diff options
Diffstat (limited to 'lib/Target')
| -rw-r--r-- | lib/Target/X86/X86InstrSSE.td | 10 |
1 files changed, 8 insertions, 2 deletions
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td index 8e5f37c867..0d5d1b449e 100644 --- a/lib/Target/X86/X86InstrSSE.td +++ b/lib/Target/X86/X86InstrSSE.td @@ -1937,6 +1937,10 @@ multiclass sse1_fp_unop_s<bits<8> opc, string OpcodeStr, def SSr : SSI<opc, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src), !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"), [(set FR32:$dst, (OpNode FR32:$src))]>; + // For scalar unary operations, fold a load into the operation + // only in OptForSize mode. It eliminates an instruction, but it also + // eliminates a whole-register clobber (the load), so it introduces a + // partial register update condition. def SSm : I<opc, MRMSrcMem, (outs FR32:$dst), (ins f32mem:$src), !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"), [(set FR32:$dst, (OpNode (load addr:$src)))]>, XS, @@ -1992,9 +1996,11 @@ multiclass sse2_fp_unop_s<bits<8> opc, string OpcodeStr, def SDr : SDI<opc, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src), !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"), [(set FR64:$dst, (OpNode FR64:$src))]>; - def SDm : SDI<opc, MRMSrcMem, (outs FR64:$dst), (ins f64mem:$src), + // See the comments in sse1_fp_unop_s for why this is OptForSize. + def SDm : I<opc, MRMSrcMem, (outs FR64:$dst), (ins f64mem:$src), !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"), - [(set FR64:$dst, (OpNode (load addr:$src)))]>; + [(set FR64:$dst, (OpNode (load addr:$src)))]>, XD, + Requires<[HasSSE2, OptForSize]>; def SDr_Int : SDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src), !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"), [(set VR128:$dst, (F64Int VR128:$src))]>; |