From: Fariborz Jahanian Date: Mon, 16 May 2005 15:24:09 +0000 (+0000) Subject: Fix vec_merge patterns for Altivec ppc. X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=fe2d45c77f86ebbed8a1ef8c6c59e61daace584f;p=gcc.git Fix vec_merge patterns for Altivec ppc. OKed by Geoff Keating. From-SVN: r99779 --- diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 5739c60f2b7..e8d0a176e3d 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,10 @@ +2005-05-16 Fariborz Jahanian + + * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh, + altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): + New values for vec_select definitions and bitmask for element + selection. + 2005-05-16 Kazu Hirata * dwarf2out.c (used_rtx_varray): Rename to used_rtx_array. diff --git a/gcc/config/rs6000/altivec.md b/gcc/config/rs6000/altivec.md index 4b4f9502ec3..418228d7bb3 100644 --- a/gcc/config/rs6000/altivec.md +++ b/gcc/config/rs6000/altivec.md @@ -673,24 +673,40 @@ (define_insn "altivec_vmrghb" [(set (match_operand:V16QI 0 "register_operand" "=v") (vec_merge:V16QI (vec_select:V16QI (match_operand:V16QI 1 "register_operand" "v") - (parallel [(const_int 8) + (parallel [(const_int 0) + (const_int 8) + (const_int 1) (const_int 9) + (const_int 2) (const_int 10) - (const_int 11) + (const_int 3) + (const_int 11) + (const_int 4) (const_int 12) + (const_int 5) (const_int 13) - (const_int 14) - (const_int 15) + (const_int 6) + (const_int 14) + (const_int 7) + (const_int 15)])) + (vec_select:V16QI (match_operand:V16QI 2 "register_operand" "v") + (parallel [(const_int 8) (const_int 0) + (const_int 9) (const_int 1) + (const_int 10) (const_int 2) - (const_int 3) + (const_int 11) + (const_int 3) + (const_int 12) (const_int 4) + (const_int 13) (const_int 5) + (const_int 14) (const_int 6) + (const_int 15) (const_int 7)])) - (match_operand:V16QI 2 "register_operand" "v") - (const_int 255)))] + (const_int 21845)))] "TARGET_ALTIVEC" "vmrghb %0,%1,%2" [(set_attr "type" "vecperm")]) @@ -698,16 +714,24 @@ (define_insn "altivec_vmrghh" [(set (match_operand:V8HI 0 "register_operand" "=v") (vec_merge:V8HI (vec_select:V8HI (match_operand:V8HI 1 "register_operand" "v") - (parallel [(const_int 4) + (parallel [(const_int 0) + (const_int 4) + (const_int 1) (const_int 5) + (const_int 2) (const_int 6) - (const_int 7) + (const_int 3) + (const_int 7)])) + (vec_select:V8HI (match_operand:V8HI 2 "register_operand" "v") + (parallel [(const_int 4) (const_int 0) + (const_int 5) (const_int 1) + (const_int 6) (const_int 2) + (const_int 7) (const_int 3)])) - (match_operand:V8HI 2 "register_operand" "v") - (const_int 15)))] + (const_int 85)))] "TARGET_ALTIVEC" "vmrghh %0,%1,%2" [(set_attr "type" "vecperm")]) @@ -715,67 +739,99 @@ (define_insn "altivec_vmrghw" [(set (match_operand:V4SI 0 "register_operand" "=v") (vec_merge:V4SI (vec_select:V4SI (match_operand:V4SI 1 "register_operand" "v") + (parallel [(const_int 0) + (const_int 2) + (const_int 1) + (const_int 3)])) + (vec_select:V4SI (match_operand:V4SI 2 "register_operand" "v") (parallel [(const_int 2) - (const_int 3) - (const_int 0) + (const_int 0) + (const_int 3) (const_int 1)])) - (match_operand:V4SI 2 "register_operand" "v") - (const_int 12)))] + (const_int 5)))] "TARGET_ALTIVEC" "vmrghw %0,%1,%2" [(set_attr "type" "vecperm")]) (define_insn "altivec_vmrglb" [(set (match_operand:V16QI 0 "register_operand" "=v") - (vec_merge:V16QI (vec_select:V16QI (match_operand:V16QI 2 "register_operand" "v") - (parallel [(const_int 0) + (vec_merge:V16QI (vec_select:V16QI (match_operand:V16QI 1 "register_operand" "v") + (parallel [(const_int 8) + (const_int 0) + (const_int 9) (const_int 1) + (const_int 10) (const_int 2) - (const_int 3) + (const_int 11) + (const_int 3) + (const_int 12) (const_int 4) + (const_int 13) (const_int 5) - (const_int 6) - (const_int 7) + (const_int 14) + (const_int 6) + (const_int 15) + (const_int 7)])) + (vec_select:V16QI (match_operand:V16QI 2 "register_operand" "v") + (parallel [(const_int 0) (const_int 8) + (const_int 1) (const_int 9) + (const_int 2) (const_int 10) - (const_int 11) + (const_int 3) + (const_int 11) + (const_int 4) (const_int 12) + (const_int 5) (const_int 13) + (const_int 6) (const_int 14) + (const_int 7) (const_int 15)])) - (match_operand:V16QI 1 "register_operand" "v") - (const_int 255)))] + (const_int 21845)))] "TARGET_ALTIVEC" "vmrglb %0,%1,%2" [(set_attr "type" "vecperm")]) (define_insn "altivec_vmrglh" [(set (match_operand:V8HI 0 "register_operand" "=v") - (vec_merge:V8HI (vec_select:V8HI (match_operand:V8HI 2 "register_operand" "v") - (parallel [(const_int 0) + (vec_merge:V8HI (vec_select:V8HI (match_operand:V8HI 1 "register_operand" "v") + (parallel [(const_int 4) + (const_int 0) + (const_int 5) (const_int 1) + (const_int 6) (const_int 2) - (const_int 3) + (const_int 7) + (const_int 3)])) + (vec_select:V8HI (match_operand:V8HI 2 "register_operand" "v") + (parallel [(const_int 0) (const_int 4) + (const_int 1) (const_int 5) + (const_int 2) (const_int 6) + (const_int 3) (const_int 7)])) - (match_operand:V8HI 1 "register_operand" "v") - (const_int 15)))] + (const_int 85)))] "TARGET_ALTIVEC" "vmrglh %0,%1,%2" [(set_attr "type" "vecperm")]) (define_insn "altivec_vmrglw" [(set (match_operand:V4SI 0 "register_operand" "=v") - (vec_merge:V4SI (vec_select:V4SI (match_operand:V4SI 2 "register_operand" "v") + (vec_merge:V4SI (vec_select:V4SI (match_operand:V4SI 1 "register_operand" "v") + (parallel [(const_int 2) + (const_int 0) + (const_int 3) + (const_int 1)])) + (vec_select:V4SI (match_operand:V4SI 2 "register_operand" "v") (parallel [(const_int 0) - (const_int 1) - (const_int 2) + (const_int 2) + (const_int 1) (const_int 3)])) - (match_operand:V4SI 1 "register_operand" "v") - (const_int 12)))] + (const_int 5)))] "TARGET_ALTIVEC" "vmrglw %0,%1,%2" [(set_attr "type" "vecperm")]) diff --git a/gcc/testsuite/gcc.dg/ppc-vec-merge.c b/gcc/testsuite/gcc.dg/ppc-vec-merge.c new file mode 100644 index 00000000000..9eef9847a82 --- /dev/null +++ b/gcc/testsuite/gcc.dg/ppc-vec-merge.c @@ -0,0 +1,600 @@ +/* { dg-do run { target powerpc*-*-* } } */ +/* { dg-options "-faltivec -O2" } */ + +int printf(const char * , ...); +extern void abort(); + +void foo(char *bS, char *bS_edge, int field_MBAFF, int top){ + char intra[16] __attribute__ ((aligned(16))); + signed short mv_const[8] __attribute__((aligned(16))); + + vector signed short v_three, v_ref_mask00, v_ref_mask01, v_vec_maskv, v_vec_maskh; + vector unsigned char v_permv, v_permh, v_bS, v_bSh, v_bSv, v_cbp_maskv, v_cbp_maskvn, v_cbp_maskh, v_cbp_maskhn, v_intra_maskh, v_intra_maskv, v_intra_maskhn, v_intra_maskvn; + vector unsigned char tmp7, tmp8, tmp9, tmp10, v_c1, v_cbp1, v_cbp2, v_pocl, v_poch; + vector signed short v0, v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6; + vector signed short idx0; + vector signed short tmp00, tmp01, tmp02, tmp03; + vector unsigned char v_zero = (vector unsigned char) {'a','b','c','d','e','f','g','h','i','j','k','l','m','n','o','p'}; + v_three = (vector signed short) vec_ld (0, (vector signed short *) mv_const); + + vector unsigned char v_coef_mask = vec_ld(0, (vector unsigned char *)mv_const); + vector unsigned char v_coef_mask_hi = vec_splat(v_coef_mask, 0); + vector unsigned char v_coef_mask_lo = vec_splat(v_coef_mask, 1); + v_coef_mask = vec_sld(v_coef_mask_hi, v_coef_mask_lo, 8); + vector unsigned char v_bit_mask = vec_sub(vec_splat_u8(7), vec_lvsl(0, (unsigned char *)0)); + v_bit_mask = vec_sld(vec_sld(v_bit_mask, v_bit_mask, 8), v_bit_mask, 8); + v_bit_mask = vec_sl(vec_splat_u8(1), v_bit_mask); + tmp5 = (vector signed short) vec_and(v_coef_mask, v_bit_mask); + + intra[0] = 1; + tmp8 = vec_ld (0, (vector unsigned char *) intra); + tmp9 = vec_ld (0, (vector unsigned char *) mv_const); + tmp10 = vec_ld (0, (vector unsigned char *) mv_const); + v_permv = vec_ld (0, (vector unsigned char *) mv_const); + v_permh = vec_ld (0, (vector unsigned char *) mv_const); + tmp6 = vec_ld (0, (vector signed short *) mv_const); + + tmp8 = vec_splat((vector unsigned char) tmp8, 0); + tmp9 = vec_splat((vector unsigned char) tmp9, 12); + tmp10 = vec_splat((vector unsigned char) tmp10, 12); + tmp9 = vec_sld ((vector unsigned char) tmp9,(vector unsigned char) tmp8, 12); + tmp10 = vec_sld ((vector unsigned char) tmp10, (vector unsigned char) tmp8, 12); + v_intra_maskv = vec_or (tmp9, tmp8); + v_intra_maskh = vec_or (tmp10, tmp8); + v_intra_maskv = (vector unsigned char) vec_cmpgt ((vector unsigned char) v_intra_maskv, (vector unsigned char) v_zero); + v_intra_maskh = (vector unsigned char) vec_cmpgt ((vector unsigned char) v_intra_maskh, (vector unsigned char) v_zero); + + tmp9 = vec_lvsl (4 + (top<<2), (unsigned char *) 0x0); + v_cbp1 = vec_perm ((vector unsigned char) tmp6, (vector unsigned char) tmp6, tmp9); + v_cbp2 = (vector unsigned char) vec_perm ((vector unsigned char) tmp5, (vector unsigned char) tmp5, (vector unsigned char) v_permv); + v_cbp1 = (vector unsigned char) vec_sld ((vector unsigned char) v_cbp1,(vector unsigned char) v_cbp2, 12); + v_cbp_maskv = vec_or (v_cbp1, v_cbp2); + + tmp9 = vec_lvsl (12 + (top<<2), (unsigned char *) 0x0); + v_cbp1 = vec_perm ((vector unsigned char) tmp6, (vector unsigned char) tmp6, tmp9); + v_cbp2 = (vector unsigned char) vec_perm ((vector unsigned char) tmp5, (vector unsigned char) tmp5, (vector unsigned char) v_permh); + v_cbp1 = (vector unsigned char) vec_sld ((vector unsigned char) v_cbp1,(vector unsigned char) v_cbp2, 12); + v_cbp_maskh = vec_or (v_cbp1, v_cbp2); + + v_cbp_maskv = (vector unsigned char) vec_cmpgt ((vector unsigned char) v_cbp_maskv, (vector unsigned char) v_zero); + v_cbp_maskh = (vector unsigned char) vec_cmpgt ((vector unsigned char) v_cbp_maskh, (vector unsigned char) v_zero); + + intra[0] =0; + intra[1] =1; + intra[2] =2; + intra[3] =3; + intra[4] =4; + intra[5] = 5; + intra[6] =6; + intra[7] =7; + intra[8] =8; + intra[9] =9; + intra[10] =9; + intra[11] =9; + intra[12] = 0xff; + + idx0 = vec_ld (0, (signed short *) intra); + + v_c1 = (vector unsigned char) {'1','2','3','4','5','6','7','8','1','2','3','4','5','6','7','8'}; + + if (field_MBAFF){ + v0 = (vector signed short) vec_and ((vector unsigned char) idx0, v_c1); + idx0 = (vector signed short) vec_sra ((vector unsigned char) idx0, v_c1); + + v1 = vec_sld (v0, v0, 15); + v1 = (vector signed short) vec_pack (v1, v0); + + v2 = vec_sld (v1, v1, 2); + v3 = vec_sld (v1, v1, 10); + + v4 = (vector signed short) vec_cmpeq ((vector signed char) v1, (vector signed char) v2); + v5 = (vector signed short) vec_cmpeq ((vector signed char) v1, (vector signed char) v3); + v6 = (vector signed short) vec_cmpeq ((vector signed char) v2, (vector signed char) v3); + } + else { + v4 = v5 = v6 = vec_nor (v_zero, v_zero); + } + + tmp1 = (vector signed short) vec_sl ((vector unsigned char) idx0, v_c1); + v_c1 = vec_mergeh ((vector unsigned char) v_zero, v_c1); + tmp1 = (vector signed short) vec_add (tmp1, (vector signed short) v_c1); + + v_pocl = vec_ld (0, (vector unsigned char *) mv_const); + v_poch = vec_ld (0, (vector unsigned char *) mv_const); + tmp2 = (vector signed short) vec_perm (v_pocl, v_poch, (vector unsigned char) tmp1); + + v_pocl = vec_ld (0, (vector unsigned char *) mv_const); + v_poch = vec_ld (16, (vector unsigned char *) mv_const); + tmp1 = (vector signed short) vec_perm (v_pocl, v_poch, (vector unsigned char) tmp1); + tmp1 = vec_sel (tmp1, tmp2, (vector unsigned short) {0xffff,0xffff,0,0,0,0,0,0}); + + tmp3 = (vector signed short) vec_splat ((vector unsigned char) idx0, 12); + v_c1 = (vector unsigned char) vec_nor (v_zero, v_zero); + tmp0 = (vector signed short) vec_cmpeq ((vector signed char) idx0, (vector signed char) v_c1); + tmp1 = vec_sel (tmp1, (vector signed short) tmp3, (vector unsigned short) tmp0); + + tmp2 = vec_sld (tmp1, tmp1, 15); + tmp1 = (vector signed short) vec_pack (tmp2, tmp1); + + tmp2 = vec_sld (tmp1, tmp1, 2); + tmp3 = vec_sld (tmp1, tmp1, 10); + + tmp0 = (vector signed short) vec_cmpeq ((vector signed char) tmp1, (vector signed char) tmp2); + tmp4 = (vector signed short) vec_cmpeq ((vector signed char) tmp1, (vector signed char) tmp3); + tmp1 = (vector signed short) vec_cmpeq ((vector signed char) tmp2, (vector signed char) tmp3); + tmp0 = vec_and (tmp0, v4); + tmp4 = vec_and (tmp4, v5); + tmp1 = vec_and (tmp1, v6); + tmp2 = vec_sld ((vector signed short) tmp0, (vector signed short) tmp0, 8); + tmp3 = vec_sld ((vector signed short) tmp4, (vector signed short) tmp4, 8); + tmp5 = vec_sld ((vector signed short) tmp1, (vector signed short) tmp1, 8); + tmp0 = vec_and (tmp0, tmp2); + tmp4 = vec_and (tmp4, tmp3); + tmp1 = vec_and (tmp1, tmp5); + v_ref_mask00 = vec_mergeh ((vector signed short) tmp0, (vector signed short) v_c1); + v_ref_mask01 = vec_mergeh ((vector signed short) tmp4, (vector signed short) tmp1); + v_ref_mask00 = (vector signed short) vec_mergeh ((vector unsigned char) v_ref_mask00, (vector unsigned char) v_ref_mask00); + v_ref_mask01 = (vector signed short) vec_mergeh ((vector unsigned char) v_ref_mask01, (vector unsigned char) v_ref_mask01); + + v0 = vec_ld (0, (vector signed short *) mv_const); + v1 = vec_ld (16, (vector signed short *) mv_const); + v4 = vec_ld (64, (vector signed short *) mv_const); + v5 = vec_ld (80, (vector signed short *) mv_const); + v8 = vec_ld (0, (vector signed short *) mv_const); + v9 = vec_ld (16, (vector signed short *) mv_const); + + tmp0 = (vector signed short) vec_perm ((vector unsigned char) v8, + (vector unsigned char) v8, (vector unsigned char) {0,1,2,3,8,9,10,11,4,5,6,7,12,13,14,15}); + tmp1 = (vector signed short) vec_mergeh ((vector signed int) v0, (vector signed int) v1); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp3 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp4 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp0 = (vector signed short) vec_perm ((vector unsigned char) v9, (vector unsigned char) v9, + (vector unsigned char) {0,1,2,3,8,9,10,11,4,5,6,7,12,13,14,15}); + tmp1 = (vector signed short) vec_mergeh ((vector signed int) v4, (vector signed int) v5); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp5 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp6 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 12); + tmp6 = vec_sld (tmp4, tmp4, 12); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp00 = (vector signed short) vec_pack ((vector unsigned short) tmp3, (vector unsigned short) tmp4); + + tmp0 = (vector signed short) vec_mergeh ((vector signed int) v0, (vector signed int) v1); + tmp1 = (vector signed short) vec_mergel ((vector signed int) v0, (vector signed int) v1); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp3 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp4 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp0 = (vector signed short) vec_mergeh ((vector signed int) v4, (vector signed int) v5); + tmp1 = (vector signed short) vec_mergel ((vector signed int) v4, (vector signed int) v5); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp5 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp6 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 12); + tmp6 = vec_sld (tmp4, tmp4, 12); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp01 = (vector signed short) vec_pack ((vector unsigned short) tmp3, (vector unsigned short) tmp4); + + v2 = vec_ld (32, (vector signed short *) mv_const); + v3 = vec_ld (48, (vector signed short *) mv_const); + v6 = vec_ld (96, (vector signed short *) mv_const); + v7 = vec_ld (112,(vector signed short *) mv_const); + + tmp0 = (vector signed short) vec_mergel ((vector signed int) v0, (vector signed int) v1); + tmp1 = (vector signed short) vec_mergeh ((vector signed int) v2, (vector signed int) v3); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp3 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp4 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp0 = (vector signed short) vec_mergel ((vector signed int) v4, (vector signed int) v5); + tmp1 = (vector signed short) vec_mergeh ((vector signed int) v6, (vector signed int) v7); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp5 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp6 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 12); + tmp6 = vec_sld (tmp4, tmp4, 12); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp02 = (vector signed short) vec_pack ((vector unsigned short) tmp3, (vector unsigned short) tmp4); + + tmp0 = (vector signed short) vec_mergeh ((vector signed int) v2, (vector signed int) v3); + tmp1 = (vector signed short) vec_mergel ((vector signed int) v2, (vector signed int) v3); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp3 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp4 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp0 = (vector signed short) vec_mergeh ((vector signed int) v6, (vector signed int) v7); + tmp1 = (vector signed short) vec_mergel ((vector signed int) v6, (vector signed int) v7); + tmp2 = vec_sld (tmp1, tmp1, 8); + tmp5 = vec_sub (vec_max (tmp0, tmp1), vec_min (tmp0, tmp1)); + tmp6 = vec_sub (vec_max (tmp0, tmp2), vec_min (tmp0, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 12); + tmp6 = vec_sld (tmp4, tmp4, 12); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp03 = (vector signed short) vec_pack ((vector unsigned short) tmp3, (vector unsigned short) tmp4); + + tmp0 = (vector signed short) vec_pack ((vector unsigned int) tmp00, (vector unsigned int) tmp01); + tmp1 = (vector signed short) vec_pack ((vector unsigned int) tmp02, (vector unsigned int) tmp03); + tmp2 = (vector signed short) vec_mergeh ((vector signed int) tmp0, (vector signed int) tmp1); + tmp3 = (vector signed short) vec_mergel ((vector signed int) tmp0, (vector signed int) tmp1); + tmp4 = (vector signed short) vec_mergeh ((vector signed int) tmp2, (vector signed int) tmp3); + tmp5 = (vector signed short) vec_mergel ((vector signed int) tmp2, (vector signed int) tmp3); + tmp4 = vec_and (v_ref_mask00, tmp4); + tmp5 = vec_and (v_ref_mask01, tmp5); + + tmp0 = vec_nor (v_ref_mask00, v_ref_mask01); + tmp1 = vec_and (v_ref_mask00, v_ref_mask01); + tmp2 = vec_and (tmp4, tmp5); + tmp2 = vec_and (tmp2, tmp1); + tmp3 = vec_nor (tmp4, tmp5); + tmp3 = vec_nor (tmp3, tmp1); + v_vec_maskv = vec_or (tmp0, tmp2); + v_vec_maskv = vec_or (v_vec_maskv, tmp3); + + intra[0] = 1; + intra[1] = 1; + intra[2] = 2; + intra[3] = 3; + intra[4] = 2; + intra[5] = 2; + intra[6] = 2; + intra[7] = 1; + intra[8] = 1; + intra[9] = 5; + intra[10] = 5; + intra[11] = 5; + + intra[13] = 0; + intra[14] = 0; + intra[15] = 0; + + idx0 = vec_ld (0, (signed short *) intra); + + v_c1 = (vector unsigned char) {'1','2','3','4','5','6','7','8','1','2','3','4','5','6','7','8'}; + + if (field_MBAFF){ + v8 = (vector signed short) vec_and ((vector unsigned char) idx0, v_c1); + idx0 = (vector signed short) vec_sra ((vector unsigned char) idx0, v_c1); + + v9 = vec_sld (v8, v8, 15); + v9 = (vector signed short) vec_pack (v9, v8); + + v10 = vec_sld (v9, v9, 2); + v11 = vec_sld (v9, v9, 10); + + v8 = (vector signed short) vec_cmpeq ((vector signed char) v9, (vector signed char) v10); + v9 = (vector signed short) vec_cmpeq ((vector signed char) v9, (vector signed char) v11); + v10 = (vector signed short) vec_cmpeq ((vector signed char) v10, (vector signed char) v11); + } + else { + v8 = v9 = v10 = vec_nor (v_zero, v_zero); + } + + tmp1 = (vector signed short) vec_sl ((vector unsigned char) idx0, v_c1); + +if (1){ + int m; + unsigned char toto2[16] __attribute__((aligned(16))); + + printf("vc1\n"); + vec_st(v_c1, 0, (unsigned char *) toto2); + for (m=0; m<16;m++) {printf("%c ", toto2[m]);} + + printf("\nv_zero\n"); + + vec_st (v_zero, 0, (unsigned char *) toto2); + for (m=0; m< 16; m++) {printf("%c ", toto2[m]);} + printf("\n"); +} + + v_c1 = vec_mergeh ((vector unsigned char) v_zero, v_c1); + tmp1 = (vector signed short) vec_add (tmp1, (vector signed short) v_c1); + +if (1){ + vector unsigned char vres = + (vector unsigned char){'a','1','b','2','c','3','d','4','e','5','f','6','g','7','h','8'}; + unsigned char toto2[16] __attribute__((aligned(16))); + int m; + + printf("vc1\n"); + vec_st(v_c1, 0, (unsigned char *) toto2); + for (m=0; m<16;m++) {printf("%c ", toto2[m]);} + printf("\n"); + if (!vec_all_eq (vres, v_c1)) + abort(); +} + + v_pocl = vec_ld (32, (vector unsigned char *) mv_const); + v_poch = vec_ld (48, (vector unsigned char *) mv_const); + tmp2 = (vector signed short) vec_perm (v_pocl, v_poch, (vector unsigned char) tmp1); + + v_pocl = vec_ld (0, (vector unsigned char *) mv_const); + v_poch = vec_ld (16, (vector unsigned char *) mv_const); + + tmp1 = (vector signed short) vec_perm (v_pocl, v_poch, (vector unsigned char) tmp1); + + tmp1 = vec_sel (tmp1, tmp2, (vector unsigned short) {0xffff,0xffff,0,0,0,0,0,0}); + + + tmp3 = (vector signed short) vec_splat ((vector unsigned char) idx0, 12); + v_c1 = (vector unsigned char) vec_nor (v_zero, v_zero); + tmp0 = (vector signed short) vec_cmpeq ((vector signed char) idx0, (vector signed char) v_c1); + tmp1 = vec_sel (tmp1, (vector signed short) tmp3, (vector unsigned short) tmp0); + + tmp2 = vec_sld (tmp1, tmp1, 15); + tmp1 = (vector signed short) vec_pack (tmp2, tmp1); + + + tmp2 = vec_sld (tmp1, tmp1, 2); + tmp3 = vec_sld (tmp1, tmp1, 10); + + tmp0 = (vector signed short) vec_cmpeq ((vector signed char) tmp1, (vector signed char) tmp2); + tmp4 = (vector signed short) vec_cmpeq ((vector signed char) tmp1, (vector signed char) tmp3); + tmp1 = (vector signed short) vec_cmpeq ((vector signed char) tmp2, (vector signed char) tmp3); + tmp0 = vec_and (tmp0, v8); + tmp4 = vec_and (tmp4, v9); + tmp1 = vec_and (tmp1, v10); + tmp2 = vec_sld ((vector signed short) tmp0, (vector signed short) tmp0, 8); + tmp3 = vec_sld ((vector signed short) tmp4, (vector signed short) tmp4, 8); + tmp5 = vec_sld ((vector signed short) tmp1, (vector signed short) tmp1, 8); + tmp0 = vec_and (tmp0, tmp2); + tmp4 = vec_and (tmp4, tmp3); + tmp1 = vec_and (tmp1, tmp5); + v_ref_mask00 = vec_mergeh ((vector signed short) tmp0, (vector signed short) v_c1); + v_ref_mask01 = vec_mergeh ((vector signed short) tmp4, (vector signed short) tmp1); + v_ref_mask00 = (vector signed short) vec_mergeh ((vector unsigned char) v_ref_mask00, (vector unsigned char) v_ref_mask00); + v_ref_mask01 = (vector signed short) vec_mergeh ((vector unsigned char) v_ref_mask01, (vector unsigned char) v_ref_mask01); + + + v_permv= vec_ld (0, (vector unsigned char *) mv_const); + v8 = vec_ld (0, (vector signed short *) mv_const); + v9 = vec_ld (16, (vector signed short *) mv_const); + tmp2 = vec_perm (v0, v0, v_permv); + tmp3 = vec_sub (vec_max (v8, v0), vec_min (v8, v0)); + tmp4 = vec_sub (vec_max (v8, tmp2), vec_min (v8, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp2 = vec_perm (v2, v2, v_permv); + tmp5 = vec_sub (vec_max (v9, v2), vec_min (v9, v2)); + tmp6 = vec_sub (vec_max (v9, tmp2), vec_min (v9, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp00 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp4); + + tmp2 = vec_perm (v1, v1, v_permv); + tmp3 = vec_sub (vec_max (v0, v1), vec_min (v0, v1)); + tmp4 = vec_sub (vec_max (v0, tmp2), vec_min (v0, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp2 = vec_perm (v3, v3, v_permv); + tmp5 = vec_sub (vec_max (v2, v3), vec_min (v2, v3)); + tmp6 = vec_sub (vec_max (v2, tmp2), vec_min (v2, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp01 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp4); + + tmp2 = vec_perm (v4, v4, v_permv); + tmp3 = vec_sub (vec_max (v1, v4), vec_min (v1, v4)); + tmp4 = vec_sub (vec_max (v1, tmp2), vec_min (v1, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp2 = vec_perm (v6, v6, v_permv); + tmp5 = vec_sub (vec_max (v3, v6), vec_min (v3, v6)); + tmp6 = vec_sub (vec_max (v3, tmp2), vec_min (v3, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp02 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp4); + + + tmp2 = vec_perm (v5, v5, v_permv); + tmp3 = vec_sub (vec_max (v4, v5), vec_min (v4, v5)); + tmp4 = vec_sub (vec_max (v4, tmp2), vec_min (v4, tmp2)); + tmp3 = (vector signed short) vec_cmpgt (tmp3, v_three); + tmp4 = (vector signed short) vec_cmpgt (tmp4, v_three); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + + tmp2 = vec_perm (v7, v7, v_permv); + tmp5 = vec_sub (vec_max (v6, v7), vec_min (v6, v7)); + tmp6 = vec_sub (vec_max (v6, tmp2), vec_min (v6, tmp2)); + tmp5 = (vector signed short) vec_cmpgt (tmp5, v_three); + tmp6 = (vector signed short) vec_cmpgt (tmp6, v_three); + tmp0 = vec_sld (tmp5, tmp5, 14); + tmp1 = vec_sld (tmp6, tmp6, 14); + tmp5 = vec_or (tmp0, tmp5); + tmp6 = vec_or (tmp1, tmp6); + + tmp3 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp5); + tmp4 = (vector signed short) vec_pack ((vector unsigned int) tmp4, (vector unsigned int) tmp6); + tmp5 = vec_sld (tmp3, tmp3, 14); + tmp6 = vec_sld (tmp4, tmp4, 14); + tmp3 = vec_or (tmp3, tmp5); + tmp4 = vec_or (tmp4, tmp6); + tmp03 = (vector signed short) vec_pack ((vector unsigned int) tmp3, (vector unsigned int) tmp4); + + tmp0 = (vector signed short) vec_pack ((vector unsigned short) tmp00, (vector unsigned short) tmp01); + tmp1 = (vector signed short) vec_pack ((vector unsigned short) tmp02, (vector unsigned short) tmp03); + tmp2 = (vector signed short) vec_mergeh ((vector signed int) tmp0, (vector signed int) tmp1); + tmp3 = (vector signed short) vec_mergel ((vector signed int) tmp0, (vector signed int) tmp1); + tmp4 = (vector signed short) vec_mergeh ((vector signed int) tmp2, (vector signed int) tmp3); + tmp5 = (vector signed short) vec_mergel ((vector signed int) tmp2, (vector signed int) tmp3); + tmp4 = vec_and (v_ref_mask00, tmp4); + tmp5 = vec_and (v_ref_mask01, tmp5); + + tmp0 = vec_nor (v_ref_mask00, v_ref_mask01); + tmp1 = vec_and (v_ref_mask00, v_ref_mask01); + tmp2 = vec_and (tmp4, tmp5); + tmp2 = vec_and (tmp2, tmp1); + tmp3 = vec_nor (tmp4, tmp5); + tmp3 = vec_nor (tmp3, tmp1); + v_vec_maskh = vec_or (tmp0, tmp2); + v_vec_maskh = vec_or (v_vec_maskh, tmp3); + + + v_intra_maskvn = vec_nor (v_intra_maskv, v_intra_maskv); + v_intra_maskhn = vec_nor (v_intra_maskh, v_intra_maskh); + v_cbp_maskvn = (vector unsigned char) vec_cmpeq ((vector unsigned char) v_cbp_maskv, (vector unsigned char) v_zero); + v_cbp_maskhn = (vector unsigned char) vec_cmpeq ((vector unsigned char) v_cbp_maskh, (vector unsigned char) v_zero); + + v_cbp_maskv = vec_and (v_cbp_maskv, v_intra_maskvn); + v_cbp_maskh = vec_and (v_cbp_maskh, v_intra_maskhn); + v_vec_maskv = vec_and (v_vec_maskv, (vector signed short) v_intra_maskvn); + v_vec_maskv = vec_and (v_vec_maskv, (vector signed short) v_cbp_maskvn); + v_vec_maskh = vec_and (v_vec_maskh, (vector signed short) v_intra_maskhn); + v_vec_maskh = vec_and (v_vec_maskh, (vector signed short) v_cbp_maskhn); + + tmp9 = vec_splat_u8(2); + tmp8 = vec_splat_u8(1); + v_bS = vec_ld (0, (vector unsigned char *) mv_const); + + v_bSv = vec_and ((vector unsigned char) v_bS, (vector unsigned char)v_intra_maskv); + tmp7 = vec_and ((vector unsigned char)tmp9, (vector unsigned char)v_cbp_maskv); + tmp6 = (vector signed short) vec_and ((vector unsigned char)tmp8, (vector unsigned char)v_vec_maskv); + tmp7 = vec_or ((vector unsigned char)tmp7, (vector unsigned char)tmp6); + v_bSv = vec_or ((vector unsigned char)tmp7, (vector unsigned char)v_bSv); + + v_bS = vec_ld (0, (vector unsigned char *) mv_const); + v_bSh = vec_and ((vector unsigned char) v_bS, (vector unsigned char)v_intra_maskh); + tmp7 = vec_and ((vector unsigned char)tmp9, (vector unsigned char)v_cbp_maskh); + tmp6 = (vector signed short) vec_and ((vector unsigned char)tmp8, (vector unsigned char)v_vec_maskh); + tmp7 = vec_or ((vector unsigned char)tmp7, (vector unsigned char)tmp6); + v_bSh = vec_or ((vector unsigned char)tmp7, (vector unsigned char)v_bSh); + + v_permh = (vector unsigned char) vec_ld (0 , (vector unsigned char *) mv_const); + v_permv = (vector unsigned char) vec_ld (0, (vector unsigned char *) mv_const); + v_bSv = vec_and (v_bSv, v_permv); + v_bSh = vec_and (v_bSh, v_permh); + + vec_st (v_bSv, 0, (unsigned char *) mv_const); + vec_st (v_bSh, 0, (unsigned char *) mv_const); + + v_bSv = vec_mergeh (v_bSv, v_bSv); + v_bSv = vec_mergeh (v_bSv, v_bSv); + v_bSh = vec_mergeh (v_bSh, v_bSh); + v_bSh = vec_mergeh (v_bSh, v_bSh); + + vec_st (v_bSv, 0, (vector unsigned char *) mv_const); + vec_st (v_bSh, 0,(vector unsigned char *) mv_const); +} + + +int main(int argc, char **argv) +{ + char toto[32] __attribute__((aligned(16))); + foo(toto, toto, 0, 0); + return 0; +}