diff options
Diffstat (limited to 'lib/Target/X86/X86InstrSSE.td')
| -rw-r--r-- | lib/Target/X86/X86InstrSSE.td | 25 |
1 files changed, 15 insertions, 10 deletions
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index 63065c9..7ed69ea 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -808,7 +808,7 @@ let isTwoAddress = 1 in { "shufps\t{$src3, $src2, $dst|$dst, $src2, $src3}", [(set VR128:$dst, (v4f32 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv4f32 addr:$src2), SHUFP_shuffle_mask:$src3)))]>; let AddedComplexity = 10 in { @@ -824,7 +824,7 @@ let isTwoAddress = 1 in { "unpckhps\t{$src2, $dst|$dst, $src2}", [(set VR128:$dst, (v4f32 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv4f32 addr:$src2), UNPCKH_shuffle_mask)))]>; def UNPCKLPSrr : PSI<0x14, MRMSrcReg, @@ -839,7 +839,7 @@ let isTwoAddress = 1 in { "unpcklps\t{$src2, $dst|$dst, $src2}", [(set VR128:$dst, (v4f32 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv4f32 addr:$src2), UNPCKL_shuffle_mask)))]>; } // AddedComplexity } // isTwoAddress @@ -1561,7 +1561,7 @@ let isTwoAddress = 1 in { "shufpd\t{$src3, $src2, $dst|$dst, $src2, $src3}", [(set VR128:$dst, (v2f64 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv2f64 addr:$src2), SHUFP_shuffle_mask:$src3)))]>; let AddedComplexity = 10 in { @@ -1577,7 +1577,7 @@ let isTwoAddress = 1 in { "unpckhpd\t{$src2, $dst|$dst, $src2}", [(set VR128:$dst, (v2f64 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv2f64 addr:$src2), UNPCKH_shuffle_mask)))]>; def UNPCKLPDrr : PDI<0x14, MRMSrcReg, @@ -1592,7 +1592,7 @@ let isTwoAddress = 1 in { "unpcklpd\t{$src2, $dst|$dst, $src2}", [(set VR128:$dst, (v2f64 (vector_shuffle - VR128:$src1, (load addr:$src2), + VR128:$src1, (memopv2f64 addr:$src2), UNPCKL_shuffle_mask)))]>; } // AddedComplexity } // isTwoAddress @@ -1782,7 +1782,7 @@ let isTwoAddress = 1 in { (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2), "pandn\t{$src2, $dst|$dst, $src2}", [(set VR128:$dst, (v2i64 (and (vnot VR128:$src1), - (load addr:$src2))))]>; + (memopv2i64 addr:$src2))))]>; } // SSE2 Integer comparison @@ -2419,6 +2419,11 @@ def : Pat<(vector_shuffle (v4f32 VR128:$src1), (undef), SHUFP_unary_shuffle_mask:$sm), (SHUFPSrri VR128:$src1, VR128:$src1, SHUFP_unary_shuffle_mask:$sm)>, Requires<[HasSSE1]>; +// Special unary SHUFPDrri case. +def : Pat<(vector_shuffle (v2f64 VR128:$src1), (undef), + SHUFP_unary_shuffle_mask:$sm), + (SHUFPDrri VR128:$src1, VR128:$src1, SHUFP_unary_shuffle_mask:$sm)>, + Requires<[HasSSE2]>; // Unary v4f32 shuffle with PSHUF* in order to fold a load. def : Pat<(vector_shuffle (memopv4f32 addr:$src1), (undef), SHUFP_unary_shuffle_mask:$sm), @@ -2583,13 +2588,13 @@ def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v16i8 immAllOnesV))), (PANDNrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>; def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v4i32 immAllOnesV))), - (load addr:$src2))), + (memopv2i64 addr:$src2))), (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v8i16 immAllOnesV))), - (load addr:$src2))), + (memopv2i64 addr:$src2))), (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v16i8 immAllOnesV))), - (load addr:$src2))), + (memopv2i64 addr:$src2))), (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>; // Use movaps / movups for SSE integer load / store (one byte shorter). |