDEX(i)] = (*vA).h[AV_HINDEX((i/2) + 4)];
+ (*vS).h[AV_HINDEX(i+1)] = (*vB).h[AV_HINDEX((i/2) + 4)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.396:VX:av:vmrglw %VD, %VA, %VB:Vector Merge Low Word
+ int i;
+ for (i = 0; i < 4; i += 2) {
+ (*vS).w[i] = (*vA).w[(i/2) + 2];
+ (*vS).w[i+1] = (*vB).w[(i/2) + 2];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Multiply Sum instructions, 6-95 ... 6-100
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.37:VAX:av:vmsummbm %VD, %VA, %VB, %VC:Vector Multiply Sum Mixed-Sign Byte Modulo
+ int i, j;
+ signed32 temp;
+ signed16 prod, a;
+ unsigned16 b;
+ for (i = 0; i < 4; i++) {
+ temp = (*vC).w[i];
+ for (j = 0; j < 4; j++) {
+ a = (signed16)(signed8)(*vA).b[i*4+j];
+ b = (*vB).b[i*4+j];
+ prod = a * b;
+ temp += (signed32)prod;
+ }
+ (*vS).w[i] = temp;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.40:VAX:av:vmsumshm %VD, %VA, %VB, %VC:Vector Multiply Sum Signed Half Word Modulo
+ int i, j;
+ signed32 temp, prod, a, b;
+ for (i = 0; i < 4; i++) {
+ temp = (*vC).w[i];
+ for (j = 0; j < 2; j++) {
+ a = (signed32)(signed16)(*vA).h[i*2+j];
+ b = (signed32)(signed16)(*vB).h[i*2+j];
+ prod = a * b;
+ temp += prod;
+ }
+ (*vS).w[i] = temp;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.41:VAX:av:vmsumshs %VD, %VA, %VB, %VC:Vector Multiply Sum Signed Half Word Saturate
+ int i, j, sat, tempsat;
+ signed64 temp;
+ signed32 prod, a, b;
+ sat = 0;
+ for (i = 0; i < 4; i++) {
+ temp = (signed64)(signed32)(*vC).w[i];
+ for (j = 0; j < 2; j++) {
+ a = (signed32)(signed16)(*vA).h[i*2+j];
+ b = (signed32)(signed16)(*vB).h[i*2+j];
+ prod = a * b;
+ temp += (signed64)prod;
+ }
+ (*vS).w[i] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.36:VAX:av:vmsumubm %VD, %VA, %VB, %VC:Vector Multiply Sum Unsigned Byte Modulo
+ int i, j;
+ unsigned32 temp;
+ unsigned16 prod, a, b;
+ for (i = 0; i < 4; i++) {
+ temp = (*vC).w[i];
+ for (j = 0; j < 4; j++) {
+ a = (*vA).b[i*4+j];
+ b = (*vB).b[i*4+j];
+ prod = a * b;
+ temp += prod;
+ }
+ (*vS).w[i] = temp;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.38:VAX:av:vmsumuhm %VD, %VA, %VB, %VC:Vector Multiply Sum Unsigned Half Word Modulo
+ int i, j;
+ unsigned32 temp, prod, a, b;
+ for (i = 0; i < 4; i++) {
+ temp = (*vC).w[i];
+ for (j = 0; j < 2; j++) {
+ a = (*vA).h[i*2+j];
+ b = (*vB).h[i*2+j];
+ prod = a * b;
+ temp += prod;
+ }
+ (*vS).w[i] = temp;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.39:VAX:av:vmsumuhs %VD, %VA, %VB, %VC:Vector Multiply Sum Unsigned Half Word Saturate
+ int i, j, sat, tempsat;
+ unsigned32 temp, prod, a, b;
+ sat = 0;
+ for (i = 0; i < 4; i++) {
+ temp = (*vC).w[i];
+ for (j = 0; j < 2; j++) {
+ a = (*vA).h[i*2+j];
+ b = (*vB).h[i*2+j];
+ prod = a * b;
+ temp += prod;
+ }
+ (*vS).w[i] = altivec_unsigned_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Multiply Even/Odd instructions, 6-101 ... 6-108
+#
+
+0.4,6.VS,11.VA,16.VB,21.776:VX:av:vmulesb %VD, %VA, %VB:Vector Multiply Even Signed Byte
+ int i;
+ signed8 a, b;
+ signed16 prod;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).b[AV_BINDEX(i*2)];
+ b = (*vB).b[AV_BINDEX(i*2)];
+ prod = a * b;
+ (*vS).h[AV_HINDEX(i)] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.840:VX:av:vmulesh %VD, %VA, %VB:Vector Multiply Even Signed Half Word
+ int i;
+ signed16 a, b;
+ signed32 prod;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).h[AV_HINDEX(i*2)];
+ b = (*vB).h[AV_HINDEX(i*2)];
+ prod = a * b;
+ (*vS).w[i] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.520:VX:av:vmuleub %VD, %VA, %VB:Vector Multiply Even Unsigned Byte
+ int i;
+ unsigned8 a, b;
+ unsigned16 prod;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).b[AV_BINDEX(i*2)];
+ b = (*vB).b[AV_BINDEX(i*2)];
+ prod = a * b;
+ (*vS).h[AV_HINDEX(i)] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.584:VX:av:vmuleuh %VD, %VA, %VB:Vector Multiply Even Unsigned Half Word
+ int i;
+ unsigned16 a, b;
+ unsigned32 prod;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).h[AV_HINDEX(i*2)];
+ b = (*vB).h[AV_HINDEX(i*2)];
+ prod = a * b;
+ (*vS).w[i] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.264:VX:av:vmulosb %VD, %VA, %VB:Vector Multiply Odd Signed Byte
+ int i;
+ signed8 a, b;
+ signed16 prod;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).b[AV_BINDEX((i*2)+1)];
+ b = (*vB).b[AV_BINDEX((i*2)+1)];
+ prod = a * b;
+ (*vS).h[AV_HINDEX(i)] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.328:VX:av:vmulosh %VD, %VA, %VB:Vector Multiply Odd Signed Half Word
+ int i;
+ signed16 a, b;
+ signed32 prod;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).h[AV_HINDEX((i*2)+1)];
+ b = (*vB).h[AV_HINDEX((i*2)+1)];
+ prod = a * b;
+ (*vS).w[i] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.8:VX:av:vmuloub %VD, %VA, %VB:Vector Multiply Odd Unsigned Byte
+ int i;
+ unsigned8 a, b;
+ unsigned16 prod;
+ for (i = 0; i < 8; i++) {
+ a = (*vA).b[AV_BINDEX((i*2)+1)];
+ b = (*vB).b[AV_BINDEX((i*2)+1)];
+ prod = a * b;
+ (*vS).h[AV_HINDEX(i)] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.72:VX:av:vmulouh %VD, %VA, %VB:Vector Multiply Odd Unsigned Half Word
+ int i;
+ unsigned16 a, b;
+ unsigned32 prod;
+ for (i = 0; i < 4; i++) {
+ a = (*vA).h[AV_HINDEX((i*2)+1)];
+ b = (*vB).h[AV_HINDEX((i*2)+1)];
+ prod = a * b;
+ (*vS).w[i] = prod;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Negative Multiply-Subtract instruction, 6-109
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.47:VX:av:vnmsubfp %VD, %VA, %VB, %VC:Vector Negative Multiply-Subtract Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, c, d, i1, i2;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_32to (&c, (*vC).w[i]);
+ sim_fpu_mul (&i1, &a, &c);
+ sim_fpu_sub (&i2, &i1, &b);
+ sim_fpu_neg (&d, &i2);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Logical OR instructions, 6-110, 6-111, 6-177
+#
+
+0.4,6.VS,11.VA,16.VB,21.1284:VX:av:vnor %VD, %VA, %VB:Vector Logical NOR
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = ~((*vA).w[i] | (*vB).w[i]);
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1156:VX:av:vor %VD, %VA, %VB:Vector Logical OR
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] | (*vB).w[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1220:VX:av:vxor %VD, %VA, %VB:Vector Logical XOR
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] ^ (*vB).w[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Permute instruction, 6-112
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.43:VX:av:vperm %VD, %VA, %VB, %VC:Vector Permute
+ int i, who;
+ for (i = 0; i < 16; i++) {
+ who = (*vC).b[AV_BINDEX(i)] & 0x1f;
+ if (who & 0x10)
+ (*vS).b[AV_BINDEX(i)] = (*vB).b[AV_BINDEX(who & 0xf)];
+ else
+ (*vS).b[AV_BINDEX(i)] = (*vA).b[AV_BINDEX(who & 0xf)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+
+#
+# Vector Pack instructions, 6-113 ... 6-121
+#
+
+0.4,6.VS,11.VA,16.VB,21.782:VX:av:vpkpx %VD, %VA, %VB:Vector Pack Pixel32
+ int i;
+ for (i = 0; i < 4; i++) {
+ (*vS).h[AV_HINDEX(i+4)] = ((((*vB).w[i]) >> 9) & 0xfc00)
+ | ((((*vB).w[i]) >> 6) & 0x03e0)
+ | ((((*vB).w[i]) >> 3) & 0x001f);
+ (*vS).h[AV_HINDEX(i)] = ((((*vA).w[i]) >> 9) & 0xfc00)
+ | ((((*vA).w[i]) >> 6) & 0x03e0)
+ | ((((*vA).w[i]) >> 3) & 0x001f);
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.398:VX:av:vpkshss %VD, %VA, %VB:Vector Pack Signed Half Word Signed Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ if (i < 8)
+ temp = (*vA).h[AV_HINDEX(i)];
+ else
+ temp = (*vB).h[AV_HINDEX(i-8)];
+ (*vS).b[AV_BINDEX(i)] = altivec_signed_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.270:VX:av:vpkshus %VD, %VA, %VB:Vector Pack Signed Half Word Unsigned Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ if (i < 8)
+ temp = (*vA).h[AV_HINDEX(i)];
+ else
+ temp = (*vB).h[AV_HINDEX(i-8)];
+ (*vS).b[AV_BINDEX(i)] = altivec_unsigned_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.462:VX:av:vpkswss %VD, %VA, %VB:Vector Pack Signed Word Signed Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ sat = 0;
+ for (i = 0; i < 8; i++) {
+ if (i < 4)
+ temp = (*vA).w[i];
+ else
+ temp = (*vB).w[i-4];
+ (*vS).h[AV_HINDEX(i)] = altivec_signed_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.334:VX:av:vpkswus %VD, %VA, %VB:Vector Pack Signed Word Unsigned Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ sat = 0;
+ for (i = 0; i < 8; i++) {
+ if (i < 4)
+ temp = (*vA).w[i];
+ else
+ temp = (*vB).w[i-4];
+ (*vS).h[AV_HINDEX(i)] = altivec_unsigned_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.14:VX:av:vpkuhum %VD, %VA, %VB:Vector Pack Unsigned Half Word Unsigned Modulo
+ int i;
+ for (i = 0; i < 16; i++)
+ if (i < 8)
+ (*vS).b[AV_BINDEX(i)] = (*vA).h[AV_HINDEX(i)];
+ else
+ (*vS).b[AV_BINDEX(i)] = (*vB).h[AV_HINDEX(i-8)];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.142:VX:av:vpkuhus %VD, %VA, %VB:Vector Pack Unsigned Half Word Unsigned Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ if (i < 8)
+ temp = (*vA).h[AV_HINDEX(i)];
+ else
+ temp = (*vB).h[AV_HINDEX(i-8)];
+ /* force positive in signed16, ok as we'll toss the bit away anyway */
+ temp &= ~0x8000;
+ (*vS).b[AV_BINDEX(i)] = altivec_unsigned_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.78:VX:av:vpkuwum %VD, %VA, %VB:Vector Pack Unsigned Word Unsigned Modulo
+ int i;
+ for (i = 0; i < 8; i++)
+ if (i < 8)
+ (*vS).h[AV_HINDEX(i)] = (*vA).w[i];
+ else
+ (*vS).h[AV_HINDEX(i)] = (*vB).w[i-8];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.206:VX:av:vpkuwus %VD, %VA, %VB:Vector Pack Unsigned Word Unsigned Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ sat = 0;
+ for (i = 0; i < 8; i++) {
+ if (i < 4)
+ temp = (*vA).w[i];
+ else
+ temp = (*vB).w[i-4];
+ /* force positive in signed32, ok as we'll toss the bit away anyway */
+ temp &= ~0x80000000;
+ (*vS).h[AV_HINDEX(i)] = altivec_unsigned_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Reciprocal instructions, 6-122, 6-123, 6-131
+#
+
+0.4,6.VS,11.0,16.VB,21.266:VX:av:vrefp %VD, %VB:Vector Reciprocal Estimate Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu op, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_div (&d, &sim_fpu_one, &op);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.330:VX:av:vrsqrtefp %VD, %VB:Vector Reciprocal Square Root Estimate Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu op, i1, one, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_sqrt (&i1, &op);
+ sim_fpu_div (&d, &sim_fpu_one, &i1);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+
+#
+# Vector Round instructions, 6-124 ... 6-127
+#
+
+0.4,6.VS,11.0,16.VB,21.714:VX:av:vrfim %VD, %VB:Vector Round to Floating-Point Integer towards Minus Infinity
+ int i;
+ unsigned32 f;
+ sim_fpu op;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_round_32(&op, sim_fpu_round_down, sim_fpu_denorm_default);
+ sim_fpu_to32 (&f, &op);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.522:VX:av:vrfin %VD, %VB:Vector Round to Floating-Point Integer Nearest
+ int i;
+ unsigned32 f;
+ sim_fpu op;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_round_32(&op, sim_fpu_round_near, sim_fpu_denorm_default);
+ sim_fpu_to32 (&f, &op);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.650:VX:av:vrfip %VD, %VB:Vector Round to Floating-Point Integer towards Plus Infinity
+ int i;
+ unsigned32 f;
+ sim_fpu op;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_round_32(&op, sim_fpu_round_up, sim_fpu_denorm_default);
+ sim_fpu_to32 (&f, &op);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.586:VX:av:vrfiz %VD, %VB:Vector Round to Floating-Point Integer towards Zero
+ int i;
+ unsigned32 f;
+ sim_fpu op;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&op, (*vB).w[i]);
+ sim_fpu_round_32(&op, sim_fpu_round_zero, sim_fpu_denorm_default);
+ sim_fpu_to32 (&f, &op);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+
+#
+# Vector Rotate Left instructions, 6-128 ... 6-130
+#
+
+0.4,6.VS,11.VA,16.VB,21.4:VX:av:vrlb %VD, %VA, %VB:Vector Rotate Left Integer Byte
+ int i;
+ unsigned16 temp;
+ for (i = 0; i < 16; i++) {
+ temp = (unsigned16)(*vA).b[i] << (((*vB).b[i]) & 7);
+ (*vS).b[i] = (temp & 0xff) | ((temp >> 8) & 0xff);
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.68:VX:av:vrlh %VD, %VA, %VB:Vector Rotate Left Integer Half Word
+ int i;
+ unsigned32 temp;
+ for (i = 0; i < 8; i++) {
+ temp = (unsigned32)(*vA).h[i] << (((*vB).h[i]) & 0xf);
+ (*vS).h[i] = (temp & 0xffff) | ((temp >> 16) & 0xffff);
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.132:VX:av:vrlw %VD, %VA, %VB:Vector Rotate Left Integer Word
+ int i;
+ unsigned64 temp;
+ for (i = 0; i < 4; i++) {
+ temp = (unsigned64)(*vA).w[i] << (((*vB).w[i]) & 0x1f);
+ (*vS).w[i] = (temp & 0xffffffff) | ((temp >> 32) & 0xffffffff);
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Conditional Select instruction, 6-133
+#
+
+0.4,6.VS,11.VA,16.VB,21.VC,26.42:VAX:av:vsel %VD, %VA, %VB, %VC:Vector Conditional Select
+ int i;
+ unsigned32 c;
+ for (i = 0; i < 4; i++) {
+ c = (*vC).w[i];
+ (*vS).w[i] = ((*vB).w[i] & c) | ((*vA).w[i] & ~c);
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK | VC_BITMASK);
+
+#
+# Vector Shift Left instructions, 6-134 ... 6-139
+#
+
+0.4,6.VS,11.VA,16.VB,21.452:VX:av:vsl %VD, %VA, %VB:Vector Shift Left
+ int sh, i, j, carry, new_carry;
+ sh = (*vB).b[0] & 7; /* don't bother checking everything */
+ carry = 0;
+ for (j = 3; j >= 0; j--) {
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ i = j;
+ else
+ i = (j + 2) % 4;
+ new_carry = (*vA).w[i] >> (32 - sh);
+ (*vS).w[i] = ((*vA).w[i] << sh) | carry;
+ carry = new_carry;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.260:VX:av:vslb %VD, %VA, %VB:Vector Shift Left Integer Byte
+ int i, sh;
+ for (i = 0; i < 16; i++) {
+ sh = ((*vB).b[i]) & 7;
+ (*vS).b[i] = (*vA).b[i] << sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.0,22.SH,26.44:VX:av:vsldol %VD, %VA, %VB:Vector Shift Left Double by Octet Immediate
+ int i, j;
+ for (j = 0, i = SH; i < 16; i++)
+ (*vS).b[j++] = (*vA).b[i];
+ for (i = 0; i < SH; i++)
+ (*vS).b[j++] = (*vB).b[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.324:VX:av:vslh %VD, %VA, %VB:Vector Shift Left Half Word
+ int i, sh;
+ for (i = 0; i < 8; i++) {
+ sh = ((*vB).h[i]) & 0xf;
+ (*vS).h[i] = (*vA).h[i] << sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1036:VX:av:vslo %VD, %VA, %VB:Vector Shift Left by Octet
+ int i, sh;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ sh = ((*vB).b[AV_BINDEX(15)] >> 3) & 0xf;
+ else
+ sh = ((*vB).b[AV_BINDEX(0)] >> 3) & 0xf;
+ for (i = 0; i < 16; i++) {
+ if (15 - i > sh)
+ (*vS).b[AV_BINDEX(i)] = (*vA).b[AV_BINDEX(i + sh)];
+ else
+ (*vS).b[AV_BINDEX(i)] = 0;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.388:VX:av:vslw %VD, %VA, %VB:Vector Shift Left Integer Word
+ int i, sh;
+ for (i = 0; i < 4; i++) {
+ sh = ((*vB).w[i]) & 0x1f;
+ (*vS).w[i] = (*vA).w[i] << sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Splat instructions, 6-140 ... 6-145
+#
+
+0.4,6.VS,11.UIMM,16.VB,21.524:VX:av:vspltb %VD, %VB, %UIMM:Vector Splat Byte
+ int i;
+ unsigned8 b;
+ b = (*vB).b[AV_BINDEX(UIMM & 0xf)];
+ for (i = 0; i < 16; i++)
+ (*vS).b[i] = b;
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.UIMM,16.VB,21.588:VX:av:vsplth %VD, %VB, %UIMM:Vector Splat Half Word
+ int i;
+ unsigned16 h;
+ h = (*vB).h[AV_HINDEX(UIMM & 0x7)];
+ for (i = 0; i < 8; i++)
+ (*vS).h[i] = h;
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.SIMM,16.0,21.780:VX:av:vspltisb %VD, %SIMM:Vector Splat Immediate Signed Byte
+ int i;
+ signed8 b = SIMM;
+ /* manual 5-bit signed extension */
+ if (b & 0x10)
+ b -= 0x20;
+ for (i = 0; i < 16; i++)
+ (*vS).b[i] = b;
+ PPC_INSN_VR(VS_BITMASK, 0);
+
+0.4,6.VS,11.SIMM,16.0,21.844:VX:av:vspltish %VD, %SIMM:Vector Splat Immediate Signed Half Word
+ int i;
+ signed16 h = SIMM;
+ /* manual 5-bit signed extension */
+ if (h & 0x10)
+ h -= 0x20;
+ for (i = 0; i < 8; i++)
+ (*vS).h[i] = h;
+ PPC_INSN_VR(VS_BITMASK, 0);
+
+0.4,6.VS,11.SIMM,16.0,21.908:VX:av:vspltisw %VD, %SIMM:Vector Splat Immediate Signed Word
+ int i;
+ signed32 w = SIMM;
+ /* manual 5-bit signed extension */
+ if (w & 0x10)
+ w -= 0x20;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = w;
+ PPC_INSN_VR(VS_BITMASK, 0);
+
+0.4,6.VS,11.UIMM,16.VB,21.652:VX:av:vspltw %VD, %VB, %UIMM:Vector Splat Word
+ int i;
+ unsigned32 w;
+ w = (*vB).w[UIMM & 0x3];
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = w;
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+
+#
+# Vector Shift Right instructions, 6-146 ... 6-154
+#
+
+0.4,6.VS,11.VA,16.VB,21.708:VX:av:vsr %VD, %VA, %VB:Vector Shift Right
+ int sh, i, j, carry, new_carry;
+ sh = (*vB).b[0] & 7; /* don't bother checking everything */
+ carry = 0;
+ for (j = 0; j < 4; j++) {
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ i = j;
+ else
+ i = (j + 2) % 4;
+ new_carry = (*vA).w[i] << (32 - sh);
+ (*vS).w[i] = ((*vA).w[i] >> sh) | carry;
+ carry = new_carry;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.772:VX:av:vsrab %VD, %VA, %VB:Vector Shift Right Algebraic Byte
+ int i, sh;
+ signed16 a;
+ for (i = 0; i < 16; i++) {
+ sh = ((*vB).b[i]) & 7;
+ a = (signed16)(signed8)(*vA).b[i];
+ (*vS).b[i] = (a >> sh) & 0xff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.836:VX:av:vsrah %VD, %VA, %VB:Vector Shift Right Algebraic Half Word
+ int i, sh;
+ signed32 a;
+ for (i = 0; i < 8; i++) {
+ sh = ((*vB).h[i]) & 0xf;
+ a = (signed32)(signed16)(*vA).h[i];
+ (*vS).h[i] = (a >> sh) & 0xffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.900:VX:av:vsraw %VD, %VA, %VB:Vector Shift Right Algebraic Word
+ int i, sh;
+ signed64 a;
+ for (i = 0; i < 4; i++) {
+ sh = ((*vB).w[i]) & 0xf;
+ a = (signed64)(signed32)(*vA).w[i];
+ (*vS).w[i] = (a >> sh) & 0xffffffff;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.516:VX:av:vsrb %VD, %VA, %VB:Vector Shift Right Byte
+ int i, sh;
+ for (i = 0; i < 16; i++) {
+ sh = ((*vB).b[i]) & 7;
+ (*vS).b[i] = (*vA).b[i] >> sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.580:VX:av:vsrh %VD, %VA, %VB:Vector Shift Right Half Word
+ int i, sh;
+ for (i = 0; i < 8; i++) {
+ sh = ((*vB).h[i]) & 0xf;
+ (*vS).h[i] = (*vA).h[i] >> sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1100:VX:av:vsro %VD, %VA, %VB:Vector Shift Right Octet
+ int i, sh;
+ if (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)
+ sh = ((*vB).b[AV_BINDEX(15)] >> 3) & 0xf;
+ else
+ sh = ((*vB).b[AV_BINDEX(0)] >> 3) & 0xf;
+ for (i = 0; i < 16; i++) {
+ if (i < sh)
+ (*vS).b[AV_BINDEX(i)] = 0;
+ else
+ (*vS).b[AV_BINDEX(i)] = (*vA).b[AV_BINDEX(i - sh)];
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.644:VX:av:vsrw %VD, %VA, %VB:Vector Shift Right Word
+ int i, sh;
+ for (i = 0; i < 4; i++) {
+ sh = ((*vB).w[i]) & 0x1f;
+ (*vS).w[i] = (*vA).w[i] >> sh;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Subtract instructions, 6-155 ... 6-165
+#
+
+0.4,6.VS,11.VA,16.VB,21.1408:VX:av:vsubcuw %VD, %VA, %VB:Vector Subtract Carryout Unsigned Word
+ int i;
+ signed64 temp, a, b;
+ for (i = 0; i < 4; i++) {
+ a = (signed64)(unsigned32)(*vA).w[i];
+ b = (signed64)(unsigned32)(*vB).w[i];
+ temp = a - b;
+ (*vS).w[i] = ~(temp >> 32) & 1;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.74:VX:av:vsubfp %VD, %VA, %VB:Vector Subtract Floating Point
+ int i;
+ unsigned32 f;
+ sim_fpu a, b, d;
+ for (i = 0; i < 4; i++) {
+ sim_fpu_32to (&a, (*vA).w[i]);
+ sim_fpu_32to (&b, (*vB).w[i]);
+ sim_fpu_sub (&d, &a, &b);
+ sim_fpu_to32 (&f, &d);
+ (*vS).w[i] = f;
+ }
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1792:VX:av:vsubsbs %VD, %VA, %VB:Vector Subtract Signed Byte Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ temp = (signed16)(signed8)(*vA).b[i] - (signed16)(signed8)(*vB).b[i];
+ (*vS).b[i] = altivec_signed_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1856:VX:av:vsubshs %VD, %VA, %VB:Vector Subtract Signed Half Word Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ sat = 0;
+ for (i = 0; i < 8; i++) {
+ temp = (signed32)(signed16)(*vA).h[i] - (signed32)(signed16)(*vB).h[i];
+ (*vS).h[i] = altivec_signed_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1920:VX:av:vsubsws %VD, %VA, %VB:Vector Subtract Signed Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ sat = 0;
+ for (i = 0; i < 4; i++) {
+ temp = (signed64)(signed32)(*vA).w[i] - (signed64)(signed32)(*vB).w[i];
+ (*vS).w[i] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1024:VX:av:vsububm %VD, %VA, %VB:Vector Subtract Unsigned Byte Modulo
+ int i;
+ for (i = 0; i < 16; i++)
+ (*vS).b[i] = (*vA).b[i] - (*vB).b[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1536:VX:av:vsububs %VD, %VA, %VB:Vector Subtract Unsigned Byte Saturate
+ int i, sat, tempsat;
+ signed16 temp;
+ sat = 0;
+ for (i = 0; i < 16; i++) {
+ temp = (signed16)(unsigned8)(*vA).b[i] - (signed16)(unsigned8)(*vB).b[i];
+ (*vS).b[i] = altivec_unsigned_saturate_8(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1088:VX:av:vsubuhm %VD, %VA, %VB:Vector Subtract Unsigned Half Word Modulo
+ int i;
+ for (i = 0; i < 8; i++)
+ (*vS).h[i] = ((*vA).h[i] - (*vB).h[i]) & 0xffff;
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1600:VX:av:vsubuhs %VD, %VA, %VB:Vector Subtract Unsigned Half Word Saturate
+ int i, sat, tempsat;
+ signed32 temp;
+ for (i = 0; i < 8; i++) {
+ temp = (signed32)(unsigned16)(*vA).h[i] - (signed32)(unsigned16)(*vB).h[i];
+ (*vS).h[i] = altivec_unsigned_saturate_16(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1152:VX:av:vsubuwm %VD, %VA, %VB:Vector Subtract Unsigned Word Modulo
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (*vA).w[i] - (*vB).w[i];
+ PPC_INSN_VR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1664:VX:av:vsubuws %VD, %VA, %VB:Vector Subtract Unsigned Word Saturate
+ int i, sat, tempsat;
+ signed64 temp;
+ for (i = 0; i < 4; i++) {
+ temp = (signed64)(unsigned32)(*vA).w[i] - (signed64)(unsigned32)(*vB).w[i];
+ (*vS).w[i] = altivec_unsigned_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Sum instructions, 6-166 ... 6-170
+#
+
+0.4,6.VS,11.VA,16.VB,21.1928:VX:av:vsumsws %VD, %VA, %VB:Vector Sum Across Signed Word Saturate
+ int i, sat;
+ signed64 temp;
+ temp = (signed64)(signed32)(*vB).w[3];
+ for (i = 0; i < 4; i++)
+ temp += (signed64)(signed32)(*vA).w[i];
+ (*vS).w[3] = altivec_signed_saturate_32(temp, &sat);
+ (*vS).w[0] = (*vS).w[1] = (*vS).w[2] = 0;
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1672:VX:av:vsum2sws %VD, %VA, %VB:Vector Sum Across Partial (1/2) Signed Word Saturate
+ int i, j, sat, tempsat;
+ signed64 temp;
+ for (j = 0; j < 4; j += 2) {
+ temp = (signed64)(signed32)(*vB).w[j+1];
+ temp += (signed64)(signed32)(*vA).w[j] + (signed64)(signed32)(*vA).w[j+1];
+ (*vS).w[j+1] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ (*vS).w[0] = (*vS).w[2] = 0;
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1800:VX:av:vsum4sbs %VD, %VA, %VB:Vector Sum Across Partial (1/4) Signed Byte Saturate
+ int i, j, sat, tempsat;
+ signed64 temp;
+ for (j = 0; j < 4; j++) {
+ temp = (signed64)(signed32)(*vB).w[j];
+ for (i = 0; i < 4; i++)
+ temp += (signed64)(signed8)(*vA).b[i+(j*4)];
+ (*vS).w[j] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1608:VX:av:vsum4shs %VD, %VA, %VB:Vector Sum Across Partial (1/4) Signed Half Word Saturate
+ int i, j, sat, tempsat;
+ signed64 temp;
+ for (j = 0; j < 4; j++) {
+ temp = (signed64)(signed32)(*vB).w[j];
+ for (i = 0; i < 2; i++)
+ temp += (signed64)(signed16)(*vA).h[i+(j*2)];
+ (*vS).w[j] = altivec_signed_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+0.4,6.VS,11.VA,16.VB,21.1544:VX:av:vsum4ubs %VD, %VA, %VB:Vector Sum Across Partial (1/4) Unsigned Byte Saturate
+ int i, j, sat, tempsat;
+ signed64 utemp;
+ signed64 temp;
+ for (j = 0; j < 4; j++) {
+ utemp = (signed64)(unsigned32)(*vB).w[j];
+ for (i = 0; i < 4; i++)
+ utemp += (signed64)(unsigned16)(*vA).b[i+(j*4)];
+ temp = utemp;
+ (*vS).w[j] = altivec_unsigned_saturate_32(temp, &tempsat);
+ sat |= tempsat;
+ }
+ ALTIVEC_SET_SAT(sat);
+ PPC_INSN_VR_VSCR(VS_BITMASK, VA_BITMASK | VB_BITMASK);
+
+
+#
+# Vector Unpack instructions, 6-171 ... 6-176
+#
+
+0.4,6.VS,11.0,16.VB,21.846:VX:av:vupkhpx %VD, %VB:Vector Unpack High Pixel16
+ int i;
+ unsigned16 h;
+ for (i = 0; i < 4; i++) {
+ h = (*vB).h[AV_HINDEX(i)];
+ (*vS).w[i] = ((h & 0x8000) ? 0xff000000 : 0)
+ | ((h & 0x7c00) << 6)
+ | ((h & 0x03e0) << 3)
+ | ((h & 0x001f));
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.526:VX:av:vupkhsb %VD, %VB:Vector Unpack High Signed Byte
+ int i;
+ for (i = 0; i < 8; i++)
+ (*vS).h[AV_HINDEX(i)] = (signed16)(signed8)(*vB).b[AV_BINDEX(i)];
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.590:VX:av:vupkhsh %VD, %VB:Vector Unpack High Signed Half Word
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (signed32)(signed16)(*vB).h[AV_HINDEX(i)];
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.974:VX:av:vupklpx %VD, %VB:Vector Unpack Low Pixel16
+ int i;
+ unsigned16 h;
+ for (i = 0; i < 4; i++) {
+ h = (*vB).h[AV_HINDEX(i + 4)];
+ (*vS).w[i] = ((h & 0x8000) ? 0xff000000 : 0)
+ | ((h & 0x7c00) << 6)
+ | ((h & 0x03e0) << 3)
+ | ((h & 0x001f));
+ }
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.654:VX:av:vupklsb %VD, %VB:Vector Unpack Low Signed Byte
+ int i;
+ for (i = 0; i < 8; i++)
+ (*vS).h[AV_HINDEX(i)] = (signed16)(signed8)(*vB).b[AV_BINDEX(i + 8)];
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
+
+0.4,6.VS,11.0,16.VB,21.718:VX:av:vupklsh %VD, %VB:Vector Unpack Low Signed Half Word
+ int i;
+ for (i = 0; i < 4; i++)
+ (*vS).w[i] = (signed32)(signed16)(*vB).h[AV_HINDEX(i + 4)];
+ PPC_INSN_VR(VS_BITMASK, VB_BITMASK);
--- /dev/null
+# e500 core instructions, for PSIM, the PowerPC simulator.
+
+# Copyright 2003 Free Software Foundation, Inc.
+
+# Contributed by Red Hat Inc; developed under contract from Motorola.
+# Written by matthew green <mrg@redhat.com>.
+
+# This file is part of GDB.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2, or (at your option)
+# any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with This program; see the file COPYING. If not, write to
+# the Free Software Foundation, 59 Temple Place - Suite 330,
+# Boston, MA 02111-1307, USA.
+
+#
+# e500 Core Complex Instructions
+#
+
+:cache:e500::signed_word *:rAh:RA:(cpu_registers(processor)->e500.gprh + RA)
+:cache:e500::signed_word *:rSh:RS:(cpu_registers(processor)->e500.gprh + RS)
+:cache:e500::signed_word *:rBh:RB:(cpu_registers(processor)->e500.gprh + RB)
+
+# Flags for model.h
+::model-macro:::
+ #define PPC_INSN_INT_SPR(OUT_MASK, IN_MASK, SPR) \
+ do { \
+ if (CURRENT_MODEL_ISSUE > 0) \
+ ppc_insn_int_spr(MY_INDEX, cpu_model(processor), OUT_MASK, IN_MASK, SPR); \
+ } while (0)
+
+# Schedule an instruction that takes 2 integer register and produces a special purpose output register plus an integer output register
+void::model-function::ppc_insn_int_spr:itable_index index, model_data *model_ptr, const unsigned32 out_mask, const unsigned32 in_mask, const unsigned nSPR
+ const unsigned32 int_mask = out_mask | in_mask;
+ model_busy *busy_ptr;
+
+ while ((model_ptr->int_busy & int_mask) != 0 || model_ptr->spr_busy[nSPR] != 0) {
+ if (WITH_TRACE && ppc_trace[trace_model])
+ model_trace_busy_p(model_ptr, int_mask, 0, 0, nSPR);
+
+ model_ptr->nr_stalls_data++;
+ model_new_cycle(model_ptr);
+ }
+
+ busy_ptr = model_wait_for_unit(index, model_ptr, &model_ptr->timing[index]);
+ busy_ptr->int_busy |= out_mask;
+ model_ptr->int_busy |= out_mask;
+ busy_ptr->spr_busy = nSPR;
+ model_ptr->spr_busy[nSPR] = 1;
+ busy_ptr->nr_writebacks = (PPC_ONE_BIT_SET_P(out_mask)) ? 3 : 2;
+ TRACE(trace_model,("Making register %s busy.\n", spr_name(nSPR)));
+\f
+#
+# SPE Modulo Fractional Multiplication handling support
+#
+:function:e500::unsigned64:ev_multiply16_smf:signed16 a, signed16 b, int *sat
+ signed32 a32 = a, b32 = b, rv32;
+ rv32 = a * b;
+ *sat = (rv32 & (3<<30)) == (3<<30);
+ return (signed64)rv32 << 1;
+
+:function:e500::unsigned64:ev_multiply32_smf:signed32 a, signed32 b, int *sat
+ signed64 rv64, a64 = a, b64 = b;
+ rv64 = a64 * b64;
+ *sat = (rv64 & ((signed64)3<<62)) == ((signed64)3<<62);
+ /* Loses top sign bit. */
+ return rv64 << 1;
+#
+# SPE Saturation handling support
+#
+:function:e500::signed32:ev_multiply16_ssf:signed16 a, signed16 b, int *sat
+ signed32 rv32;
+ if (a == 0xffff8000 && b == 0xffff8000)
+ {
+ rv32 = 0x7fffffffL;
+ * sat = 1;
+ return rv32;
+ }
+ else
+ {
+ signed32 a32 = a, b32 = b;
+
+ rv32 = a * b;
+ * sat = (rv32 & (3<<30)) == (3<<30);
+ return (signed64)rv32 << 1;
+ }
+
+:function:e500::signed64:ev_multiply32_ssf:signed32 a, signed32 b, int *sat
+ signed64 rv64;
+ if (a == 0x80000000 && b == 0x80000000)
+ {
+ rv64 = 0x7fffffffffffffffLL;
+ * sat = 1;
+ return rv64;
+ }
+ else
+ {
+ signed64 a64 = a, b64 = b;
+ rv64 = a64 * b64;
+ *sat = (rv64 & ((signed64)3<<62)) == ((signed64)3<<62);
+ /* Loses top sign bit. */
+ return rv64 << 1;
+ }
+\f
+#
+# SPE FP handling support
+#
+
+:function:e500::void:ev_check_guard:sim_fpu *a, int fg, int fx, cpu *processor
+ unsigned64 guard;
+ guard = sim_fpu_guard(a, 0);
+ if (guard & 1)
+ EV_SET_SPEFSCR_BITS(fg);
+ if (guard & ~1)
+ EV_SET_SPEFSCR_BITS(fx);
+
+:function:e500::void:booke_sim_fpu_32to:sim_fpu *dst, unsigned32 packed
+ sim_fpu_32to (dst, packed);
+
+ /* Set normally unused fields to allow booke arithmetic. */
+ if (dst->class == sim_fpu_class_infinity)
+ {
+ dst->normal_exp = 128;
+ dst->fraction = ((unsigned64)1 << 60);
+ }
+ else if (dst->class == sim_fpu_class_qnan
+ || dst->class == sim_fpu_class_snan)
+ {
+ dst->normal_exp = 128;
+ /* This is set, but without the implicit bit, so we have to or
+ in the implicit bit. */
+ dst->fraction |= ((unsigned64)1 << 60);
+ }
+
+:function:e500::int:booke_sim_fpu_add:sim_fpu *d, sim_fpu *a, sim_fpu *b, int inv, int over, int under, cpu *processor
+ int invalid_operand, overflow_result, underflow_result;
+ int dest_exp;
+
+ invalid_operand = 0;
+ overflow_result = 0;
+ underflow_result = 0;
+
+ /* Treat NaN, Inf, and denorm like normal numbers, and signal invalid
+ operand if it hasn't already been done. */
+ if (EV_IS_INFDENORMNAN (a))
+ {
+ a->class = sim_fpu_class_number;
+
+ EV_SET_SPEFSCR_BITS (inv);
+ invalid_operand = 1;
+ }
+ if (EV_IS_INFDENORMNAN (b))
+ {
+ b->class = sim_fpu_class_number;
+
+ if (! invalid_operand)
+ {
+ EV_SET_SPEFSCR_BITS (inv);
+ invalid_operand = 1;
+ }
+ }
+
+ sim_fpu_add (d, a, b);
+
+ dest_exp = booke_sim_fpu_exp (d);
+ /* If this is a denorm, force to zero, and signal underflow if
+ we haven't already indicated invalid operand. */
+ if (dest_exp <= -127)
+ {
+ int sign = d->sign;
+
+ *d = sim_fpu_zero;
+ d->sign = sign;
+ if (! invalid_operand)
+ {
+ EV_SET_SPEFSCR_BITS (under);
+ underflow_result = 1;
+ }
+ }
+ /* If this is Inf/NaN, force to pmax/nmax, and signal overflow if
+ we haven't already indicated invalid operand. */
+ else if (dest_exp >= 127)
+ {
+ int sign = d->sign;
+
+ *d = sim_fpu_max32;
+ d->sign = sign;
+ if (! invalid_operand)
+ {
+ EV_SET_SPEFSCR_BITS (over);
+ overflow_result = 1;
+ }
+ }
+ /* Destination sign is sign of operand with larger magnitude, or
+ the sign of the first operand if operands have the same
+ magnitude. Thus if the result is zero, we force it to have
+ the sign of the first operand. */
+ else if (d->fraction == 0)
+ d->sign = a->sign;
+
+ return invalid_operand || overflow_result || underflow_result;
+
+:function:e500::unsigned32:ev_fs_add:unsigned32 aa, unsigned32 bb, int inv, int over, int under, int fg, int fx, cpu *processor
+ sim_fpu a, b, d;
+ unsigned32 w;
+ int exception;
+
+ booke_sim_fpu_32to (&a, aa);
+ booke_sim_fpu_32to (&b, bb);
+
+ exception = booke_sim_fpu_add (&d, &a, &b, inv, over, under,
+ processor);
+
+ sim_fpu_to32 (&w, &d);
+ if (! exception)
+ ev_check_guard(&d, fg, fx, processor);
+ return w;
+
+:function:e500::unsigned32:ev_fs_sub:unsigned32 aa, unsigned32 bb, int inv, int over, int under, int fg, int fx, cpu *processor
+ sim_fpu a, b, d;
+ unsigned32 w;
+ int exception;
+
+ booke_sim_fpu_32to (&a, aa);
+ booke_sim_fpu_32to (&b, bb);
+
+ /* Invert sign of second operand, and add. */
+ b.sign = ! b.sign;
+ exception = booke_sim_fpu_add (&d, &a, &b, inv, over, under,
+ processor);
+
+ sim_fpu_to32 (&w, &d);
+ if (! exception)
+ ev_check_guard(&d, fg, fx, processor);
+ return w;
+
+# sim_fpu_exp leaves the normal_exp field undefined for Inf and NaN.
+# The booke algorithms require exp values, so we fake them here.
+# fixme: It also apparently does the same for zero, but should not.
+:function:e500::unsigned32:booke_sim_fpu_exp:sim_fpu *x
+ int y = sim_fpu_is (x);
+ if (y == SIM_FPU_IS_PZERO || y == SIM_FPU_IS_NZERO)
+ return 0;
+ else if (y == SIM_FPU_IS_SNAN || y == SIM_FPU_IS_QNAN
+ || y == SIM_FPU_IS_NINF || y == SIM_FPU_IS_PINF)
+ return 128;
+ else
+ return sim_fpu_exp (x);
+
+:function:e500::unsigned32:ev_fs_mul:unsigned32 aa, unsigned32 bb, int inv, int over, int under, int fg, int fx, cpu *processor
+ sim_fpu a, b, d;
+ unsigned32 w;
+ int sa, sb, ea, eb, ei;
+ sim_fpu_32to (&a, aa);
+ sim_fpu_32to (&b, bb);
+ sa = sim_fpu_sign(&a);
+ sb = sim_fpu_sign(&b);
+ ea = booke_sim_fpu_exp(&a);
+ eb = booke_sim_fpu_exp(&b);
+ ei = ea + eb + 127;
+ if (sim_fpu_is_zero (&a) || sim_fpu_is_zero (&b))
+ w = 0;
+ else if (sa == sb) {
+ if (ei >= 254) {
+ w = EV_PMAX;
+ EV_SET_SPEFSCR_BITS(over);
+ } else if (ei < 1) {
+ d = sim_fpu_zero;
+ sim_fpu_to32 (&w, &d);
+ w &= 0x7fffffff; /* Clear sign bit. */
+ } else {
+ goto normal_mul;
+ }
+ } else {
+ if (ei >= 254) {
+ w = EV_NMAX;
+ EV_SET_SPEFSCR_BITS(over);
+ } else if (ei < 1) {
+ d = sim_fpu_zero;
+ sim_fpu_to32 (&w, &d);
+ w |= 0x80000000; /* Set sign bit. */
+ } else {
+ normal_mul:
+ if (EV_IS_INFDENORMNAN(&a) || EV_IS_INFDENORMNAN(&b))
+ EV_SET_SPEFSCR_BITS(inv);
+ sim_fpu_mul (&d, &a, &b);
+ sim_fpu_to32 (&w, &d);
+ }
+ }
+ return w;
+
+:function:e500::unsigned32:ev_fs_div:unsigned32 aa, unsigned32 bb, int inv, int over, int under, int dbz, int fg, int fx, cpu *processor
+ sim_fpu a, b, d;
+ unsigned32 w;
+ int sa, sb, ea, eb, ei;
+
+ sim_fpu_32to (&a, aa);
+ sim_fpu_32to (&b, bb);
+ sa = sim_fpu_sign(&a);
+ sb = sim_fpu_sign(&b);
+ ea = booke_sim_fpu_exp(&a);
+ eb = booke_sim_fpu_exp(&b);
+ ei = ea - eb + 127;
+
+ /* Special cases to handle behaviour of e500 hardware.
+ cf case 107543. */
+ if (sim_fpu_is_nan (&a) || sim_fpu_is_nan (&b)
+ || sim_fpu_is_zero (&a) || sim_fpu_is_zero (&b))
+ {
+ if (sim_fpu_is_snan (&a) || sim_fpu_is_snan (&b))
+ {
+ if (bb == 0x3f800000)
+ w = EV_PMAX;
+ else if (aa == 0x7fc00001)
+ w = 0x3fbffffe;
+ else
+ goto normal_div;
+ }
+ else
+ goto normal_div;
+ }
+ else if (sim_fpu_is_infinity (&a) && sim_fpu_is_infinity (&b))
+ {
+ if (sa == sb)
+ sim_fpu_32to (&d, 0x3f800000);
+ else
+ sim_fpu_32to (&d, 0xbf800000);
+ sim_fpu_to32 (&w, &d);
+ }
+ else if (sa == sb) {
+ if (ei > 254) {
+ w = EV_PMAX;
+ EV_SET_SPEFSCR_BITS(over);
+ } else if (ei <= 1) {
+ d = sim_fpu_zero;
+ sim_fpu_to32 (&w, &d);
+ w &= 0x7fffffff; /* Clear sign bit. */
+ } else {
+ goto normal_div;
+ }
+ } else {
+ if (ei > 254) {
+ w = EV_NMAX;
+ EV_SET_SPEFSCR_BITS(over);
+ } else if (ei <= 1) {
+ d = sim_fpu_zero;
+ sim_fpu_to32 (&w, &d);
+ w |= 0x80000000; /* Set sign bit. */
+ } else {
+ normal_div:
+ if (EV_IS_INFDENORMNAN(&a) || EV_IS_INFDENORMNAN(&b))
+ EV_SET_SPEFSCR_BITS(inv);
+ if (sim_fpu_is_zero (&b))
+ {
+ if (sim_fpu_is_zero (&a))
+ EV_SET_SPEFSCR_BITS(dbz);
+ else
+ EV_SET_SPEFSCR_BITS(inv);
+ w = sa ? EV_NMAX : EV_PMAX;
+ }
+ else
+ {
+ sim_fpu_div (&d, &a, &b);
+ sim_fpu_to32 (&w, &d);
+ ev_check_guard(&d, fg, fx, processor);
+ }
+ }
+ }
+ return w;
+
+\f
+#
+# A.2.7 Integer SPE Simple Instructions
+#
+
+0.4,6.RS,11.RA,16.RB,21.512:X:e500:evaddw %RS,%RA,%RB:Vector Add Word
+ unsigned32 w1, w2;
+ w1 = *rBh + *rAh;
+ w2 = *rB + *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evaddw: *rSh = %08x; *rS = %08x; w1 = %08x w2 = %08x\n", *rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.IMM,16.RB,21.514:X:e500:evaddiw %RS,%RB,%IMM:Vector Add Immediate Word
+ unsigned32 w1, w2;
+ w1 = *rBh + IMM;
+ w2 = *rB + IMM;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evaddiw: *rSh = %08x; *rS = %08x; w1 = %08x w2 = %08x\n", *rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.516:X:e500:evsubfw %RS,%RA,%RB:Vector Subtract from Word
+ unsigned32 w1, w2;
+ w1 = *rBh - *rAh;
+ w2 = *rB - *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evsubfw: *rSh = %08x; *rS = %08x; w1 = %08x w2 = %08x\n", *rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.IMM,16.RB,21.518:X:e500:evsubifw %RS,%RB,%IMM:Vector Subtract Immediate from Word
+ unsigned32 w1, w2;
+ w1 = *rBh - IMM;
+ w2 = *rB - IMM;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evsubifw: *rSh = %08x; *rS = %08x; IMM = %d\n", *rSh, *rS, IMM);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.520:X:e500:evabs %RS,%RA:Vector Absolute Value
+ signed32 w1, w2;
+ w1 = *rAh;
+ if (w1 < 0 && w1 != 0x80000000)
+ w1 = -w1;
+ w2 = *rA;
+ if (w2 < 0 && w2 != 0x80000000)
+ w2 = -w2;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.521:X:e500:evneg %RS,%RA:Vector Negate
+ signed32 w1, w2;
+ w1 = *rAh;
+ /* the negative most negative number is the most negative number */
+ if (w1 != 0x80000000)
+ w1 = -w1;
+ w2 = *rA;
+ if (w2 != 0x80000000)
+ w2 = -w2;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.522:X:e500:evextsb %RS,%RA:Vector Extend Signed Byte
+ unsigned64 w1, w2;
+ w1 = *rAh & 0xff;
+ if (w1 & 0x80)
+ w1 |= 0xffffff00;
+ w2 = *rA & 0xff;
+ if (w2 & 0x80)
+ w2 |= 0xffffff00;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK , 0);
+
+0.4,6.RS,11.RA,16.0,21.523:X:e500:evextsb %RS,%RA:Vector Extend Signed Half Word
+ unsigned64 w1, w2;
+ w1 = *rAh & 0xffff;
+ if (w1 & 0x8000)
+ w1 |= 0xffff0000;
+ w2 = *rA & 0xffff;
+ if (w2 & 0x8000)
+ w2 |= 0xffff0000;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.529:X:e500:evand %RS,%RA,%RB:Vector AND
+ unsigned32 w1, w2;
+ w1 = *rBh & *rAh;
+ w2 = *rB & *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.535:X:e500:evor %RS,%RA,%RB:Vector OR
+ unsigned32 w1, w2;
+ w1 = *rBh | *rAh;
+ w2 = *rB | *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.534:X:e500:evxor %RS,%RA,%RB:Vector XOR
+ unsigned32 w1, w2;
+ w1 = *rBh ^ *rAh;
+ w2 = *rB ^ *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.542:X:e500:evnand %RS,%RA,%RB:Vector NAND
+ unsigned32 w1, w2;
+ w1 = ~(*rBh & *rAh);
+ w2 = ~(*rB & *rA);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.536:X:e500:evnor %RS,%RA,%RB:Vector NOR
+ unsigned32 w1, w2;
+ w1 = ~(*rBh | *rAh);
+ w2 = ~(*rB | *rA);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.537:X:e500:eveqv %RS,%RA,%RB:Vector Equivalent
+ unsigned32 w1, w2;
+ w1 = (~*rBh) ^ *rAh;
+ w2 = (~*rB) ^ *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.530:X:e500:evandc %RS,%RA,%RB:Vector AND with Compliment
+ unsigned32 w1, w2;
+ w1 = (~*rBh) & *rAh;
+ w2 = (~*rB) & *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evandc: *rSh = %08x; *rS = %08x\n", *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.539:X:e500:evorc %RS,%RA,%RB:Vector OR with Compliment
+ unsigned32 w1, w2;
+ w1 = (~*rBh) | *rAh;
+ w2 = (~*rB) | *rA;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evorc: *rSh = %08x; *rS = %08x\n", *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.552:X:e500:evrlw %RS,%RA,%RB:Vector Rotate Left Word
+ unsigned32 nh, nl, w1, w2;
+ nh = *rBh & 0x1f;
+ nl = *rB & 0x1f;
+ w1 = ((unsigned32)*rAh) << nh | ((unsigned32)*rAh) >> (32 - nh);
+ w2 = ((unsigned32)*rA) << nl | ((unsigned32)*rA) >> (32 - nl);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evrlw: nh %d nl %d *rSh = %08x; *rS = %08x\n", nh, nl, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.554:X:e500:evrlwi %RS,%RA,%UIMM:Vector Rotate Left Word Immediate
+ unsigned32 w1, w2, imm;
+ imm = (unsigned32)UIMM;
+ w1 = ((unsigned32)*rAh) << imm | ((unsigned32)*rAh) >> (32 - imm);
+ w2 = ((unsigned32)*rA) << imm | ((unsigned32)*rA) >> (32 - imm);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.548:X:e500:evslw %RS,%RA,%RB:Vector Shift Left Word
+ unsigned32 nh, nl, w1, w2;
+ nh = *rBh & 0x1f;
+ nl = *rB & 0x1f;
+ w1 = ((unsigned32)*rAh) << nh;
+ w2 = ((unsigned32)*rA) << nl;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.550:X:e500:evslwi %RS,%RA,%UIMM:Vector Shift Left Word Immediate
+ unsigned32 w1, w2, imm = UIMM;
+ w1 = ((unsigned32)*rAh) << imm;
+ w2 = ((unsigned32)*rA) << imm;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.545:X:e500:evsrws %RS,%RA,%RB:Vector Shift Right Word Signed
+ signed32 w1, w2;
+ unsigned32 nh, nl;
+ nh = *rBh & 0x1f;
+ nl = *rB & 0x1f;
+ w1 = ((signed32)*rAh) >> nh;
+ w2 = ((signed32)*rA) >> nl;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evsrws: nh %d nl %d *rSh = %08x; *rS = %08x\n", nh, nl, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.544:X:e500:evsrwu %RS,%RA,%RB:Vector Shift Right Word Unsigned
+ unsigned32 w1, w2, nh, nl;
+ nh = *rBh & 0x1f;
+ nl = *rB & 0x1f;
+ w1 = ((unsigned32)*rAh) >> nh;
+ w2 = ((unsigned32)*rA) >> nl;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.547:X:e500:evsrwis %RS,%RA,%UIMM:Vector Shift Right Word Immediate Signed
+ signed32 w1, w2;
+ unsigned32 imm = UIMM;
+ w1 = ((signed32)*rAh) >> imm;
+ w2 = ((signed32)*rA) >> imm;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.546:X:e500:evsrwiu %RS,%RA,%UIMM:Vector Shift Right Word Immediate Unsigned
+ unsigned32 w1, w2, imm = UIMM;
+ w1 = ((unsigned32)*rAh) >> imm;
+ w2 = ((unsigned32)*rA) >> imm;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.525:X:e500:evcntlzw %RS,%RA:Vector Count Leading Zeros Word
+ unsigned32 w1, w2, mask, c1, c2;
+ for (c1 = 0, mask = 0x80000000, w1 = *rAh;
+ !(w1 & mask) && mask != 0; mask >>= 1)
+ c1++;
+ for (c2 = 0, mask = 0x80000000, w2 = *rA;
+ !(w2 & mask) && mask != 0; mask >>= 1)
+ c2++;
+ EV_SET_REG2(*rSh, *rS, c1, c2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.526:X:e500:evcntlsw %RS,%RA:Vector Count Leading Sign Bits Word
+ unsigned32 w1, w2, mask, sign_bit, c1, c2;
+ for (c1 = 0, mask = 0x80000000, w1 = *rAh, sign_bit = w1 & mask;
+ ((w1 & mask) == sign_bit) && mask != 0;
+ mask >>= 1, sign_bit >>= 1)
+ c1++;
+ for (c2 = 0, mask = 0x80000000, w2 = *rA, sign_bit = w2 & mask;
+ ((w2 & mask) == sign_bit) && mask != 0;
+ mask >>= 1, sign_bit >>= 1)
+ c2++;
+ EV_SET_REG2(*rSh, *rS, c1, c2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.524:X:e500:evrndw %RS,%RA:Vector Round Word
+ unsigned32 w1, w2;
+ w1 = ((unsigned32)*rAh + 0x8000) & 0xffff0000;
+ w2 = ((unsigned32)*rA + 0x8000) & 0xffff0000;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ //printf("evrndw: *rSh = %08x; *rS = %08x\n", *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.556:X:e500:evmergehi %RS,%RA,%RB:Vector Merge Hi
+ unsigned32 w1, w2;
+ w1 = *rAh;
+ w2 = *rBh;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.557:X:e500:evmergelo %RS,%RA,%RB:Vector Merge Low
+ unsigned32 w1, w2;
+ w1 = *rA;
+ w2 = *rB;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.559:X:e500:evmergelohi %RS,%RA,%RB:Vector Merge Low Hi
+ unsigned32 w1, w2;
+ w1 = *rA;
+ w2 = *rBh;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.558:X:e500:evmergehilo %RS,%RA,%RB:Vector Merge Hi Low
+ unsigned32 w1, w2;
+ w1 = *rAh;
+ w2 = *rB;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.SIMM,16.0,21.553:X:e500:evsplati %RS,%SIMM:Vector Splat Immediate
+ unsigned32 w;
+ w = SIMM & 0x1f;
+ if (w & 0x10)
+ w |= 0xffffffe0;
+ EV_SET_REG2(*rSh, *rS, w, w);
+ PPC_INSN_INT(RS_BITMASK, 0, 0);
+
+0.4,6.RS,11.SIMM,16.0,21.555:X:e500:evsplatfi %RS,%SIMM:Vector Splat Fractional Immediate
+ unsigned32 w;
+ w = SIMM << 27;
+ EV_SET_REG2(*rSh, *rS, w, w);
+ PPC_INSN_INT(RS_BITMASK, 0, 0);
+
+0.4,6.BF,9.0,11.RA,16.RB,21.561:X:e500:evcmpgts %BF,%RA,%RB:Vector Compare Greater Than Signed
+ signed32 ah, al, bh, bl;
+ int w, ch, cl;
+ ah = *rAh;
+ al = *rA;
+ bh = *rBh;
+ bl = *rB;
+ if (ah > bh)
+ ch = 1;
+ else
+ ch = 0;
+ if (al > bl)
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9.0,11.RA,16.RB,21.560:X:e500:evcmpgtu %BF,%RA,%RB:Vector Compare Greater Than Unsigned
+ unsigned32 ah, al, bh, bl;
+ int w, ch, cl;
+ ah = *rAh;
+ al = *rA;
+ bh = *rBh;
+ bl = *rB;
+ if (ah > bh)
+ ch = 1;
+ else
+ ch = 0;
+ if (al > bl)
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9.0,11.RA,16.RB,21.563:X:e500:evcmplts %BF,%RA,%RB:Vector Compare Less Than Signed
+ signed32 ah, al, bh, bl;
+ int w, ch, cl;
+ ah = *rAh;
+ al = *rA;
+ bh = *rBh;
+ bl = *rB;
+ if (ah < bh)
+ ch = 1;
+ else
+ ch = 0;
+ if (al < bl)
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9.0,11.RA,16.RB,21.562:X:e500:evcmpltu %BF,%RA,%RB:Vector Compare Less Than Unsigned
+ unsigned32 ah, al, bh, bl;
+ int w, ch, cl;
+ ah = *rAh;
+ al = *rA;
+ bh = *rBh;
+ bl = *rB;
+ if (ah < bh)
+ ch = 1;
+ else
+ ch = 0;
+ if (al < bl)
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9.0,11.RA,16.RB,21.564:X:e500:evcmpeq %BF,%RA,%RB:Vector Compare Equal
+ unsigned32 ah, al, bh, bl;
+ int w, ch, cl;
+ ah = *rAh;
+ al = *rA;
+ bh = *rBh;
+ bl = *rB;
+ if (ah == bh)
+ ch = 1;
+ else
+ ch = 0;
+ if (al == bl)
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ //printf("evcmpeq: ch %d cl %d BF %d, CR is now %08x\n", ch, cl, BF, CR);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.RS,11.RA,16.RB,21.79,29.CRFS:X:e500:evsel %RS,%RA,%RB,%CRFS:Vector Select
+ unsigned32 w1, w2;
+ int cr;
+ cr = CR_FIELD(CRFS);
+ if (cr & 8)
+ w1 = *rAh;
+ else
+ w1 = *rBh;
+ if (cr & 4)
+ w2 = *rA;
+ else
+ w2 = *rB;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.527:X:e500:brinc %RS,%RA,%RB:Bit Reversed Increment
+ unsigned32 w1, w2, a, d, mask;
+ mask = (*rB) & 0xffff;
+ a = (*rA) & 0xffff;
+ d = EV_BITREVERSE16(1 + EV_BITREVERSE16(a | ~mask));
+ *rS = ((*rA) & 0xffff0000) | (d & 0xffff);
+ //printf("brinc: *rS = %08x\n", *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+#
+# A.2.8 Integer SPE Complex Instructions
+#
+
+0.4,6.RS,11.RA,16.RB,21.1031:EVX:e500:evmhossf %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Fractional
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int movl, movh;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ tl = ev_multiply16_ssf (al, bl, &movl);
+ th = ev_multiply16_ssf (ah, bh, &movh);
+ EV_SET_REG2 (*rSh, *rS, EV_SATURATE (movh, 0x7fffffff, th),
+ EV_SATURATE (movl, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl, movh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1063:EVX:e500:evmhossfa %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Fractional Accumulate
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int movl, movh;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ tl = ev_multiply16_ssf (al, bl, &movl);
+ th = ev_multiply16_ssf (ah, bh, &movh);
+ EV_SET_REG2 (*rSh, *rS, EV_SATURATE (movh, 0x7fffffff, th),
+ EV_SATURATE (movl, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl, movh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1039:EVX:e500:evmhosmf %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Fractional
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int dummy;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ tl = ev_multiply16_smf (al, bl, & dummy);
+ th = ev_multiply16_smf (ah, bh, & dummy);
+ EV_SET_REG2 (*rSh, *rS, th, tl);
+ PPC_INSN_INT (RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1071:EVX:e500:evmhosmfa %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Fractional Accumulate
+ signed32 al, ah, bl, bh;
+ signed32 tl, th;
+ int dummy;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ tl = ev_multiply16_smf (al, bl, & dummy);
+ th = ev_multiply16_smf (ah, bh, & dummy);
+ EV_SET_REG2_ACC (*rSh, *rS, th, tl);
+ PPC_INSN_INT (RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1037:EVX:e500:evmhosmi %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Integer
+ signed32 al, ah, bl, bh, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2(*rSh, *rS, th, tl);
+ //printf("evmhosmi: *rSh = %08x; *rS = %08x\n", *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1069:EVX:e500:evmhosmia %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Integer Accumulate
+ signed32 al, ah, bl, bh, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2_ACC(*rSh, *rS, th, tl);
+ //printf("evmhosmia: ACC = %08x; *rSh = %08x; *rS = %08x\n", ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1036:EVX:e500:evmhoumi %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Modulo Integer
+ unsigned32 al, ah, bl, bh, tl, th;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1068:EVX:e500:evmhoumia %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Modulo Integer Accumulate
+ unsigned32 al, ah, bl, bh, tl, th;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2_ACC(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1027:EVX:e500:evmhessf %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Fractional
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int movl, movh;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ tl = ev_multiply16_ssf (al, bl, &movl);
+ th = ev_multiply16_ssf (ah, bh, &movh);
+ EV_SET_REG2 (*rSh, *rS, EV_SATURATE (movh, 0x7fffffff, th),
+ EV_SATURATE (movl, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl, movh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1059:EVX:e500:evmhessfa %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Fractional Accumulate
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int movl, movh;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ tl = ev_multiply16_ssf (al, bl, &movl);
+ th = ev_multiply16_ssf (ah, bh, &movh);
+ EV_SET_REG2_ACC (*rSh, *rS, EV_SATURATE (movh, 0x7fffffff, th),
+ EV_SATURATE (movl, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl, movh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1035:EVX:e500:evmhesmf %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Fractional
+ signed16 al, ah, bl, bh;
+ signed64 tl, th;
+ int movl, movh;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ tl = ev_multiply16_smf (al, bl, &movl);
+ th = ev_multiply16_smf (ah, bh, &movh);
+ EV_SET_REG2 (*rSh, *rS, th, tl);
+ EV_SET_SPEFSCR_OV (movl, movh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1067:EVX:e500:evmhesmfa %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Fractional Accumulate
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+ int dummy;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ tl = ev_multiply16_smf (al, bl, & dummy);
+ th = ev_multiply16_smf (ah, bh, & dummy);
+ EV_SET_REG2_ACC (*rSh, *rS, th, tl);
+ PPC_INSN_INT (RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1033:EVX:e500:evmhesmi %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Integer
+ signed16 al, ah, bl, bh;
+ signed32 tl, th;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2 (*rSh, *rS, th, tl);
+ PPC_INSN_INT (RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1065:EVX:e500:evmhesmia %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Integer Accumulate
+ signed32 al, ah, bl, bh, tl, th;
+ al = (signed32)(signed16)EV_HIHALF(*rA);
+ ah = (signed32)(signed16)EV_HIHALF(*rAh);
+ bl = (signed32)(signed16)EV_HIHALF(*rB);
+ bh = (signed32)(signed16)EV_HIHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2_ACC(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1032:EVX:e500:evmheumi %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Modulo Integer
+ unsigned32 al, ah, bl, bh, tl, th;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1064:EVX:e500:evmheumia %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Modulo Integer Accumulate
+ unsigned32 al, ah, bl, bh, tl, th;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ tl = al * bl;
+ th = ah * bh;
+ EV_SET_REG2_ACC(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1287:EVX:e500:evmhossfaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Fractional and Accumulate into Words
+ signed16 al, ah, bl, bh;
+ signed32 t1, t2;
+ signed64 tl, th;
+ int movl, movh, ovl, ovh;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ t1 = ev_multiply16_ssf (ah, bh, &movh);
+ t2 = ev_multiply16_ssf (al, bl, &movl);
+ th = EV_ACCHIGH + EV_SATURATE (movh, 0x7fffffff, t1);
+ tl = EV_ACCLOW + EV_SATURATE (movl, 0x7fffffff, t2);
+ ovh = EV_SAT_P_S32 (th);
+ ovl = EV_SAT_P_S32 (tl);
+ EV_SET_REG2_ACC (*rSh, *rS, EV_SATURATE_ACC (ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC (ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1285:EVX:e500:evmhossiaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Integer and Accumulate into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ //printf("evmhossiaaw: ovh %d ovl %d al %d ah %d bl %d bh %d t1 %qd t2 %qd tl %qd th %qd\n", ovh, ovl, al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhossiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1295:EVX:e500:evmhosmfaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Fractional and Accumulate into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ((signed64)ah * bh) << 1;
+ t2 = ((signed64)al * bl) << 1;
+ th = EV_ACCHIGH + (t1 & 0xffffffff);
+ tl = EV_ACCLOW + (t2 & 0xffffffff);
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1293:EVX:e500:evmhosmiaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Integer and Accumulate into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ //printf("evmhosmiaaw: al %d ah %d bl %d bh %d t1 %qd t2 %qd tl %qd th %qd\n", al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhosmiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1284:EVX:e500:evmhousiaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Saturate Integer and Accumulate into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ signed64 tl, th;
+ int ovl, ovh;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = (signed64)EV_ACCHIGH + (signed64)t1;
+ tl = (signed64)EV_ACCLOW + (signed64)t2;
+ ovh = EV_SAT_P_U32(th);
+ ovl = EV_SAT_P_U32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0, 0xffffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0, 0xffffffff, tl));
+ //printf("evmhousiaaw: al %u ah %u bl %u bh %u t1 %qu t2 %qu tl %qu th %qu\n", al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhousiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1292:EVX:e500:evmhoumiaaw %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Modulo Integer and Accumulate into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned32 t1, t2;
+ signed64 tl, th;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ //printf("evmhoumiaaw: al %u ah %u bl %u bh %u t1 %qu t2 %qu tl %qu th %qu\n", al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhoumiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1283:EVX:e500:evmhessfaaw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Fractional and Accumulate into Words
+ signed16 al, ah, bl, bh;
+ signed32 t1, t2;
+ signed64 tl, th;
+ int movl, movh, ovl, ovh;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ t1 = ev_multiply16_ssf (ah, bh, &movh);
+ t2 = ev_multiply16_ssf (al, bl, &movl);
+ th = EV_ACCHIGH + EV_SATURATE (movh, 0x7fffffff, t1);
+ tl = EV_ACCLOW + EV_SATURATE (movl, 0x7fffffff, t2);
+ ovh = EV_SAT_P_S32 (th);
+ ovl = EV_SAT_P_S32 (tl);
+ EV_SET_REG2_ACC (*rSh, *rS, EV_SATURATE_ACC (ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC (ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1281:EVX:e500:evmhessiaaw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Integer and Accumulate into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = (signed32)(signed16)EV_HIHALF(*rA);
+ ah = (signed32)(signed16)EV_HIHALF(*rAh);
+ bl = (signed32)(signed16)EV_HIHALF(*rB);
+ bh = (signed32)(signed16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ //printf("evmhessiaaw: ovh %d ovl %d al %d ah %d bl %d bh %d t1 %qd t2 %qd tl %qd th %qd\n", ovh, ovl, al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhessiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1291:EVX:e500:evmhesmfaaw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Fractional and Accumulate into Words
+ signed16 al, ah, bl, bh;
+ signed32 t1, t2, th, tl;
+ int dummy;
+
+ al = (signed16)EV_HIHALF(*rA);
+ ah = (signed16)EV_HIHALF(*rAh);
+ bl = (signed16)EV_HIHALF(*rB);
+ bh = (signed16)EV_HIHALF(*rBh);
+ t1 = ev_multiply16_smf (ah, bh, &dummy);
+ t2 = ev_multiply16_smf (al, bl, &dummy);
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th, tl);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1289:EVX:e500:evmhesmiaaw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Integer and Accumulate into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_HIHALF(*rA);
+ ah = (signed32)(signed16)EV_HIHALF(*rAh);
+ bl = (signed32)(signed16)EV_HIHALF(*rB);
+ bh = (signed32)(signed16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1280:EVX:e500:evmheusiaaw %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Saturate Integer and Accumulate into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ signed64 tl, th;
+ int ovl, ovh;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = (signed64)EV_ACCHIGH + (signed64)t1;
+ tl = (signed64)EV_ACCLOW + (signed64)t2;
+ ovh = EV_SAT_P_U32(th);
+ ovl = EV_SAT_P_U32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0, 0xffffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0, 0xffffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1288:EVX:e500:evmheumiaaw %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Modulo Integer and Accumulate into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned32 t1, t2;
+ unsigned64 tl, th;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH + t1;
+ tl = EV_ACCLOW + t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1415:EVX:e500:evmhossfanw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Fractional and Accumulate Negative into Words
+ signed16 al, ah, bl, bh;
+ signed32 t1, t2;
+ signed64 tl, th;
+ int movl, movh, ovl, ovh;
+
+ al = (signed16) EV_LOHALF (*rA);
+ ah = (signed16) EV_LOHALF (*rAh);
+ bl = (signed16) EV_LOHALF (*rB);
+ bh = (signed16) EV_LOHALF (*rBh);
+ t1 = ev_multiply16_ssf (ah, bh, &movh);
+ t2 = ev_multiply16_ssf (al, bl, &movl);
+ th = EV_ACCHIGH - EV_SATURATE (movh, 0x7fffffff, t1);
+ tl = EV_ACCLOW - EV_SATURATE (movl, 0x7fffffff, t2);
+ ovh = EV_SAT_P_S32 (th);
+ ovl = EV_SAT_P_S32 (tl);
+ EV_SET_REG2_ACC (*rSh, *rS, EV_SATURATE_ACC (ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC (ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1413:EVX:e500:evmhossianw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Saturate Integer and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ //printf("evmhossianw: ACC = %08x; *rSh = %08x; *rS = %08x\n", ACC, *rSh, *rS);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1423:EVX:e500:evmhosmfanw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Fractional and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ((signed64)ah * bh) << 1;
+ t2 = ((signed64)al * bl) << 1;
+ th = EV_ACCHIGH - (t1 & 0xffffffff);
+ tl = EV_ACCLOW - (t2 & 0xffffffff);
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1421:EVX:e500:evmhosmianw %RS,%RA,%RB:Vector Multiply Half Words Odd Signed Modulo Integer and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_LOHALF(*rA);
+ ah = (signed32)(signed16)EV_LOHALF(*rAh);
+ bl = (signed32)(signed16)EV_LOHALF(*rB);
+ bh = (signed32)(signed16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1412:EVX:e500:evmhousianw %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Saturate Integer and Accumulate Negative into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ signed64 tl, th;
+ int ovl, ovh;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = (signed64)EV_ACCHIGH - (signed64)t1;
+ tl = (signed64)EV_ACCLOW - (signed64)t2;
+ ovl = EV_SAT_P_U32(tl);
+ ovh = EV_SAT_P_U32(th);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0, 0xffffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0, 0xffffffff, tl));
+ //printf("evmhousianw: ovh %d ovl %d al %d ah %d bl %d bh %d t1 %qd t2 %qd tl %qd th %qd\n", ovh, ovl, al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmoussianw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1420:EVX:e500:evmhoumianw %RS,%RA,%RB:Vector Multiply Half Words Odd Unsigned Modulo Integer and Accumulate Negative into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned32 t1, t2;
+ unsigned64 tl, th;
+ al = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_LOHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_LOHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1411:EVX:e500:evmhessfanw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Fractional and Accumulate Negative into Words
+ signed16 al, ah, bl, bh;
+ signed32 t1, t2;
+ signed64 tl, th;
+ int movl, movh, ovl, ovh;
+
+ al = (signed16) EV_HIHALF (*rA);
+ ah = (signed16) EV_HIHALF (*rAh);
+ bl = (signed16) EV_HIHALF (*rB);
+ bh = (signed16) EV_HIHALF (*rBh);
+ t1 = ev_multiply16_ssf (ah, bh, &movh);
+ t2 = ev_multiply16_ssf (al, bl, &movl);
+ th = EV_ACCHIGH - EV_SATURATE (movh, 0x7fffffff, t1);
+ tl = EV_ACCLOW - EV_SATURATE (movl, 0x7fffffff, t2);
+ ovh = EV_SAT_P_S32 (th);
+ ovl = EV_SAT_P_S32 (tl);
+ EV_SET_REG2_ACC (*rSh, *rS, EV_SATURATE_ACC (ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC (ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV (movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR (RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1409:EVX:e500:evmhessianw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Saturate Integer and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = (signed32)(signed16)EV_HIHALF(*rA);
+ ah = (signed32)(signed16)EV_HIHALF(*rAh);
+ bl = (signed32)(signed16)EV_HIHALF(*rB);
+ bh = (signed32)(signed16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1419:EVX:e500:evmhesmfanw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Fractional and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ t1 = ((signed64)ah * bh) << 1;
+ t2 = ((signed64)al * bl) << 1;
+ th = EV_ACCHIGH - (t1 & 0xffffffff);
+ tl = EV_ACCLOW - (t2 & 0xffffffff);
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1417:EVX:e500:evmhesmianw %RS,%RA,%RB:Vector Multiply Half Words Even Signed Modulo Integer and Accumulate Negative into Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ al = (signed32)(signed16)EV_HIHALF(*rA);
+ ah = (signed32)(signed16)EV_HIHALF(*rAh);
+ bl = (signed32)(signed16)EV_HIHALF(*rB);
+ bh = (signed32)(signed16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ //printf("evmhesmianw: al %d ah %d bl %d bh %d t1 %qd t2 %qd tl %qd th %qd\n", al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmhesmianw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1408:EVX:e500:evmheusianw %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Saturate Integer and Accumulate Negative into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ signed64 tl, th;
+ int ovl, ovh;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = (signed64)EV_ACCHIGH - (signed64)t1;
+ tl = (signed64)EV_ACCLOW - (signed64)t2;
+ ovl = EV_SAT_P_U32(tl);
+ ovh = EV_SAT_P_U32(th);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0, 0xffffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0, 0xffffffff, tl));
+ //printf("evmheusianw: ovh %d ovl %d al %u ah %u bl %u bh %u t1 %qu t2 %qu tl %qd th %qd\n", ovh, ovl, al, ah, bl, bh, t1, t2, tl, th);
+ //printf("evmheusianw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1416:EVX:e500:evmheumianw %RS,%RA,%RB:Vector Multiply Half Words Even Unsigned Modulo Integer and Accumulate Negative into Words
+ unsigned32 al, ah, bl, bh;
+ unsigned32 t1, t2;
+ unsigned64 tl, th;
+ al = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ ah = (unsigned32)(unsigned16)EV_HIHALF(*rAh);
+ bl = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ bh = (unsigned32)(unsigned16)EV_HIHALF(*rBh);
+ t1 = ah * bh;
+ t2 = al * bl;
+ th = EV_ACCHIGH - t1;
+ tl = EV_ACCLOW - t2;
+ EV_SET_REG2_ACC(*rSh, *rS, th & 0xffffffff, tl & 0xffffffff);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1327:EVX:e500:evmhogsmfaa %RS,%RA,%RB:Multiply Half Words Odd Guarded Signed Modulo Fractional and Accumulate
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_LOHALF(*rA);
+ b = (signed32)(signed16)EV_LOHALF(*rB);
+ t1 = EV_MUL16_SSF(a, b);
+ if (t1 & ((unsigned64)1 << 32))
+ t1 |= 0xfffffffe00000000;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1325:EVX:e500:evmhogsmiaa %RS,%RA,%RB:Multiply Half Words Odd Guarded Signed Modulo Integer and Accumulate
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_LOHALF(*rA);
+ b = (signed32)(signed16)EV_LOHALF(*rB);
+ t1 = (signed64)a * (signed64)b;
+ t2 = (signed64)ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ //printf("evmhogsmiaa: a %d b %d t1 %qd t2 %qd\n", a, b, t1, t2);
+ //printf("evmhogsmiaa: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1324:EVX:e500:evmhogumiaa %RS,%RA,%RB:Multiply Half Words Odd Guarded Unsigned Modulo Integer and Accumulate
+ unsigned32 a, b;
+ unsigned64 t1, t2;
+ a = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ b = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ t1 = a * b;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1323:EVX:e500:evmhegsmfaa %RS,%RA,%RB:Multiply Half Words Even Guarded Signed Modulo Fractional and Accumulate
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_HIHALF(*rA);
+ b = (signed32)(signed16)EV_HIHALF(*rB);
+ t1 = EV_MUL16_SSF(a, b);
+ if (t1 & ((unsigned64)1 << 32))
+ t1 |= 0xfffffffe00000000;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1321:EVX:e500:evmhegsmiaa %RS,%RA,%RB:Multiply Half Words Even Guarded Signed Modulo Integer and Accumulate
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_HIHALF(*rA);
+ b = (signed32)(signed16)EV_HIHALF(*rB);
+ t1 = (signed64)(a * b);
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1320:EVX:e500:evmhegumiaa %RS,%RA,%RB:Multiply Half Words Even Guarded Unsigned Modulo Integer and Accumulate
+ unsigned32 a, b;
+ unsigned64 t1, t2;
+ a = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ b = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ t1 = a * b;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1455:EVX:e500:evmhogsmfan %RS,%RA,%RB:Multiply Half Words Odd Guarded Signed Modulo Fractional and Accumulate Negative
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_LOHALF(*rA);
+ b = (signed32)(signed16)EV_LOHALF(*rB);
+ t1 = EV_MUL16_SSF(a, b);
+ if (t1 & ((unsigned64)1 << 32))
+ t1 |= 0xfffffffe00000000;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1453:EVX:e500:evmhogsmian %RS,%RA,%RB:Multiply Half Words Odd Guarded Signed Modulo Integer and Accumulate Negative
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_LOHALF(*rA);
+ b = (signed32)(signed16)EV_LOHALF(*rB);
+ t1 = (signed64)a * (signed64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ //printf("evmhogsmian: a %d b %d t1 %qd t2 %qd\n", a, b, t1, t2);
+ //printf("evmhogsmian: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1452:EVX:e500:evmhogumian %RS,%RA,%RB:Multiply Half Words Odd Guarded Unsigned Modulo Integer and Accumulate Negative
+ unsigned32 a, b;
+ unsigned64 t1, t2;
+ a = (unsigned32)(unsigned16)EV_LOHALF(*rA);
+ b = (unsigned32)(unsigned16)EV_LOHALF(*rB);
+ t1 = (unsigned64)a * (unsigned64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1451:EVX:e500:evmhegsmfan %RS,%RA,%RB:Multiply Half Words Even Guarded Signed Modulo Fractional and Accumulate Negative
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_HIHALF(*rA);
+ b = (signed32)(signed16)EV_HIHALF(*rB);
+ t1 = EV_MUL16_SSF(a, b);
+ if (t1 & ((unsigned64)1 << 32))
+ t1 |= 0xfffffffe00000000;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1449:EVX:e500:evmhegsmian %RS,%RA,%RB:Multiply Half Words Even Guarded Signed Modulo Integer and Accumulate Negative
+ signed32 a, b;
+ signed64 t1, t2;
+ a = (signed32)(signed16)EV_HIHALF(*rA);
+ b = (signed32)(signed16)EV_HIHALF(*rB);
+ t1 = (signed64)a * (signed64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1448:EVX:e500:evmhegumian %RS,%RA,%RB:Multiply Half Words Even Guarded Unsigned Modulo Integer and Accumulate Negative
+ unsigned32 a, b;
+ unsigned64 t1, t2;
+ a = (unsigned32)(unsigned16)EV_HIHALF(*rA);
+ b = (unsigned32)(unsigned16)EV_HIHALF(*rB);
+ t1 = (unsigned64)a * (unsigned64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1095:EVX:e500:evmwhssf %RS,%RA,%RB:Vector Multiply Word High Signed Saturate Fractional
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int movl, movh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(al, bl, &movl);
+ t2 = ev_multiply32_ssf(ah, bh, &movh);
+ EV_SET_REG2(*rSh, *rS, EV_SATURATE(movh, 0x7fffffff, t2 >> 32),
+ EV_SATURATE(movl, 0x7fffffff, t1 >> 32));
+ EV_SET_SPEFSCR_OV(movl, movh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1127:EVX:e500:evmwhssfa %RS,%RA,%RB:Vector Multiply Word High Signed Saturate Fractional and Accumulate
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int movl, movh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(al, bl, &movl);
+ t2 = ev_multiply32_ssf(ah, bh, &movh);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(movh, 0x7fffffff, t2 >> 32),
+ EV_SATURATE(movl, 0x7fffffff, t1 >> 32));
+ EV_SET_SPEFSCR_OV(movl, movh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1103:EVX:e500:evmwhsmf %RS,%RA,%RB:Vector Multiply Word High Signed Modulo Fractional
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = EV_MUL32_SSF(al, bl);
+ t2 = EV_MUL32_SSF(ah, bh);
+ EV_SET_REG2(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1135:EVX:e500:evmwhsmfa %RS,%RA,%RB:Vector Multiply Word High Signed Modulo Fractional and Accumulate
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = EV_MUL32_SSF(al, bl);
+ t2 = EV_MUL32_SSF(ah, bh);
+ EV_SET_REG2_ACC(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1101:EVX:e500:evmwhsmi %RS,%RA,%RB:Vector Multiply Word High Signed Modulo Integer
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)al * (signed64)bl;
+ t2 = (signed64)ah * (signed64)bh;
+ EV_SET_REG2(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1133:EVX:e500:evmwhsmia %RS,%RA,%RB:Vector Multiply Word High Signed Modulo Integer and Accumulate
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)al * (signed64)bl;
+ t2 = (signed64)ah * (signed64)bh;
+ EV_SET_REG2_ACC(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1100:EVX:e500:evmwhumi %RS,%RA,%RB:Vector Multiply Word High Unsigned Modulo Integer
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)al * (unsigned64)bl;
+ t2 = (unsigned64)ah * (unsigned64)bh;
+ EV_SET_REG2(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1132:EVX:e500:evmwhumia %RS,%RA,%RB:Vector Multiply Word High Unsigned Modulo Integer and Accumulate
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)al * (unsigned64)bl;
+ t2 = (unsigned64)ah * (unsigned64)bh;
+ EV_SET_REG2_ACC(*rSh, *rS, t2 >> 32, t1 >> 32);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1091:EVX:e500:evmwlssf %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Fractional
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int movl, movh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(al, bl, &movl);
+ t2 = ev_multiply32_ssf(ah, bh, &movh);
+ EV_SET_REG2(*rSh, *rS, EV_SATURATE(movh, 0xffffffff, t2),
+ EV_SATURATE(movl, 0xffffffff, t1));
+ EV_SET_SPEFSCR_OV(movl, movh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1123:EVX:e500:evmwlssfa %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Fractional and Accumulate
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int movl, movh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(al, bl, &movl);
+ t2 = ev_multiply32_ssf(ah, bh, &movh);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(movh, 0xffffffff, t2),
+ EV_SATURATE(movl, 0xffffffff, t1));
+ EV_SET_SPEFSCR_OV(movl, movh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1099:EVX:e500:evmwlsmf %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Fractional
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = EV_MUL32_SSF(al, bl);
+ t2 = EV_MUL32_SSF(ah, bh);
+ EV_SET_REG2(*rSh, *rS, t2, t1);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1131:EVX:e500:evmwlsmfa %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Fractional and Accumulate
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = EV_MUL32_SSF(al, bl);
+ t2 = EV_MUL32_SSF(ah, bh);
+ EV_SET_REG2_ACC(*rSh, *rS, t2, t1);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1096:EVX:e500:evmwlumi %RS,%RA,%RB:Vector Multiply Word Low Unsigned Modulo Integer
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)al * (unsigned64)bl;
+ t2 = (unsigned64)ah * (unsigned64)bh;
+ EV_SET_REG2(*rSh, *rS, t2, t1);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1128:EVX:e500:evmwlumia %RS,%RA,%RB:Vector Multiply Word Low Unsigned Modulo Integer and Accumulate
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)al * (unsigned64)bl;
+ t2 = (unsigned64)ah * (unsigned64)bh;
+ EV_SET_REG2_ACC(*rSh, *rS, t2, t1);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1347:EVX:e500:evmwlssfaaw %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Fractional and Accumulate in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int movl, movh, ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(ah, bh, &movh);
+ t2 = ev_multiply32_ssf(al, bl, &movl);
+ th = EV_ACCHIGH + EV_SATURATE(movh, 0xffffffff, t1);
+ tl = EV_ACCLOW + EV_SATURATE(movl, 0xffffffff, t2);
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1345:EVX:e500:evmwlssiaaw %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Integer and Accumulate in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)ah * (signed64)bh;
+ t2 = (signed64)al * (signed64)bl;
+ th = EV_ACCHIGH + (t1 & 0xffffffff);
+ tl = EV_ACCLOW + (t2 & 0xffffffff);
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1355:EVX:e500:evmwlsmfaaw %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Fractional and Accumulate in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int mov;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_smf(ah, bh, &mov);
+ t2 = ev_multiply32_smf(al, bl, &mov);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH + (t1 & 0xffffffff),
+ EV_ACCLOW + (t2 & 0xffffffff));
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1353:EVX:e500:evmwlsmiaaw %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Integer and Accumulate in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)ah * (signed64)bh;
+ t2 = (signed64)al * (signed64)bl;
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH + (t1 & 0xffffffff),
+ EV_ACCLOW + (t2 & 0xffffffff));
+ //printf("evmwlsmiaaw: al %d ah %d bl %d bh %d t1 %qd t2 %qd\n", al, ah, bl, bh, t1, t2);
+ //printf("evmwlsmiaaw: *rSh = %08x; *rS = %08x\n", *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1344:EVX:e500:evmwlusiaaw %RS,%RA,%RB:Vector Multiply Word Low Unsigned Saturate Integer and Accumulate in Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)ah * (unsigned64)bh;
+ t2 = (unsigned64)al * (unsigned64)bl;
+ th = EV_ACCHIGH + (t1 & 0xffffffff);
+ tl = EV_ACCLOW + (t2 & 0xffffffff);
+ ovh = (th >> 32);
+ ovl = (tl >> 32);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(ovh, 0xffffffff, th),
+ EV_SATURATE(ovl, 0xffffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1352:EVX:e500:evmwlumiaaw %RS,%RA,%RB:Vector Multiply Word Low Unsigned Modulo Integer and Accumulate in Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)ah * (unsigned64)bh;
+ t2 = (unsigned64)al * (unsigned64)bl;
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH + (t1 & 0xffffffff),
+ EV_ACCLOW + (t2 & 0xffffffff));
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1475:EVX:e500:evmwlssfanw %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Fractional and Accumulate Negative in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int movl, movh, ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_ssf(ah, bh, &movh);
+ t2 = ev_multiply32_ssf(al, bl, &movl);
+ th = EV_ACCHIGH - EV_SATURATE(movh, 0xffffffff, t1);
+ tl = EV_ACCLOW - EV_SATURATE(movl, 0xffffffff, t2);
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(movl | ovl, movh | ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1473:EVX:e500:evmwlssianw %RS,%RA,%RB:Vector Multiply Word Low Signed Saturate Integer and Accumulate Negative in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)ah * (signed64)bh;
+ t2 = (signed64)al * (signed64)bl;
+ th = EV_ACCHIGH - (t1 & 0xffffffff);
+ tl = EV_ACCLOW - (t2 & 0xffffffff);
+ ovh = EV_SAT_P_S32(th);
+ ovl = EV_SAT_P_S32(tl);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, th, 0x80000000, 0x7fffffff, th),
+ EV_SATURATE_ACC(ovl, tl, 0x80000000, 0x7fffffff, tl));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1483:EVX:e500:evmwlsmfanw %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Fractional and Accumulate Negative in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ int mov;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = ev_multiply32_smf(ah, bh, &mov);
+ t2 = ev_multiply32_smf(al, bl, &mov);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH - (t1 & 0xffffffff),
+ EV_ACCLOW - (t2 & 0xffffffff));
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1481:EVX:e500:evmwlsmianw %RS,%RA,%RB:Vector Multiply Word Low Signed Modulo Integer and Accumulate Negative in Words
+ signed32 al, ah, bl, bh;
+ signed64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (signed64)ah * (signed64)bh;
+ t2 = (signed64)al * (signed64)bl;
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH - (t1 & 0xffffffff),
+ EV_ACCLOW - (t2 & 0xffffffff));
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1472:EVX:e500:evmwlusianw %RS,%RA,%RB:Vector Multiply Word Low Unsigned Saturate Integer and Accumulate Negative in Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2, tl, th;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)ah * (unsigned64)bh;
+ t2 = (unsigned64)al * (unsigned64)bl;
+ th = EV_ACCHIGH - (t1 & 0xffffffff);
+ tl = EV_ACCLOW - (t2 & 0xffffffff);
+ ovh = (th >> 32);
+ ovl = (tl >> 32);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(ovh, 0xffffffff, th),
+ EV_SATURATE(ovl, 0xffffffff, tl));
+ //printf("evmwlusianw: ovl %d ovh %d al %d ah %d bl %d bh %d t1 %qd t2 %qd th %qd tl %qd\n", ovl, ovh, al, ah, al, bh, t1, t2, th, tl);
+ //printf("evmwlusianw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1480:EVX:e500:evmwlumianw %RS,%RA,%RB:Vector Multiply Word Low Unsigned Modulo Integer and Accumulate Negative in Words
+ unsigned32 al, ah, bl, bh;
+ unsigned64 t1, t2;
+ al = *rA;
+ ah = *rAh;
+ bl = *rB;
+ bh = *rBh;
+ t1 = (unsigned64)ah * (unsigned64)bh;
+ t2 = (unsigned64)al * (unsigned64)bl;
+ EV_SET_REG2_ACC(*rSh, *rS, EV_ACCHIGH - (t1 & 0xffffffff),
+ EV_ACCLOW - (t2 & 0xffffffff));
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1107:EVX:e500:evmwssf %RS,%RA,%RB:Vector Multiply Word Signed Saturate Fractional
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = ev_multiply32_ssf(a, b, &movl);
+ EV_SET_REG1(*rSh, *rS, EV_SATURATE(movl, 0x7fffffffffffffff, t));
+ EV_SET_SPEFSCR_OV(movl, 0);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1139:EVX:e500:evmwssfa %RS,%RA,%RB:Vector Multiply Word Signed Saturate Fractional and Accumulate
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = ev_multiply32_ssf(a, b, &movl);
+ EV_SET_REG1_ACC(*rSh, *rS, EV_SATURATE(movl, 0x7fffffffffffffff, t));
+ EV_SET_SPEFSCR_OV(movl, 0);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1115:EVX:e500:evmwsmf %RS,%RA,%RB:Vector Multiply Word Signed Modulo Fractional
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = ev_multiply32_smf(a, b, &movl);
+ EV_SET_REG1(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1147:EVX:e500:evmwsmfa %RS,%RA,%RB:Vector Multiply Word Signed Modulo Fractional and Accumulate
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = ev_multiply32_smf(a, b, &movl);
+ EV_SET_REG1_ACC(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1113:EVX:e500:evmwsmi %RS,%RA,%RB:Vector Multiply Word Signed Modulo Integer
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = (signed64)a * (signed64)b;
+ EV_SET_REG1(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1145:EVX:e500:evmwsmia %RS,%RA,%RB:Vector Multiply Word Signed Modulo Integer and Accumulate
+ signed32 a, b;
+ signed64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = (signed64)a * (signed64)b;
+ EV_SET_REG1_ACC(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1112:EVX:e500:evmwumi %RS,%RA,%RB:Vector Multiply Word Unigned Modulo Integer
+ unsigned32 a, b;
+ unsigned64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = (signed64)a * (signed64)b;
+ EV_SET_REG1(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1144:EVX:e500:evmwumia %RS,%RA,%RB:Vector Multiply Word Unigned Modulo Integer and Accumulate
+ unsigned32 a, b;
+ unsigned64 t;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t = (signed64)a * (signed64)b;
+ EV_SET_REG1_ACC(*rSh, *rS, t);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1363:EVX:e500:evmwssfaa %RS,%RA,%RB:Vector Multiply Word Signed Saturate Fractional Add and Accumulate
+ signed64 t1, t2;
+ signed32 a, b;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t1 = ev_multiply32_ssf(a, b, &movl);
+ t2 = ACC + EV_SATURATE(movl, 0x7fffffffffffffff, t1);
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ EV_SET_SPEFSCR_OV(movl, 0);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1371:EVX:e500:evmwsmfaa %RS,%RA,%RB:Vector Multiply Word Signed Modulo Fractional Add and Accumulate
+ signed64 t1, t2;
+ signed32 a, b;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t1 = ev_multiply32_smf(a, b, &movl);
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1369:EVX:e500:evmwsmiaa %RS,%RA,%RB:Vector Multiply Word Signed Modulo Integer And and Accumulate
+ signed64 t1, t2;
+ signed32 a, b;
+ a = *rA;
+ b = *rB;
+ t1 = (signed64)a * (signed64)b;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1368:EVX:e500:evmwumiaa %RS,%RA,%RB:Vector Multiply Word Unsigned Modulo Integer Add and Accumulate
+ unsigned64 t1, t2;
+ unsigned32 a, b;
+ a = *rA;
+ b = *rB;
+ t1 = (unsigned64)a * (unsigned64)b;
+ t2 = ACC + t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.RB,21.1491:EVX:e500:evmwssfan %RS,%RA,%RB:Vector Multiply Word Signed Saturate Fractional and Accumulate Negative
+ signed64 t1, t2;
+ signed32 a, b;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t1 = ev_multiply32_ssf(a, b, &movl);
+ t2 = ACC - EV_SATURATE(movl, 0x7fffffffffffffff, t1);
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ EV_SET_SPEFSCR_OV(movl, 0);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.1499:EVX:e500:evmwsmfan %RS,%RA,%RB:Vector Multiply Word Signed Modulo Fractional and Accumulate Negative
+ signed64 t1, t2;
+ signed32 a, b;
+ int movl;
+ a = *rA;
+ b = *rB;
+ t1 = ev_multiply32_smf(a, b, &movl);
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1497:EVX:e500:evmwsmian %RS,%RA,%RB:Vector Multiply Word Signed Modulo Integer and Accumulate Negative
+ signed64 t1, t2;
+ signed32 a, b;
+ a = *rA;
+ b = *rB;
+ t1 = (signed64)a * (signed64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1496:EVX:e500:evmwumian %RS,%RA,%RB:Vector Multiply Word Unsigned Modulo Integer and Accumulate Negative
+ unsigned64 t1, t2;
+ unsigned32 a, b;
+ a = *rA;
+ b = *rB;
+ t1 = (unsigned64)a * (unsigned64)b;
+ t2 = ACC - t1;
+ EV_SET_REG1_ACC(*rSh, *rS, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.0,21.1217:EVX:e500:evaddssiaaw %RS,%RA:Vector Add Signed Saturate Integer to Accumulator Word
+ signed64 t1, t2;
+ signed32 al, ah;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH + (signed64)ah;
+ t2 = (signed64)EV_ACCLOW + (signed64)al;
+ ovh = EV_SAT_P_S32(t1);
+ ovl = EV_SAT_P_S32(t2);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, t1 & ((unsigned64)1 << 32), 0x80000000, 0x7fffffff, t1),
+ EV_SATURATE_ACC(ovl, t2 & ((unsigned64)1 << 32), 0x80000000, 0x7fffffff, t2));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.0,21.1225:EVX:e500:evaddsmiaaw %RS,%RA:Vector Add Signed Modulo Integer to Accumulator Word
+ signed64 t1, t2;
+ signed32 al, ah;
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH + (signed64)ah;
+ t2 = (signed64)EV_ACCLOW + (signed64)al;
+ EV_SET_REG2_ACC(*rSh, *rS, t1, t2);
+ //printf("evaddsmiaaw: al %d ah %d t1 %qd t2 %qd\n", al, ah, t1, t2);
+ //printf("evaddsmiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.1216:EVX:e500:evaddusiaaw %RS,%RA:Vector Add Unsigned Saturate Integer to Accumulator Word
+ signed64 t1, t2;
+ unsigned32 al, ah;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH + (signed64)ah;
+ t2 = (signed64)EV_ACCLOW + (signed64)al;
+ ovh = EV_SAT_P_U32(t1);
+ ovl = EV_SAT_P_U32(t2);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(ovh, 0xffffffff, t1),
+ EV_SATURATE(ovl, 0xffffffff, t2));
+ //printf("evaddusiaaw: ovl %d ovh %d al %d ah %d t1 %qd t2 %qd\n", ovl, ovh, al, ah, t1, t2);
+ //printf("evaddusiaaw: ACC = %08x.%08x; *rSh = %08x; *rS = %08x\n", (int)(ACC >> 32), (int)ACC, *rSh, *rS);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.0,21.1224:EVX:e500:evaddumiaaw %RS,%RA:Vector Add Unsigned Modulo Integer to Accumulator Word
+ unsigned64 t1, t2;
+ unsigned32 al, ah;
+ al = *rA;
+ ah = *rAh;
+ t1 = (unsigned64)EV_ACCHIGH + (unsigned64)ah;
+ t2 = EV_ACCLOW + al;
+ EV_SET_REG2_ACC(*rSh, *rS, t1, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.0,21.1219:EVX:e500:evsubfssiaaw %RS,%RA:Vector Subtract Signed Saturate Integer to Accumulator Word
+ signed64 t1, t2;
+ signed32 al, ah;
+ int ovl, ovh;
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH - (signed64)ah;
+ t2 = (signed64)EV_ACCLOW - (signed64)al;
+ ovh = EV_SAT_P_S32(t1);
+ ovl = EV_SAT_P_S32(t2);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE_ACC(ovh, t1, 0x80000000, 0x7fffffff, t1),
+ EV_SATURATE_ACC(ovl, t2, 0x80000000, 0x7fffffff, t2));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.0,21.1227:EVX:e500:evsubfsmiaaw %RS,%RA:Vector Subtract Signed Modulo Integer to Accumulator Word
+ signed64 t1, t2;
+ signed32 al, ah;
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH - (signed64)ah;
+ t2 = (signed64)EV_ACCLOW - (signed64)al;
+ EV_SET_REG2_ACC(*rSh, *rS, t1, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.1218:EVX:e500:evsubfusiaaw %RS,%RA:Vector Subtract Unsigned Saturate Integer to Accumulator Word
+ signed64 t1, t2;
+ unsigned32 al, ah;
+ int ovl, ovh;
+
+ al = *rA;
+ ah = *rAh;
+ t1 = (signed64)EV_ACCHIGH - (signed64)ah;
+ t2 = (signed64)EV_ACCLOW - (signed64)al;
+ ovh = EV_SAT_P_U32(t1);
+ ovl = EV_SAT_P_U32(t2);
+ EV_SET_REG2_ACC(*rSh, *rS, EV_SATURATE(ovh, 0, t1),
+ EV_SATURATE(ovl, 0, t2));
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.0,21.1226:EVX:e500:evsubfumiaaw %RS,%RA:Vector Subtract Unsigned Modulo Integer to Accumulator Word
+ unsigned64 t1, t2;
+ unsigned32 al, ah;
+ al = *rA;
+ ah = *rAh;
+ t1 = (unsigned64)EV_ACCHIGH - (unsigned64)ah;
+ t2 = (unsigned64)EV_ACCLOW - (unsigned64)al;
+ EV_SET_REG2_ACC(*rSh, *rS, t1, t2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.0,21.1220:EVX:e500:evmra %RS,%RA:Initialize Accumulator
+ EV_SET_REG2_ACC(*rSh, *rS, *rAh, *rA);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.1222:EVX:e500:evdivws %RS,%RA,%RB:Vector Divide Word Signed
+ signed32 dividendh, dividendl, divisorh, divisorl;
+ signed32 w1, w2;
+ int ovh, ovl;
+ dividendh = *rAh;
+ dividendl = *rA;
+ divisorh = *rBh;
+ divisorl = *rB;
+ if (dividendh < 0 && divisorh == 0) {
+ w1 = 0x80000000;
+ ovh = 1;
+ } else if (dividendh > 0 && divisorh == 0) {
+ w1 = 0x7fffffff;
+ ovh = 1;
+ } else if (dividendh == 0x80000000 && divisorh == -1) {
+ w1 = 0x7fffffff;
+ ovh = 1;
+ } else {
+ w1 = dividendh / divisorh;
+ ovh = 0;
+ }
+ if (dividendl < 0 && divisorl == 0) {
+ w2 = 0x80000000;
+ ovl = 1;
+ } else if (dividendl > 0 && divisorl == 0) {
+ w2 = 0x7fffffff;
+ ovl = 1;
+ } else if (dividendl == 0x80000000 && divisorl == -1) {
+ w2 = 0x7fffffff;
+ ovl = 1;
+ } else {
+ w2 = dividendl / divisorl;
+ ovl = 0;
+ }
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+
+0.4,6.RS,11.RA,16.RB,21.1223:EVX:e500:evdivwu %RS,%RA,%RB:Vector Divide Word Unsigned
+ unsigned32 dividendh, dividendl, divisorh, divisorl;
+ unsigned32 w1, w2;
+ int ovh, ovl;
+ dividendh = *rAh;
+ dividendl = *rA;
+ divisorh = *rBh;
+ divisorl = *rB;
+ if (divisorh == 0) {
+ w1 = 0xffffffff;
+ ovh = 1;
+ } else {
+ w1 = dividendh / divisorh;
+ ovh = 0;
+ }
+ if (divisorl == 0) {
+ w2 = 0xffffffff;
+ ovl = 1;
+ } else {
+ w2 = dividendl / divisorl;
+ ovl = 0;
+ }
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ EV_SET_SPEFSCR_OV(ovl, ovh);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK, spr_spefscr);
+
+
+#
+# A.2.9 Floating Point SPE Instructions
+#
+
+0.4,6.RS,11.RA,16.0,21.644:EVX:e500:evfsabs %RS,%RA:Vector Floating-Point Absolute Value
+ unsigned32 w1, w2;
+ w1 = *rAh & 0x7fffffff;
+ w2 = *rA & 0x7fffffff;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.645:EVX:e500:evfsnabs %RS,%RA:Vector Floating-Point Negative Absolute Value
+ unsigned32 w1, w2;
+ w1 = *rAh | 0x80000000;
+ w2 = *rA | 0x80000000;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.646:EVX:e500:evfsneg %RS,%RA:Vector Floating-Point Negate
+ unsigned32 w1, w2;
+ w1 = *rAh;
+ w2 = *rA;
+ w1 = (w1 & 0x7fffffff) | ((~w1) & 0x80000000);
+ w2 = (w2 & 0x7fffffff) | ((~w2) & 0x80000000);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.640:EVX:e500:evfsadd %RS,%RA,%RB:Vector Floating-Point Add
+ unsigned32 w1, w2;
+ w1 = ev_fs_add (*rAh, *rBh, spefscr_finvh, spefscr_fovfh, spefscr_funfh, spefscr_fgh, spefscr_fxh, processor);
+ w2 = ev_fs_add (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.641:EVX:e500:evfssub %RS,%RA,%RB:Vector Floating-Point Subtract
+ unsigned32 w1, w2;
+ w1 = ev_fs_sub (*rAh, *rBh, spefscr_finvh, spefscr_fovfh, spefscr_funfh, spefscr_fgh, spefscr_fxh, processor);
+ w2 = ev_fs_sub (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.648:EVX:e500:evfsmul %RS,%RA,%RB:Vector Floating-Point Multiply
+ unsigned32 w1, w2;
+ w1 = ev_fs_mul (*rAh, *rBh, spefscr_finvh, spefscr_fovfh, spefscr_funfh, spefscr_fgh, spefscr_fxh, processor);
+ w2 = ev_fs_mul (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.649:EVX:e500:evfsdiv %RS,%RA,%RB:Vector Floating-Point Divide
+ signed32 w1, w2;
+ w1 = ev_fs_div (*rAh, *rBh, spefscr_finvh, spefscr_fovfh, spefscr_funfh, spefscr_fdbzh, spefscr_fgh, spefscr_fxh, processor);
+ w2 = ev_fs_div (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fdbz, spefscr_fg, spefscr_fx, processor);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.652:EVX:e500:evfscmpgt %BF,%RA,%RB:Vector Floating-Point Compare Greater Than
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (EV_IS_INFDENORMNAN(&al) || EV_IS_INFDENORMNAN(&bl))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (EV_IS_INFDENORMNAN(&ah) || EV_IS_INFDENORMNAN(&bh))
+ EV_SET_SPEFSCR_BITS(spefscr_finvh);
+ if (sim_fpu_is_gt(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_gt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.653:EVX:e500:evfscmplt %BF,%RA,%RB:Vector Floating-Point Compare Less Than
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (EV_IS_INFDENORMNAN(&al) || EV_IS_INFDENORMNAN(&bl))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (EV_IS_INFDENORMNAN(&ah) || EV_IS_INFDENORMNAN(&bh))
+ EV_SET_SPEFSCR_BITS(spefscr_finvh);
+ if (sim_fpu_is_lt(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_lt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.654:EVX:e500:evfscmpeq %BF,%RA,%RB:Vector Floating-Point Compare Equal
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (EV_IS_INFDENORMNAN(&al) || EV_IS_INFDENORMNAN(&bl))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (EV_IS_INFDENORMNAN(&ah) || EV_IS_INFDENORMNAN(&bh))
+ EV_SET_SPEFSCR_BITS(spefscr_finvh);
+ if (sim_fpu_is_eq(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_eq(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.668:EVX:e500:evfststgt %BF,%RA,%RB:Vector Floating-Point Test Greater Than
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (sim_fpu_is_gt(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_gt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9./,11.RA,16.RB,21.669:EVX:e500:evfststlt %BF,%RA,%RB:Vector Floating-Point Test Less Than
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (sim_fpu_is_lt(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_lt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9./,11.RA,16.RB,21.670:EVX:e500:evfststeq %BF,%RA,%RB:Vector Floating-Point Test Equal
+ sim_fpu al, ah, bl, bh;
+ int w, ch, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&ah, *rAh);
+ sim_fpu_32to (&bl, *rB);
+ sim_fpu_32to (&bh, *rBh);
+ if (sim_fpu_is_eq(&ah, &bh))
+ ch = 1;
+ else
+ ch = 0;
+ if (sim_fpu_is_eq(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = ch << 3 | cl << 2 | (ch | cl) << 1 | (ch & cl);
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.RS,11.0,16.RB,21.656:EVX:e500:evfscfui %RS,%RB:Vector Convert Floating-Point from Unsigned Integer
+ unsigned32 f, w1, w2;
+ sim_fpu b;
+
+ sim_fpu_u32to (&b, *rBh, sim_fpu_round_default);
+ sim_fpu_to32 (&w1, &b);
+ sim_fpu_u32to (&b, *rB, sim_fpu_round_default);
+ sim_fpu_to32 (&w2, &b);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.664:EVX:e500:evfsctuiz %RS,%RB:Vector Convert Floating-Point to Unsigned Integer with Round toward Zero
+ unsigned32 w1, w2;
+ sim_fpu b;
+
+ sim_fpu_32to (&b, *rBh);
+ sim_fpu_to32u (&w1, &b, sim_fpu_round_zero);
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_zero);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.657:EVX:e500:evfscfsi %RS,%RB:Vector Convert Floating-Point from Signed Integer
+ signed32 w1, w2;
+ sim_fpu b, x, y;
+
+ sim_fpu_i32to (&b, *rBh, sim_fpu_round_default);
+ sim_fpu_to32 (&w1, &b);
+ sim_fpu_i32to (&b, *rB, sim_fpu_round_default);
+ sim_fpu_to32 (&w2, &b);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.658:EVX:e500:evfscfuf %RS,%RB:Vector Convert Floating-Point from Unsigned Fraction
+ unsigned32 w1, w2, bh, bl;
+ sim_fpu b, x, y;
+ bh = *rBh;
+ if (bh == 0xffffffff)
+ sim_fpu_to32 (&w1, &sim_fpu_one);
+ else {
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_u32to (&y, bh, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w1, &b);
+ }
+ bl = *rB;
+ if (bl == 0xffffffff)
+ sim_fpu_to32 (&w2, &sim_fpu_one);
+ else {
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_u32to (&y, bl, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w2, &b);
+ }
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.659:EVX:e500:evfscfsf %RS,%RB:Vector Convert Floating-Point from Signed Fraction
+ unsigned32 w1, w2;
+ sim_fpu b, x, y;
+
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_i32to (&y, *rBh, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w1, &b);
+
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_i32to (&y, *rB, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w2, &b);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.660:EVX:e500:evfsctui %RS,%RB:Vector Convert Floating-Point to Unsigned Integer
+ unsigned32 w1, w2;
+ sim_fpu b;
+
+ sim_fpu_32to (&b, *rBh);
+ sim_fpu_to32u (&w1, &b, sim_fpu_round_default);
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_default);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.661:EVX:e500:evfsctsi %RS,%RB:Vector Convert Floating-Point to Signed Integer
+ signed32 w1, w2;
+ sim_fpu b;
+
+ sim_fpu_32to (&b, *rBh);
+ sim_fpu_to32i (&w1, &b, sim_fpu_round_default);
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_default);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.666:EVX:e500:evfsctsiz %RS,%RB:Vector Convert Floating-Point to Signed Integer with Round toward Zero
+ signed32 w1, w2;
+ sim_fpu b;
+
+ sim_fpu_32to (&b, *rBh);
+ sim_fpu_to32i (&w1, &b, sim_fpu_round_zero);
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_zero);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.662:EVX:e500:evfsctuf %RS,%RB:Vector Convert Floating-Point to Unsigned Fraction
+ unsigned32 w1, w2;
+ sim_fpu b, x, y;
+
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_32to (&y, *rBh);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32u (&w1, &b, sim_fpu_round_default);
+
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_32to (&y, *rB);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_default);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.663:EVX:e500:evfsctsf %RS,%RB:Vector Convert Floating-Point to Signed Fraction
+ signed32 w1, w2;
+ sim_fpu b, x, y;
+
+ sim_fpu_32to (&y, *rBh);
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32i (&w1, &b, sim_fpu_round_near);
+
+ sim_fpu_32to (&y, *rB);
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_near);
+
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.0,21.708:EVX:e500:efsabs %RS,%RA:Floating-Point Absolute Value
+ unsigned32 w1, w2;
+ w1 = *rSh;
+ w2 = *rA & 0x7fffffff;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.709:EVX:e500:efsnabs %RS,%RA:Floating-Point Negative Absolute Value
+ unsigned32 w1, w2;
+ w1 = *rSh;
+ w2 = *rA | 0x80000000;
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.0,21.710:EVX:e500:efsneg %RS,%RA:Floating-Point Negate
+ unsigned32 w1, w2;
+ w1 = *rSh;
+ w2 = (*rA & 0x7fffffff) | ((~*rA) & 0x80000000);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.RB,21.704:EVX:e500:efsadd %RS,%RA,%RB:Floating-Point Add
+ unsigned32 w;
+ w = ev_fs_add (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG(*rS, w);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.705:EVX:e500:efssub %RS,%RA,%RB:Floating-Point Subtract
+ unsigned32 w;
+ w = ev_fs_sub (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG(*rS, w);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.712:EVX:e500:efsmul %RS,%RA,%RB:Floating-Point Multiply
+ unsigned32 w;
+ w = ev_fs_mul (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fgh, spefscr_fxh, processor);
+ EV_SET_REG(*rS, w);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.RS,11.RA,16.RB,21.713:EVX:e500:efsdiv %RS,%RA,%RB:Floating-Point Divide
+ unsigned32 w;
+ w = ev_fs_div (*rA, *rB, spefscr_finv, spefscr_fovf, spefscr_funf, spefscr_fdbz, spefscr_fg, spefscr_fx, processor);
+ EV_SET_REG(*rS, w);
+ PPC_INSN_INT_SPR(RS_BITMASK, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.716:EVX:e500:efscmpgt %BF,%RA,%RB:Floating-Point Compare Greater Than
+ sim_fpu a, b;
+ int w, cl;
+ sim_fpu_32to (&a, *rA);
+ sim_fpu_32to (&b, *rB);
+ if (EV_IS_INFDENORMNAN(&a) || EV_IS_INFDENORMNAN(&b))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (sim_fpu_is_gt(&a, &b))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.717:EVX:e500:efscmplt %BF,%RA,%RB:Floating-Point Compare Less Than
+ sim_fpu al, bl;
+ int w, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&bl, *rB);
+ if (EV_IS_INFDENORMNAN(&al) || EV_IS_INFDENORMNAN(&bl))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (sim_fpu_is_lt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.718:EVX:e500:efscmpeq %BF,%RA,%RB:Floating-Point Compare Equal
+ sim_fpu al, bl;
+ int w, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&bl, *rB);
+ if (EV_IS_INFDENORMNAN(&al) || EV_IS_INFDENORMNAN(&bl))
+ EV_SET_SPEFSCR_BITS(spefscr_finv);
+ if (sim_fpu_is_eq(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_SPR(0, RA_BITMASK | RB_BITMASK, spr_spefscr);
+
+0.4,6.BF,9./,11.RA,16.RB,21.732:EVX:e500:efststgt %BF,%RA,%RB:Floating-Point Test Greater Than
+ sim_fpu al, bl;
+ int w, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&bl, *rB);
+ if (sim_fpu_is_gt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9./,11.RA,16.RB,21.733:EVX:e500:efststlt %BF,%RA,%RB:Floating-Point Test Less Than
+ sim_fpu al, bl;
+ int w, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&bl, *rB);
+ if (sim_fpu_is_lt(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.BF,9./,11.RA,16.RB,21.734:EVX:e500:efststeq %BF,%RA,%RB:Floating-Point Test Equal
+ sim_fpu al, bl;
+ int w, cl;
+ sim_fpu_32to (&al, *rA);
+ sim_fpu_32to (&bl, *rB);
+ if (sim_fpu_is_eq(&al, &bl))
+ cl = 1;
+ else
+ cl = 0;
+ w = cl << 2 | cl << 1;
+ CR_SET(BF, w);
+ PPC_INSN_INT_CR(0, RA_BITMASK | RB_BITMASK, BF_BITMASK);
+
+0.4,6.RS,11.0,16.RB,21.721:EVX:e500:efscfsi %RS,%RB:Convert Floating-Point from Signed Integer
+ signed32 f, w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_i32to (&b, *rB, sim_fpu_round_default);
+ sim_fpu_to32 (&w2, &b);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.720:EVX:e500:efscfui %RS,%RB:Convert Floating-Point from Unsigned Integer
+ unsigned32 w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_u32to (&b, *rB, sim_fpu_round_default);
+ sim_fpu_to32 (&w2, &b);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.723:EVX:e500:efscfsf %RS,%RB:Convert Floating-Point from Signed Fraction
+ unsigned32 w1, w2;
+ sim_fpu b, x, y;
+ w1 = *rSh;
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_i32to (&y, *rB, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w2, &b);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.722:EVX:e500:efscfuf %RS,%RB:Convert Floating-Point from Unsigned Fraction
+ unsigned32 w1, w2, bl;
+ sim_fpu b, x, y;
+ w1 = *rSh;
+ bl = *rB;
+ if (bl == 0xffffffff)
+ sim_fpu_to32 (&w2, &sim_fpu_one);
+ else {
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_u32to (&y, bl, sim_fpu_round_default);
+ sim_fpu_div (&b, &y, &x);
+ sim_fpu_to32 (&w2, &b);
+ }
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.725:EVX:e500:efsctsi %RS,%RB:Convert Floating-Point to Signed Integer
+ signed64 temp;
+ signed32 w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_default);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.730:EVX:e500:efsctsiz %RS,%RB:Convert Floating-Point to Signed Integer with Round toward Zero
+ signed64 temp;
+ signed32 w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_zero);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.724:EVX:e500:efsctui %RS,%RB:Convert Floating-Point to Unsigned Integer
+ unsigned64 temp;
+ signed32 w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_default);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.728:EVX:e500:efsctuiz %RS,%RB:Convert Floating-Point to Unsigned Integer with Round toward Zero
+ unsigned64 temp;
+ signed32 w1, w2;
+ sim_fpu b;
+ w1 = *rSh;
+ sim_fpu_32to (&b, *rB);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_zero);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.727:EVX:e500:efsctsf %RS,%RB:Convert Floating-Point to Signed Fraction
+ unsigned32 w1, w2;
+ sim_fpu b, x, y;
+ w1 = *rSh;
+ sim_fpu_32to (&y, *rB);
+ sim_fpu_u32to (&x, 0x80000000, sim_fpu_round_default);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32i (&w2, &b, sim_fpu_round_default);
+ sim_fpu_to32 (&w2, &b);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+0.4,6.RS,11.0,16.RB,21.726:EVX:e500:efsctuf %RS,%RB:Convert Floating-Point to Unsigned Fraction
+ unsigned32 w1, w2;
+ sim_fpu b, x, y;
+ w1 = *rSh;
+ sim_fpu_u64to (&x, 0x100000000, sim_fpu_round_default);
+ sim_fpu_32to (&y, *rB);
+ sim_fpu_mul (&b, &y, &x);
+ sim_fpu_to32u (&w2, &b, sim_fpu_round_default);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, RB_BITMASK, 0);
+
+
+#
+# A.2.10 Vector Load/Store Instructions
+#
+
+0.4,6.RS,11.RA,16.UIMM,21.769:EVX:e500:evldd %RS,%RA,%UIMM:Vector Load Double Word into Double Word
+ unsigned64 m;
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ m = MEM(unsigned, EA, 8);
+ EV_SET_REG1(*rSh, *rS, m);
+ //printf("evldd(%d<-%d + %u): m %08x.%08x, *rSh %x *rS %x\n", RS, RA, UIMM, (int)(m >> 32), (int)m, *rSh, *rS);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.768:EVX:e500:evlddx %RS,%RA,%RB:Vector Load Double Word into Double Word Indexed
+ unsigned64 m;
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ m = MEM(unsigned, EA, 8);
+ EV_SET_REG1(*rSh, *rS, m);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.771:EVX:e500:evldw %RS,%RA,%UIMM:Vector Load Double into Two Words
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1, w2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ w1 = MEM(unsigned, EA, 4);
+ w2 = MEM(unsigned, EA + 4, 4);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.770:EVX:e500:evldwx %RS,%RA,%RB:Vector Load Double into Two Words Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1, w2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ w1 = MEM(unsigned, EA, 4);
+ w2 = MEM(unsigned, EA + 4, 4);
+ EV_SET_REG2(*rSh, *rS, w1, w2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.773:EVX:e500:evldh %RS,%RA,%UIMM:Vector Load Double into 4 Half Words
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ h1 = MEM(unsigned, EA, 2);
+ h2 = MEM(unsigned, EA + 2, 2);
+ h3 = MEM(unsigned, EA + 4, 2);
+ h4 = MEM(unsigned, EA + 6, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.772:EVX:e500:evldhx %RS,%RA,%RB:Vector Load Double into 4 Half Words Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = MEM(unsigned, EA, 2);
+ h2 = MEM(unsigned, EA + 2, 2);
+ h3 = MEM(unsigned, EA + 4, 2);
+ h4 = MEM(unsigned, EA + 6, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.785:EVX:e500:evlwhe %RS,%RA,%UIMM:Vector Load Word into Two Half Words Even
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 2);
+ h1 = MEM(unsigned, EA, 2);
+ h2 = 0;
+ h3 = MEM(unsigned, EA + 2, 2);
+ h4 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.784:EVX:e500:evlwhex %RS,%RA,%RB:Vector Load Word into Two Half Words Even Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = MEM(unsigned, EA, 2);
+ h2 = 0;
+ h3 = MEM(unsigned, EA + 2, 2);
+ h4 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.789:EVX:e500:evlwhou %RS,%RA,%UIMM:Vector Load Word into Two Half Words Odd Unsigned zero-extended
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 2);
+ h1 = 0;
+ h2 = MEM(unsigned, EA, 2);
+ h3 = 0;
+ h4 = MEM(unsigned, EA + 2, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.788:EVX:e500:evlwhoux %RS,%RA,%RB:Vector Load Word into Two Half Words Odd Unsigned Indexed zero-extended
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = 0;
+ h2 = MEM(unsigned, EA, 2);
+ h3 = 0;
+ h4 = MEM(unsigned, EA + 2, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.791:EVX:e500:evlwhos %RS,%RA,%UIMM:Vector Load Word into Half Words Odd Signed with sign extension
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 2);
+ h2 = MEM(unsigned, EA, 2);
+ if (h2 & 0x8000)
+ h1 = 0xffff;
+ else
+ h1 = 0;
+ h4 = MEM(unsigned, EA + 2, 2);
+ if (h4 & 0x8000)
+ h3 = 0xffff;
+ else
+ h3 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.790:EVX:e500:evlwhosx %RS,%RA,%RB:Vector Load Word into Half Words Odd Signed Indexed with sign extension
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h2 = MEM(unsigned, EA, 2);
+ if (h2 & 0x8000)
+ h1 = 0xffff;
+ else
+ h1 = 0;
+ h4 = MEM(unsigned, EA + 2, 2);
+ if (h4 & 0x8000)
+ h3 = 0xffff;
+ else
+ h3 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h3, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.793:EVX:e500:evlwwsplat %RS,%RA,%UIMM:Vector Load Word into Word and Splat
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 2);
+ w1 = MEM(unsigned, EA, 4);
+ EV_SET_REG2(*rSh, *rS, w1, w1);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.792:EVX:e500:evlwwsplatx %RS,%RA,%RB:Vector Load Word into Word and Splat Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ w1 = MEM(unsigned, EA, 4);
+ EV_SET_REG2(*rSh, *rS, w1, w1);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.797:EVX:e500:evlwhsplat %RS,%RA,%UIMM:Vector Load Word into 2 Half Words and Splat
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 2);
+ h1 = MEM(unsigned, EA, 2);
+ h2 = MEM(unsigned, EA + 2, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h1, h2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.796:EVX:e500:evlwhsplatx %RS,%RA,%RB:Vector Load Word into 2 Half Words and Splat Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = MEM(unsigned, EA, 2);
+ h2 = MEM(unsigned, EA + 2, 2);
+ EV_SET_REG4(*rSh, *rS, h1, h1, h2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.777:EVX:e500:evlhhesplat %RS,%RA,%UIMM:Vector Load Half Word into Half Words Even and Splat
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 1);
+ h = MEM(unsigned, EA, 2);
+ EV_SET_REG4(*rSh, *rS, h, 0, h, 0);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.776:EVX:e500:evlhhesplatx %RS,%RA,%RB:Vector Load Half Word into Half Words Even and Splat Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h = MEM(unsigned, EA, 2);
+ EV_SET_REG4(*rSh, *rS, h, 0, h, 0);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.781:EVX:e500:evlhhousplat %RS,%RA,%UIMM:Vector Load Half Word into Half Word Odd Unsigned and Splat
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 1);
+ h = MEM(unsigned, EA, 2);
+ EV_SET_REG4(*rSh, *rS, 0, h, 0, h);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.780:EVX:e500:evlhhousplatx %RS,%RA,%RB:Vector Load Half Word into Half Word Odd Unsigned and Splat Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h = MEM(unsigned, EA, 2);
+ EV_SET_REG4(*rSh, *rS, 0, h, 0, h);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.783:EVX:e500:evlhhossplat %RS,%RA,%UIMM:Vector Load Half Word into Half Word Odd Signed and Splat
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 1);
+ h2 = MEM(unsigned, EA, 2);
+ if (h2 & 0x8000)
+ h1 = 0xffff;
+ else
+ h1 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h1, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.782:EVX:e500:evlhhossplatx %RS,%RA,%RB:Vector Load Half Word into Half Word Odd Signed and Splat Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h2 = MEM(unsigned, EA, 2);
+ if (h2 & 0x8000)
+ h1 = 0xffff;
+ else
+ h1 = 0;
+ EV_SET_REG4(*rSh, *rS, h1, h2, h1, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+
+0.4,6.RS,11.RA,16.UIMM,21.801:EVX:e500:evstdd %RS,%RA,%UIMM:Vector Store Double of Double
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ STORE(EA, 4, (*rSh));
+ STORE(EA + 4, 4, (*rS));
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.800:EVX:e500:evstddx %RS,%RA,%RB:Vector Store Double of Double Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ STORE(EA, 4, (*rSh));
+ STORE(EA + 4, 4, (*rS));
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.803:EVX:e500:evstdw %RS,%RA,%UIMM:Vector Store Double of Two Words
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1, w2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ w1 = *rSh;
+ w2 = *rS;
+ STORE(EA + 0, 4, w1);
+ STORE(EA + 4, 4, w2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.802:EVX:e500:evstdwx %RS,%RA,%RB:Vector Store Double of Two Words Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w1, w2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ w1 = *rSh;
+ w2 = *rS;
+ STORE(EA + 0, 4, w1);
+ STORE(EA + 4, 4, w2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.805:EVX:e500:evstdh %RS,%RA,%UIMM:Vector Store Double of Four Half Words
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ h1 = EV_HIHALF(*rSh);
+ h2 = EV_LOHALF(*rSh);
+ h3 = EV_HIHALF(*rS);
+ h4 = EV_LOHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ STORE(EA + 4, 2, h3);
+ STORE(EA + 6, 2, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.804:EVX:e500:evstdhx %RS,%RA,%RB:Vector Store Double of Four Half Words Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2, h3, h4;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = EV_HIHALF(*rSh);
+ h2 = EV_LOHALF(*rSh);
+ h3 = EV_HIHALF(*rS);
+ h4 = EV_LOHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ STORE(EA + 4, 2, h3);
+ STORE(EA + 6, 2, h4);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.825:EVX:e500:evstwwe %RS,%RA,%UIMM:Vector Store Word of Word from Even
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ w = *rSh;
+ STORE(EA, 4, w);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.824:EVX:e500:evstwwex %RS,%RA,%RB:Vector Store Word of Word from Even Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ w = *rSh;
+ STORE(EA, 4, w);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.829:EVX:e500:evstwwo %RS,%RA,%UIMM:Vector Store Word of Word from Odd
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ w = *rS;
+ STORE(EA, 4, w);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.828:EVX:e500:evstwwox %RS,%RA,%RB:Vector Store Word of Word from Odd Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned32 w;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ w = *rS;
+ STORE(EA, 4, w);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.817:EVX:e500:evstwhe %RS,%RA,%UIMM:Vector Store Word of Two Half Words from Even
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ h1 = EV_HIHALF(*rSh);
+ h2 = EV_HIHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.816:EVX:e500:evstwhex %RS,%RA,%RB:Vector Store Word of Two Half Words from Even Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = EV_HIHALF(*rSh);
+ h2 = EV_HIHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+0.4,6.RS,11.RA,16.UIMM,21.821:EVX:e500:evstwho %RS,%RA,%UIMM:Vector Store Word of Two Half Words from Odd
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + (UIMM << 3);
+ h1 = EV_LOHALF(*rSh);
+ h2 = EV_LOHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1), 0);
+
+0.4,6.RS,11.RA,16.RB,21.820:EVX:e500:evstwhox %RS,%RA,%RB:Vector Store Word of Two Half Words from Odd Indexed
+ unsigned_word b;
+ unsigned_word EA;
+ unsigned16 h1, h2;
+ if (RA_is_0) b = 0;
+ else b = *rA;
+ EA = b + *rB;
+ h1 = EV_LOHALF(*rSh);
+ h2 = EV_LOHALF(*rS);
+ STORE(EA + 0, 2, h1);
+ STORE(EA + 2, 2, h2);
+ PPC_INSN_INT(RS_BITMASK, (RA_BITMASK & ~1) | RB_BITMASK, 0);
+
+
+#
+# 4.5.1 Integer Select Instruction
+#
+
+0.31,6.RS,11.RA,16.RB,21.CRB,26.30:X:e500:isel %RS,%RA,%RB,%CRB:Integer Select
+ if (CR & (1 << (31 - (unsigned)CRB)))
+ if (RA_is_0)
+ EV_SET_REG1(*rSh, *rS, 0);
+ else
+ EV_SET_REG2(*rSh, *rS, *rAh, *rA);
+ else
+ EV_SET_REG2(*rSh, *rS, *rBh, *rB);
+ PPC_INSN_INT(RS_BITMASK, RA_BITMASK | RB_BITMASK, 0);
projects / binutils-gdb.git / commitdiff