if (shader->bo == NULL) {
shader->bo = (struct r600_resource*)
- pipe_buffer_create(ctx->screen, PIPE_BIND_CUSTOM, PIPE_USAGE_IMMUTABLE, shader->shader.bc.ndw * 4);
+ pipe_buffer_create(ctx->screen, 0, PIPE_USAGE_IMMUTABLE, shader->shader.bc.ndw * 4);
if (shader->bo == NULL) {
return -ENOMEM;
}
/* handle some special cases */
if (i == 1 || i == 3) {
switch (ctx->parse.FullToken.FullInstruction.Instruction.Opcode) {
- case TGSI_OPCODE_SUB:
- r600_bytecode_src_toggle_neg(&alu.src[1]);
- break;
case TGSI_OPCODE_DABS:
r600_bytecode_src_set_abs(&alu.src[0]);
break;
r600_bytecode_src(&alu.src[0], &ctx->src[1], i);
r600_bytecode_src(&alu.src[1], &ctx->src[0], i);
}
- /* handle some special cases */
- switch (inst->Instruction.Opcode) {
- case TGSI_OPCODE_SUB:
- r600_bytecode_src_toggle_neg(&alu.src[1]);
- break;
- case TGSI_OPCODE_ABS:
- r600_bytecode_src_set_abs(&alu.src[0]);
- break;
- default:
- break;
- }
if (i == lasti || trans_only) {
alu.last = 1;
}
int lasti = tgsi_last_instruction(inst->Dst[0].Register.WriteMask);
int t1 = ctx->temp_reg;
- for (k = 0; k < 2; k++) {
- if (!(inst->Dst[0].Register.WriteMask & (0x3 << (k * 2))))
- continue;
+ /* t1 would get overwritten below if we actually tried to
+ * multiply two pairs of doubles at a time. */
+ assert(inst->Dst[0].Register.WriteMask == TGSI_WRITEMASK_XY ||
+ inst->Dst[0].Register.WriteMask == TGSI_WRITEMASK_ZW);
- for (i = 0; i < 4; i++) {
- memset(&alu, 0, sizeof(struct r600_bytecode_alu));
- alu.op = ctx->inst_info->op;
- for (j = 0; j < inst->Instruction.NumSrcRegs; j++) {
- r600_bytecode_src(&alu.src[j], &ctx->src[j], k * 2 + ((i == 3) ? 0 : 1));
- }
- alu.dst.sel = t1;
- alu.dst.chan = i;
- alu.dst.write = 1;
- if (i == 3)
- alu.last = 1;
- r = r600_bytecode_add_alu(ctx->bc, &alu);
- if (r)
- return r;
+ k = inst->Dst[0].Register.WriteMask == TGSI_WRITEMASK_XY ? 0 : 1;
+
+ for (i = 0; i < 4; i++) {
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ alu.op = ctx->inst_info->op;
+ for (j = 0; j < inst->Instruction.NumSrcRegs; j++) {
+ r600_bytecode_src(&alu.src[j], &ctx->src[j], k * 2 + ((i == 3) ? 0 : 1));
}
+ alu.dst.sel = t1;
+ alu.dst.chan = i;
+ alu.dst.write = 1;
+ if (i == 3)
+ alu.last = 1;
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
}
for (i = 0; i <= lasti; i++) {
return 0;
}
+static int tgsi_pk2h(struct r600_shader_ctx *ctx)
+{
+ struct tgsi_full_instruction *inst = &ctx->parse.FullToken.FullInstruction;
+ struct r600_bytecode_alu alu;
+ int r, i;
+ int lasti = tgsi_last_instruction(inst->Dst[0].Register.WriteMask);
+
+ /* temp.xy = f32_to_f16(src) */
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ alu.op = ALU_OP1_FLT32_TO_FLT16;
+ alu.dst.chan = 0;
+ alu.dst.sel = ctx->temp_reg;
+ alu.dst.write = 1;
+ r600_bytecode_src(&alu.src[0], &ctx->src[0], 0);
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+ alu.dst.chan = 1;
+ r600_bytecode_src(&alu.src[0], &ctx->src[0], 1);
+ alu.last = 1;
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+
+ /* dst.x = temp.y * 0x10000 + temp.x */
+ for (i = 0; i < lasti + 1; i++) {
+ if (!(inst->Dst[0].Register.WriteMask & (1 << i)))
+ continue;
+
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ alu.op = ALU_OP3_MULADD_UINT24;
+ alu.is_op3 = 1;
+ tgsi_dst(ctx, &inst->Dst[0], i, &alu.dst);
+ alu.last = i == lasti;
+ alu.src[0].sel = ctx->temp_reg;
+ alu.src[0].chan = 1;
+ alu.src[1].sel = V_SQ_ALU_SRC_LITERAL;
+ alu.src[1].value = 0x10000;
+ alu.src[2].sel = ctx->temp_reg;
+ alu.src[2].chan = 0;
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+ }
+
+ return 0;
+}
+
+static int tgsi_up2h(struct r600_shader_ctx *ctx)
+{
+ struct tgsi_full_instruction *inst = &ctx->parse.FullToken.FullInstruction;
+ struct r600_bytecode_alu alu;
+ int r, i;
+ int lasti = tgsi_last_instruction(inst->Dst[0].Register.WriteMask);
+
+ /* temp.x = src.x */
+ /* note: no need to mask out the high bits */
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ alu.op = ALU_OP1_MOV;
+ alu.dst.chan = 0;
+ alu.dst.sel = ctx->temp_reg;
+ alu.dst.write = 1;
+ r600_bytecode_src(&alu.src[0], &ctx->src[0], 0);
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+
+ /* temp.y = src.x >> 16 */
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ alu.op = ALU_OP2_LSHR_INT;
+ alu.dst.chan = 1;
+ alu.dst.sel = ctx->temp_reg;
+ alu.dst.write = 1;
+ r600_bytecode_src(&alu.src[0], &ctx->src[0], 0);
+ alu.src[1].sel = V_SQ_ALU_SRC_LITERAL;
+ alu.src[1].value = 16;
+ alu.last = 1;
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+
+ /* dst.wz = dst.xy = f16_to_f32(temp.xy) */
+ for (i = 0; i < lasti + 1; i++) {
+ if (!(inst->Dst[0].Register.WriteMask & (1 << i)))
+ continue;
+ memset(&alu, 0, sizeof(struct r600_bytecode_alu));
+ tgsi_dst(ctx, &inst->Dst[0], i, &alu.dst);
+ alu.op = ALU_OP1_FLT16_TO_FLT32;
+ alu.src[0].sel = ctx->temp_reg;
+ alu.src[0].chan = i % 2;
+ alu.last = i == lasti;
+ r = r600_bytecode_add_alu(ctx->bc, &alu);
+ if (r)
+ return r;
+ }
+
+ return 0;
+}
+
static const struct r600_shader_tgsi_instruction r600_shader_tgsi_instruction[] = {
[TGSI_OPCODE_ARL] = { ALU_OP0_NOP, tgsi_r600_arl},
[TGSI_OPCODE_MOV] = { ALU_OP1_MOV, tgsi_op2},
[TGSI_OPCODE_SLT] = { ALU_OP2_SETGT, tgsi_op2_swap},
[TGSI_OPCODE_SGE] = { ALU_OP2_SETGE, tgsi_op2},
[TGSI_OPCODE_MAD] = { ALU_OP3_MULADD, tgsi_op3},
- [TGSI_OPCODE_SUB] = { ALU_OP2_ADD, tgsi_op2},
[TGSI_OPCODE_LRP] = { ALU_OP0_NOP, tgsi_lrp},
[TGSI_OPCODE_FMA] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_SQRT] = { ALU_OP1_SQRT_IEEE, tgsi_trans_srcx_replicate},
[TGSI_OPCODE_POW] = { ALU_OP0_NOP, tgsi_pow},
[TGSI_OPCODE_XPD] = { ALU_OP0_NOP, tgsi_xpd},
[32] = { ALU_OP0_NOP, tgsi_unsupported},
- [TGSI_OPCODE_ABS] = { ALU_OP1_MOV, tgsi_op2},
+ [33] = { ALU_OP0_NOP, tgsi_unsupported},
[34] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_DPH] = { ALU_OP2_DOT4, tgsi_dp},
[TGSI_OPCODE_COS] = { ALU_OP1_COS, tgsi_trig},
[TGSI_OPCODE_SLT] = { ALU_OP2_SETGT, tgsi_op2_swap},
[TGSI_OPCODE_SGE] = { ALU_OP2_SETGE, tgsi_op2},
[TGSI_OPCODE_MAD] = { ALU_OP3_MULADD, tgsi_op3},
- [TGSI_OPCODE_SUB] = { ALU_OP2_ADD, tgsi_op2},
[TGSI_OPCODE_LRP] = { ALU_OP0_NOP, tgsi_lrp},
[TGSI_OPCODE_FMA] = { ALU_OP3_FMA, tgsi_op3},
[TGSI_OPCODE_SQRT] = { ALU_OP1_SQRT_IEEE, tgsi_trans_srcx_replicate},
[TGSI_OPCODE_POW] = { ALU_OP0_NOP, tgsi_pow},
[TGSI_OPCODE_XPD] = { ALU_OP0_NOP, tgsi_xpd},
[32] = { ALU_OP0_NOP, tgsi_unsupported},
- [TGSI_OPCODE_ABS] = { ALU_OP1_MOV, tgsi_op2},
+ [33] = { ALU_OP0_NOP, tgsi_unsupported},
[34] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_DPH] = { ALU_OP2_DOT4, tgsi_dp},
[TGSI_OPCODE_COS] = { ALU_OP1_COS, tgsi_trig},
[TGSI_OPCODE_DDX] = { FETCH_OP_GET_GRADIENTS_H, tgsi_tex},
[TGSI_OPCODE_DDY] = { FETCH_OP_GET_GRADIENTS_V, tgsi_tex},
[TGSI_OPCODE_KILL] = { ALU_OP2_KILLGT, tgsi_kill}, /* unconditional kill */
- [TGSI_OPCODE_PK2H] = { ALU_OP0_NOP, tgsi_unsupported},
+ [TGSI_OPCODE_PK2H] = { ALU_OP0_NOP, tgsi_pk2h},
[TGSI_OPCODE_PK2US] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_PK4B] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_PK4UB] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_TEX] = { FETCH_OP_SAMPLE, tgsi_tex},
[TGSI_OPCODE_TXD] = { FETCH_OP_SAMPLE_G, tgsi_tex},
[TGSI_OPCODE_TXP] = { FETCH_OP_SAMPLE, tgsi_tex},
- [TGSI_OPCODE_UP2H] = { ALU_OP0_NOP, tgsi_unsupported},
+ [TGSI_OPCODE_UP2H] = { ALU_OP0_NOP, tgsi_up2h},
[TGSI_OPCODE_UP2US] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_UP4B] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_UP4UB] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_SLT] = { ALU_OP2_SETGT, tgsi_op2_swap},
[TGSI_OPCODE_SGE] = { ALU_OP2_SETGE, tgsi_op2},
[TGSI_OPCODE_MAD] = { ALU_OP3_MULADD, tgsi_op3},
- [TGSI_OPCODE_SUB] = { ALU_OP2_ADD, tgsi_op2},
[TGSI_OPCODE_LRP] = { ALU_OP0_NOP, tgsi_lrp},
[TGSI_OPCODE_FMA] = { ALU_OP3_FMA, tgsi_op3},
[TGSI_OPCODE_SQRT] = { ALU_OP1_SQRT_IEEE, cayman_emit_float_instr},
[TGSI_OPCODE_POW] = { ALU_OP0_NOP, cayman_pow},
[TGSI_OPCODE_XPD] = { ALU_OP0_NOP, tgsi_xpd},
[32] = { ALU_OP0_NOP, tgsi_unsupported},
- [TGSI_OPCODE_ABS] = { ALU_OP1_MOV, tgsi_op2},
+ [33] = { ALU_OP0_NOP, tgsi_unsupported},
[34] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_DPH] = { ALU_OP2_DOT4, tgsi_dp},
[TGSI_OPCODE_COS] = { ALU_OP1_COS, cayman_trig},
[TGSI_OPCODE_DDX] = { FETCH_OP_GET_GRADIENTS_H, tgsi_tex},
[TGSI_OPCODE_DDY] = { FETCH_OP_GET_GRADIENTS_V, tgsi_tex},
[TGSI_OPCODE_KILL] = { ALU_OP2_KILLGT, tgsi_kill}, /* unconditional kill */
- [TGSI_OPCODE_PK2H] = { ALU_OP0_NOP, tgsi_unsupported},
+ [TGSI_OPCODE_PK2H] = { ALU_OP0_NOP, tgsi_pk2h},
[TGSI_OPCODE_PK2US] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_PK4B] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_PK4UB] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_TEX] = { FETCH_OP_SAMPLE, tgsi_tex},
[TGSI_OPCODE_TXD] = { FETCH_OP_SAMPLE_G, tgsi_tex},
[TGSI_OPCODE_TXP] = { FETCH_OP_SAMPLE, tgsi_tex},
- [TGSI_OPCODE_UP2H] = { ALU_OP0_NOP, tgsi_unsupported},
+ [TGSI_OPCODE_UP2H] = { ALU_OP0_NOP, tgsi_up2h},
[TGSI_OPCODE_UP2US] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_UP4B] = { ALU_OP0_NOP, tgsi_unsupported},
[TGSI_OPCODE_UP4UB] = { ALU_OP0_NOP, tgsi_unsupported},