[X86] Split X87/SSE compare classes into WriteFCom + WriteFComX

Most X87 compare instructions write to the X87 status word, while the SSE (U)COMI compares write to rFLAGS. These are often handled very differently on CPUs (e.g. rFLAGS outputs typically involve a fpu2gpr transfer), and we shouldn't be grouping all these instructions behind a single class - so this patch splits off the SSE compares into a new WriteFComX class (and currently keeps the same behaviours). If there's a need to distinguish between X87 instructions more closely we can investigate that in the future, but as we don't handle any of the X87 side effects at the moment its unlikely to have any notable effect.
llvm · Jan 17, 2020 · 8eb4d25 · 8eb4d25
1 parent 1dc2f25
commit 8eb4d25
Show file tree

Hide file tree

Showing 14 changed files with 19 additions and 7 deletions.
diff --git a/llvm/lib/Target/X86/X86InstrAVX512.td b/llvm/lib/Target/X86/X86InstrAVX512.td
@@ -8552,7 +8552,7 @@ let Predicates = [HasVLX] in {
 //  Unordered/Ordered scalar fp compare with Sae and set EFLAGS
 multiclass avx512_ord_cmp_sae<bits<8> opc, X86VectorVTInfo _,
                             string OpcodeStr, Domain d,
-                            X86FoldableSchedWrite sched = WriteFCom> {
+                            X86FoldableSchedWrite sched = WriteFComX> {
   let hasSideEffects = 0, Uses = [MXCSR] in
   def rrb: AVX512<opc, MRMSrcReg, (outs), (ins _.RC:$src1, _.RC:$src2),
                   !strconcat(OpcodeStr, "\t{{sae}, $src2, $src1|$src1, $src2, {sae}}"), []>,

diff --git a/llvm/lib/Target/X86/X86InstrSSE.td b/llvm/lib/Target/X86/X86InstrSSE.td
@@ -1830,7 +1830,7 @@ let Constraints = "$src1 = $dst" in {
 multiclass sse12_ord_cmp<bits<8> opc, RegisterClass RC, SDNode OpNode,
                          ValueType vt, X86MemOperand x86memop,
                          PatFrag ld_frag, string OpcodeStr, Domain d,
-                         X86FoldableSchedWrite sched = WriteFCom> {
+                         X86FoldableSchedWrite sched = WriteFComX> {
 let hasSideEffects = 0, Uses = [MXCSR], mayRaiseFPException = 1,
     ExeDomain = d in {
   def rr: SI<opc, MRMSrcReg, (outs), (ins RC:$src1, RC:$src2),
@@ -1851,7 +1851,7 @@ multiclass sse12_ord_cmp_int<bits<8> opc, RegisterClass RC, SDNode OpNode,
                              ValueType vt, Operand memop,
                              ComplexPattern mem_cpat, string OpcodeStr,
                              Domain d,
-                             X86FoldableSchedWrite sched = WriteFCom> {
+                             X86FoldableSchedWrite sched = WriteFComX> {
 let Uses = [MXCSR], mayRaiseFPException = 1, ExeDomain = d in {
   def rr_Int: SI<opc, MRMSrcReg, (outs), (ins RC:$src1, RC:$src2),
                      !strconcat(OpcodeStr, "\t{$src2, $src1|$src1, $src2}"),

diff --git a/llvm/lib/Target/X86/X86SchedBroadwell.td b/llvm/lib/Target/X86/X86SchedBroadwell.td
@@ -260,7 +260,8 @@ defm : BWWriteResPair<WriteFCmp64X, [BWPort1],  3, [1], 1, 5>; // Floating point
 defm : BWWriteResPair<WriteFCmp64Y, [BWPort1],  3, [1], 1, 6>; // Floating point double compare (YMM/ZMM).
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 
-defm : BWWriteResPair<WriteFCom,    [BWPort1],  3>; // Floating point compare to flags.
+defm : BWWriteResPair<WriteFCom,    [BWPort1],  3>; // Floating point compare to flags (X87).
+defm : BWWriteResPair<WriteFComX,   [BWPort1],  3>; // Floating point compare to flags (SSE).
 
 defm : BWWriteResPair<WriteFMul,    [BWPort01], 3, [1], 1, 5>; // Floating point multiplication.
 defm : BWWriteResPair<WriteFMulX,   [BWPort01], 3, [1], 1, 5>; // Floating point multiplication (XMM).

diff --git a/llvm/lib/Target/X86/X86SchedHaswell.td b/llvm/lib/Target/X86/X86SchedHaswell.td
@@ -261,6 +261,7 @@ defm : HWWriteResPair<WriteFCmp64Y, [HWPort1],  3, [1], 1, 7>;
 defm : HWWriteResPair<WriteFCmp64Z, [HWPort1],  3, [1], 1, 7>; // Unsupported = 1
 
 defm : HWWriteResPair<WriteFCom,    [HWPort1],  3>;
+defm : HWWriteResPair<WriteFComX,   [HWPort1],  3>;
 
 defm : HWWriteResPair<WriteFMul,    [HWPort01],  5, [1], 1, 5>;
 defm : HWWriteResPair<WriteFMulX,   [HWPort01],  5, [1], 1, 6>;

diff --git a/llvm/lib/Target/X86/X86SchedSandyBridge.td b/llvm/lib/Target/X86/X86SchedSandyBridge.td
@@ -238,6 +238,7 @@ defm : SBWriteResPair<WriteFCmp64Y, [SBPort1],  3, [1], 1, 7>;
 defm : SBWriteResPair<WriteFCmp64Z, [SBPort1],  3, [1], 1, 7>; // Unsupported = 1
 
 defm : SBWriteResPair<WriteFCom,    [SBPort1],  3>;
+defm : SBWriteResPair<WriteFComX,   [SBPort1],  3>;
 
 defm : SBWriteResPair<WriteFMul,    [SBPort0],  5, [1], 1, 6>;
 defm : SBWriteResPair<WriteFMulX,   [SBPort0],  5, [1], 1, 6>;

diff --git a/llvm/lib/Target/X86/X86SchedSkylakeClient.td b/llvm/lib/Target/X86/X86SchedSkylakeClient.td
@@ -255,7 +255,8 @@ defm : SKLWriteResPair<WriteFCmp64X,  [SKLPort01],  4, [1], 1, 6>;
 defm : SKLWriteResPair<WriteFCmp64Y,  [SKLPort01],  4, [1], 1, 7>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 
-defm : SKLWriteResPair<WriteFCom,      [SKLPort0],  2>; // Floating point compare to flags.
+defm : SKLWriteResPair<WriteFCom,      [SKLPort0],  2>; // Floating point compare to flags (X87).
+defm : SKLWriteResPair<WriteFComX,     [SKLPort0],  2>; // Floating point compare to flags (SSE).
 
 defm : SKLWriteResPair<WriteFMul,     [SKLPort01],  4, [1], 1, 5>; // Floating point multiplication.
 defm : SKLWriteResPair<WriteFMulX,    [SKLPort01],  4, [1], 1, 6>;

diff --git a/llvm/lib/Target/X86/X86SchedSkylakeServer.td b/llvm/lib/Target/X86/X86SchedSkylakeServer.td
@@ -255,7 +255,8 @@ defm : SKXWriteResPair<WriteFCmp64X,   [SKXPort01],  4, [1], 1, 6>;
 defm : SKXWriteResPair<WriteFCmp64Y,   [SKXPort01],  4, [1], 1, 7>;
 defm : SKXWriteResPair<WriteFCmp64Z,   [SKXPort05],  4, [1], 1, 7>;
 
-defm : SKXWriteResPair<WriteFCom,       [SKXPort0],  2>; // Floating point compare to flags.
+defm : SKXWriteResPair<WriteFCom,       [SKXPort0],  2>; // Floating point compare to flags (X87).
+defm : SKXWriteResPair<WriteFComX,      [SKXPort0],  2>; // Floating point compare to flags (SSE).
 
 defm : SKXWriteResPair<WriteFMul,      [SKXPort01],  4, [1], 1, 5>; // Floating point multiplication.
 defm : SKXWriteResPair<WriteFMulX,     [SKXPort01],  4, [1], 1, 6>;

diff --git a/llvm/lib/Target/X86/X86Schedule.td b/llvm/lib/Target/X86/X86Schedule.td
@@ -250,7 +250,8 @@ defm WriteFCmp64  : X86SchedWritePair<ReadAfterVecLd>;  // Floating point double
 defm WriteFCmp64X : X86SchedWritePair<ReadAfterVecXLd>; // Floating point double compare (XMM).
 defm WriteFCmp64Y : X86SchedWritePair<ReadAfterVecYLd>; // Floating point double compare (YMM).
 defm WriteFCmp64Z : X86SchedWritePair<ReadAfterVecYLd>; // Floating point double compare (ZMM).
-defm WriteFCom    : X86SchedWritePair<ReadAfterVecLd>;  // Floating point compare to flags.
+defm WriteFCom    : X86SchedWritePair<ReadAfterVecLd>;  // Floating point compare to flags (X87).
+defm WriteFComX   : X86SchedWritePair<ReadAfterVecLd>;  // Floating point compare to flags (SSE).
 defm WriteFMul    : X86SchedWritePair<ReadAfterVecLd>;  // Floating point multiplication.
 defm WriteFMulX   : X86SchedWritePair<ReadAfterVecXLd>; // Floating point multiplication (XMM).
 defm WriteFMulY   : X86SchedWritePair<ReadAfterVecYLd>; // Floating point multiplication (YMM).

diff --git a/llvm/lib/Target/X86/X86ScheduleAtom.td b/llvm/lib/Target/X86/X86ScheduleAtom.td
@@ -244,6 +244,7 @@ defm : AtomWriteResPair<WriteFCmp64X,       [AtomPort01], [AtomPort01],  6,  7,
 defm : X86WriteResPairUnsupported<WriteFCmp64Y>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 defm : AtomWriteResPair<WriteFCom,           [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
+defm : AtomWriteResPair<WriteFComX,          [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : AtomWriteResPair<WriteFMul,           [AtomPort0],  [AtomPort0],  4,  4,  [4],  [4]>;
 defm : AtomWriteResPair<WriteFMulX,          [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : X86WriteResPairUnsupported<WriteFMulY>;

diff --git a/llvm/lib/Target/X86/X86ScheduleBdVer2.td b/llvm/lib/Target/X86/X86ScheduleBdVer2.td
@@ -766,6 +766,7 @@ defm : PdWriteResYMMPair<WriteFCmp64Y,      [PdFPU0, PdFPFMA],  2, [1, 2]>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 
 defm : PdWriteResXMMPair<WriteFCom,         [PdFPU0, PdFPFMA, PdEX0], 1, [], 2>;
+defm : PdWriteResXMMPair<WriteFComX,        [PdFPU0, PdFPFMA, PdEX0], 1, [], 2>;
 
 def PdWriteFCOMPm : SchedWriteRes<[PdFPU1, PdFPFMA]> {
   let Latency = 6;

diff --git a/llvm/lib/Target/X86/X86ScheduleBtVer2.td b/llvm/lib/Target/X86/X86ScheduleBtVer2.td
@@ -541,6 +541,7 @@ defm : JWriteResFpuPair<WriteFCmp64X,      [JFPU0, JFPA],  2>;
 defm : JWriteResYMMPair<WriteFCmp64Y,      [JFPU0, JFPA],  2, [2,2], 2>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 defm : JWriteResFpuPair<WriteFCom,  [JFPU0, JFPA, JALU0],  3>;
+defm : JWriteResFpuPair<WriteFComX, [JFPU0, JFPA, JALU0],  3>;
 defm : JWriteResFpuPair<WriteFMul,         [JFPU1, JFPM],  2>;
 defm : JWriteResFpuPair<WriteFMulX,        [JFPU1, JFPM],  2>;
 defm : JWriteResYMMPair<WriteFMulY,        [JFPU1, JFPM],  2, [2,2], 2>;

diff --git a/llvm/lib/Target/X86/X86ScheduleSLM.td b/llvm/lib/Target/X86/X86ScheduleSLM.td
@@ -214,6 +214,7 @@ defm : SLMWriteResPair<WriteFCmp64X,  [SLM_FPC_RSV1], 3>;
 defm : SLMWriteResPair<WriteFCmp64Y,  [SLM_FPC_RSV1], 3>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 defm : SLMWriteResPair<WriteFCom,     [SLM_FPC_RSV1], 3>;
+defm : SLMWriteResPair<WriteFComX,    [SLM_FPC_RSV1], 3>;
 defm : SLMWriteResPair<WriteFMul,     [SLM_FPC_RSV0, SLMFPMultiplier], 5, [1,2]>;
 defm : SLMWriteResPair<WriteFMulX,    [SLM_FPC_RSV0, SLMFPMultiplier], 5, [1,2]>;
 defm : SLMWriteResPair<WriteFMulY,    [SLM_FPC_RSV0, SLMFPMultiplier], 5, [1,2]>;

diff --git a/llvm/lib/Target/X86/X86ScheduleZnver1.td b/llvm/lib/Target/X86/X86ScheduleZnver1.td
@@ -295,6 +295,7 @@ defm : ZnWriteResFpuPair<WriteFCmp64X,   [ZnFPU0],  3>;
 defm : ZnWriteResFpuPair<WriteFCmp64Y,   [ZnFPU0],  3>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 defm : ZnWriteResFpuPair<WriteFCom,      [ZnFPU0],  3>;
+defm : ZnWriteResFpuPair<WriteFComX,     [ZnFPU0],  3>;
 defm : ZnWriteResFpuPair<WriteFBlend,    [ZnFPU01], 1>;
 defm : ZnWriteResFpuPair<WriteFBlendY,   [ZnFPU01], 1>;
 defm : X86WriteResPairUnsupported<WriteFBlendZ>;

diff --git a/llvm/lib/Target/X86/X86ScheduleZnver2.td b/llvm/lib/Target/X86/X86ScheduleZnver2.td
@@ -281,6 +281,7 @@ defm : Zn2WriteResFpuPair<WriteFCmp64X,   [Zn2FPU0],  3>;
 defm : Zn2WriteResFpuPair<WriteFCmp64Y,   [Zn2FPU0],  3>;
 defm : X86WriteResPairUnsupported<WriteFCmp64Z>;
 defm : Zn2WriteResFpuPair<WriteFCom,      [Zn2FPU0],  3>;
+defm : Zn2WriteResFpuPair<WriteFComX,     [Zn2FPU0],  3>;
 defm : Zn2WriteResFpuPair<WriteFBlend,    [Zn2FPU01], 1>;
 defm : Zn2WriteResFpuPair<WriteFBlendY,   [Zn2FPU01], 1>;
 defm : X86WriteResPairUnsupported<WriteFBlendZ>;