[llvm] bf527a1 - AMDGPU/GlobalISel: Fix f64 G_FDIV lowering
Matt Arsenault via llvm-commits
llvm-commits at lists.llvm.org
Mon May 18 12:14:19 PDT 2020
Author: Matt Arsenault
Date: 2020-05-18T15:14:08-04:00
New Revision: bf527a1dc410eddf5b2e176b9092bfb1de75df0e
URL: https://github.com/llvm/llvm-project/commit/bf527a1dc410eddf5b2e176b9092bfb1de75df0e
DIFF: https://github.com/llvm/llvm-project/commit/bf527a1dc410eddf5b2e176b9092bfb1de75df0e.diff
LOG: AMDGPU/GlobalISel: Fix f64 G_FDIV lowering
This was using an integer multiply instead of FP.
Added:
Modified:
llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fdiv.mir
Removed:
################################################################################
diff --git a/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
index 4f2e7e664ed6..554075ac8ab4 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
@@ -2994,7 +2994,7 @@ bool AMDGPULegalizerInfo::legalizeFDIV64(MachineInstr &MI,
.setMIFlags(Flags);
auto Fma3 = B.buildFMA(S64, Fma1, Fma2, Fma1, Flags);
- auto Mul = B.buildMul(S64, DivScale1.getReg(0), Fma3, Flags);
+ auto Mul = B.buildFMul(S64, DivScale1.getReg(0), Fma3, Flags);
auto Fma4 = B.buildFMA(S64, NegDivScale0, Mul, DivScale1.getReg(0), Flags);
Register Scale;
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fdiv.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fdiv.mir
index b1db5a9e2d71..7d5841679769 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fdiv.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fdiv.mir
@@ -366,24 +366,16 @@ body: |
; SI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; SI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[COPY]](s64), [[COPY1]](s64), 1
; SI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; SI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; SI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; SI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; SI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; SI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; SI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; SI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
- ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
- ; SI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
- ; SI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV5]](s32), [[UV11]]
- ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV7]](s32), [[UV9]]
+ ; SI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
+ ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+ ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
+ ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
+ ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV1]](s32), [[UV7]]
+ ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV3]](s32), [[UV5]]
; SI: [[XOR:%[0-9]+]]:_(s1) = G_XOR [[ICMP]], [[ICMP1]]
- ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[XOR]](s1)
+ ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[XOR]](s1)
; SI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY1]](s64), [[COPY]](s64)
; SI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; VI-LABEL: name: test_fdiv_s64
@@ -398,17 +390,9 @@ body: |
; VI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; VI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[COPY]](s64), [[COPY1]](s64), 1
; VI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; VI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; VI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; VI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; VI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; VI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; VI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; VI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; VI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; VI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; VI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY1]](s64), [[COPY]](s64)
; VI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-LABEL: name: test_fdiv_s64
@@ -423,17 +407,9 @@ body: |
; GFX9: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX9: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[COPY]](s64), [[COPY1]](s64), 1
; GFX9: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX9: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX9: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX9: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX9: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX9: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX9: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX9: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX9: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX9: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX9: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY1]](s64), [[COPY]](s64)
; GFX9: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-UNSAFE-LABEL: name: test_fdiv_s64
@@ -454,17 +430,9 @@ body: |
; GFX10: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX10: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[COPY]](s64), [[COPY1]](s64), 1
; GFX10: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX10: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX10: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX10: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX10: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX10: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX10: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX10: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX10: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX10: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX10: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX10: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY1]](s64), [[COPY]](s64)
; GFX10: $vgpr0_vgpr1 = COPY [[INT6]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
@@ -1019,24 +987,16 @@ body: |
; SI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; SI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV]](s64), [[UV2]](s64), 1
; SI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; SI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV6]]
- ; SI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV5]], [[UV6]]
- ; SI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV7]]
- ; SI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV4]], [[UV6]]
- ; SI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; SI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; SI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; SI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV]](s64)
- ; SI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV2]](s64)
- ; SI: [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
- ; SI: [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV9]](s32), [[UV15]]
- ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV11]](s32), [[UV13]]
+ ; SI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV]](s64)
+ ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV2]](s64)
+ ; SI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
+ ; SI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
+ ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV5]](s32), [[UV11]]
+ ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV7]](s32), [[UV9]]
; SI: [[XOR:%[0-9]+]]:_(s1) = G_XOR [[ICMP]], [[ICMP1]]
- ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[XOR]](s1)
+ ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[XOR]](s1)
; SI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[UV2]](s64), [[UV]](s64)
; SI: [[INT7:%[0-9]+]]:_(s64), [[INT8:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 0
; SI: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[INT7]]
@@ -1046,24 +1006,16 @@ body: |
; SI: [[FMA7:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMA6]], [[C]]
; SI: [[INT10:%[0-9]+]]:_(s64), [[INT11:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 1
; SI: [[FMA8:%[0-9]+]]:_(s64) = G_FMA [[FMA6]], [[FMA7]], [[FMA6]]
- ; SI: [[UV16:%[0-9]+]]:_(s32), [[UV17:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
- ; SI: [[UV18:%[0-9]+]]:_(s32), [[UV19:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA8]](s64)
- ; SI: [[MUL3:%[0-9]+]]:_(s32) = G_MUL [[UV16]], [[UV18]]
- ; SI: [[MUL4:%[0-9]+]]:_(s32) = G_MUL [[UV17]], [[UV18]]
- ; SI: [[MUL5:%[0-9]+]]:_(s32) = G_MUL [[UV16]], [[UV19]]
- ; SI: [[UMULH1:%[0-9]+]]:_(s32) = G_UMULH [[UV16]], [[UV18]]
- ; SI: [[ADD2:%[0-9]+]]:_(s32) = G_ADD [[MUL4]], [[MUL5]]
- ; SI: [[ADD3:%[0-9]+]]:_(s32) = G_ADD [[ADD2]], [[UMULH1]]
- ; SI: [[MV1:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL3]](s32), [[ADD3]](s32)
- ; SI: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[MV1]], [[INT10]]
- ; SI: [[UV20:%[0-9]+]]:_(s32), [[UV21:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV1]](s64)
- ; SI: [[UV22:%[0-9]+]]:_(s32), [[UV23:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV3]](s64)
- ; SI: [[UV24:%[0-9]+]]:_(s32), [[UV25:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT7]](s64)
- ; SI: [[UV26:%[0-9]+]]:_(s32), [[UV27:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
- ; SI: [[ICMP2:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV21]](s32), [[UV27]]
- ; SI: [[ICMP3:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV23]](s32), [[UV25]]
+ ; SI: [[FMUL1:%[0-9]+]]:_(s64) = G_FMUL [[INT10]], [[FMA8]]
+ ; SI: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMUL1]], [[INT10]]
+ ; SI: [[UV12:%[0-9]+]]:_(s32), [[UV13:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV1]](s64)
+ ; SI: [[UV14:%[0-9]+]]:_(s32), [[UV15:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[UV3]](s64)
+ ; SI: [[UV16:%[0-9]+]]:_(s32), [[UV17:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT7]](s64)
+ ; SI: [[UV18:%[0-9]+]]:_(s32), [[UV19:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
+ ; SI: [[ICMP2:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV13]](s32), [[UV19]]
+ ; SI: [[ICMP3:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV15]](s32), [[UV17]]
; SI: [[XOR1:%[0-9]+]]:_(s1) = G_XOR [[ICMP2]], [[ICMP3]]
- ; SI: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[MV1]](s64), [[XOR1]](s1)
+ ; SI: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[FMUL1]](s64), [[XOR1]](s1)
; SI: [[INT13:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT12]](s64), [[UV3]](s64), [[UV1]](s64)
; SI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[INT6]](s64), [[INT13]](s64)
; SI: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
@@ -1081,17 +1033,9 @@ body: |
; VI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; VI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV]](s64), [[UV2]](s64), 1
; VI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; VI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; VI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; VI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV6]]
- ; VI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV5]], [[UV6]]
- ; VI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV7]]
- ; VI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV4]], [[UV6]]
- ; VI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; VI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; VI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; VI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; VI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[UV2]](s64), [[UV]](s64)
; VI: [[INT7:%[0-9]+]]:_(s64), [[INT8:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 0
; VI: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[INT7]]
@@ -1101,17 +1045,9 @@ body: |
; VI: [[FMA7:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMA6]], [[C]]
; VI: [[INT10:%[0-9]+]]:_(s64), [[INT11:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 1
; VI: [[FMA8:%[0-9]+]]:_(s64) = G_FMA [[FMA6]], [[FMA7]], [[FMA6]]
- ; VI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
- ; VI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA8]](s64)
- ; VI: [[MUL3:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV10]]
- ; VI: [[MUL4:%[0-9]+]]:_(s32) = G_MUL [[UV9]], [[UV10]]
- ; VI: [[MUL5:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV11]]
- ; VI: [[UMULH1:%[0-9]+]]:_(s32) = G_UMULH [[UV8]], [[UV10]]
- ; VI: [[ADD2:%[0-9]+]]:_(s32) = G_ADD [[MUL4]], [[MUL5]]
- ; VI: [[ADD3:%[0-9]+]]:_(s32) = G_ADD [[ADD2]], [[UMULH1]]
- ; VI: [[MV1:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL3]](s32), [[ADD3]](s32)
- ; VI: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[MV1]], [[INT10]]
- ; VI: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[MV1]](s64), [[INT11]](s1)
+ ; VI: [[FMUL1:%[0-9]+]]:_(s64) = G_FMUL [[INT10]], [[FMA8]]
+ ; VI: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMUL1]], [[INT10]]
+ ; VI: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[FMUL1]](s64), [[INT11]](s1)
; VI: [[INT13:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT12]](s64), [[UV3]](s64), [[UV1]](s64)
; VI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[INT6]](s64), [[INT13]](s64)
; VI: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
@@ -1129,17 +1065,9 @@ body: |
; GFX9: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX9: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV]](s64), [[UV2]](s64), 1
; GFX9: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX9: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX9: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX9: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV6]]
- ; GFX9: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV5]], [[UV6]]
- ; GFX9: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV7]]
- ; GFX9: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV4]], [[UV6]]
- ; GFX9: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX9: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX9: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX9: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX9: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[UV2]](s64), [[UV]](s64)
; GFX9: [[INT7:%[0-9]+]]:_(s64), [[INT8:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 0
; GFX9: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[INT7]]
@@ -1149,17 +1077,9 @@ body: |
; GFX9: [[FMA7:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMA6]], [[C]]
; GFX9: [[INT10:%[0-9]+]]:_(s64), [[INT11:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 1
; GFX9: [[FMA8:%[0-9]+]]:_(s64) = G_FMA [[FMA6]], [[FMA7]], [[FMA6]]
- ; GFX9: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
- ; GFX9: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA8]](s64)
- ; GFX9: [[MUL3:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV10]]
- ; GFX9: [[MUL4:%[0-9]+]]:_(s32) = G_MUL [[UV9]], [[UV10]]
- ; GFX9: [[MUL5:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV11]]
- ; GFX9: [[UMULH1:%[0-9]+]]:_(s32) = G_UMULH [[UV8]], [[UV10]]
- ; GFX9: [[ADD2:%[0-9]+]]:_(s32) = G_ADD [[MUL4]], [[MUL5]]
- ; GFX9: [[ADD3:%[0-9]+]]:_(s32) = G_ADD [[ADD2]], [[UMULH1]]
- ; GFX9: [[MV1:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL3]](s32), [[ADD3]](s32)
- ; GFX9: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[MV1]], [[INT10]]
- ; GFX9: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[MV1]](s64), [[INT11]](s1)
+ ; GFX9: [[FMUL1:%[0-9]+]]:_(s64) = G_FMUL [[INT10]], [[FMA8]]
+ ; GFX9: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMUL1]], [[INT10]]
+ ; GFX9: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[FMUL1]](s64), [[INT11]](s1)
; GFX9: [[INT13:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT12]](s64), [[UV3]](s64), [[UV1]](s64)
; GFX9: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[INT6]](s64), [[INT13]](s64)
; GFX9: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
@@ -1188,17 +1108,9 @@ body: |
; GFX10: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX10: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV]](s64), [[UV2]](s64), 1
; GFX10: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX10: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX10: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX10: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV6]]
- ; GFX10: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV5]], [[UV6]]
- ; GFX10: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV4]], [[UV7]]
- ; GFX10: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV4]], [[UV6]]
- ; GFX10: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX10: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX10: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX10: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX10: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[UV2]](s64), [[UV]](s64)
; GFX10: [[INT7:%[0-9]+]]:_(s64), [[INT8:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 0
; GFX10: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[INT7]]
@@ -1208,17 +1120,9 @@ body: |
; GFX10: [[FMA7:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMA6]], [[C]]
; GFX10: [[INT10:%[0-9]+]]:_(s64), [[INT11:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[UV1]](s64), [[UV3]](s64), 1
; GFX10: [[FMA8:%[0-9]+]]:_(s64) = G_FMA [[FMA6]], [[FMA7]], [[FMA6]]
- ; GFX10: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT10]](s64)
- ; GFX10: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA8]](s64)
- ; GFX10: [[MUL3:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV10]]
- ; GFX10: [[MUL4:%[0-9]+]]:_(s32) = G_MUL [[UV9]], [[UV10]]
- ; GFX10: [[MUL5:%[0-9]+]]:_(s32) = G_MUL [[UV8]], [[UV11]]
- ; GFX10: [[UMULH1:%[0-9]+]]:_(s32) = G_UMULH [[UV8]], [[UV10]]
- ; GFX10: [[ADD2:%[0-9]+]]:_(s32) = G_ADD [[MUL4]], [[MUL5]]
- ; GFX10: [[ADD3:%[0-9]+]]:_(s32) = G_ADD [[ADD2]], [[UMULH1]]
- ; GFX10: [[MV1:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL3]](s32), [[ADD3]](s32)
- ; GFX10: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[MV1]], [[INT10]]
- ; GFX10: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[MV1]](s64), [[INT11]](s1)
+ ; GFX10: [[FMUL1:%[0-9]+]]:_(s64) = G_FMUL [[INT10]], [[FMA8]]
+ ; GFX10: [[FMA9:%[0-9]+]]:_(s64) = G_FMA [[FNEG1]], [[FMUL1]], [[INT10]]
+ ; GFX10: [[INT12:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA9]](s64), [[FMA8]](s64), [[FMUL1]](s64), [[INT11]](s1)
; GFX10: [[INT13:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT12]](s64), [[UV3]](s64), [[UV1]](s64)
; GFX10: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[INT6]](s64), [[INT13]](s64)
; GFX10: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
@@ -2310,24 +2214,16 @@ body: |
; SI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; SI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; SI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; SI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; SI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; SI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; SI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; SI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; SI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; SI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[C]](s64)
- ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
- ; SI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
- ; SI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV5]](s32), [[UV11]]
- ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV7]](s32), [[UV9]]
+ ; SI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[C]](s64)
+ ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
+ ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
+ ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
+ ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV1]](s32), [[UV7]]
+ ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV3]](s32), [[UV5]]
; SI: [[XOR:%[0-9]+]]:_(s1) = G_XOR [[ICMP]], [[ICMP1]]
- ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[XOR]](s1)
+ ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[XOR]](s1)
; SI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; SI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; VI-LABEL: name: test_fdiv_s64_constant_one_rcp
@@ -2341,17 +2237,9 @@ body: |
; VI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; VI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; VI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; VI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; VI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; VI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; VI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; VI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; VI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; VI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; VI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; VI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; VI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; VI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-LABEL: name: test_fdiv_s64_constant_one_rcp
@@ -2365,17 +2253,9 @@ body: |
; GFX9: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX9: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; GFX9: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX9: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX9: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX9: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX9: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX9: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX9: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX9: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX9: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX9: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX9: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; GFX9: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-UNSAFE-LABEL: name: test_fdiv_s64_constant_one_rcp
@@ -2393,17 +2273,9 @@ body: |
; GFX10: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C]]
; GFX10: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; GFX10: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX10: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX10: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX10: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX10: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX10: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX10: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX10: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX10: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX10: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX10: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX10: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; GFX10: $vgpr0_vgpr1 = COPY [[INT6]](s64)
%0:_(s64) = G_FCONSTANT double 1.0
@@ -2437,24 +2309,16 @@ body: |
; SI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C1]]
; SI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; SI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; SI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; SI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; SI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; SI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; SI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; SI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; SI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[C]](s64)
- ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
- ; SI: [[UV8:%[0-9]+]]:_(s32), [[UV9:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
- ; SI: [[UV10:%[0-9]+]]:_(s32), [[UV11:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV5]](s32), [[UV11]]
- ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV7]](s32), [[UV9]]
+ ; SI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; SI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[C]](s64)
+ ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](s64)
+ ; SI: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT]](s64)
+ ; SI: [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
+ ; SI: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV1]](s32), [[UV7]]
+ ; SI: [[ICMP1:%[0-9]+]]:_(s1) = G_ICMP intpred(eq), [[UV3]](s32), [[UV5]]
; SI: [[XOR:%[0-9]+]]:_(s1) = G_XOR [[ICMP]], [[ICMP1]]
- ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[XOR]](s1)
+ ; SI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[XOR]](s1)
; SI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; SI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; VI-LABEL: name: test_fdiv_s64_constant_negative_one_rcp
@@ -2469,17 +2333,9 @@ body: |
; VI: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C1]]
; VI: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; VI: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; VI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; VI: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; VI: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; VI: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; VI: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; VI: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; VI: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; VI: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; VI: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; VI: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; VI: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; VI: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; VI: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-LABEL: name: test_fdiv_s64_constant_negative_one_rcp
@@ -2494,17 +2350,9 @@ body: |
; GFX9: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C1]]
; GFX9: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; GFX9: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX9: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX9: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX9: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX9: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX9: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX9: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX9: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX9: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX9: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX9: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX9: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX9: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; GFX9: $vgpr0_vgpr1 = COPY [[INT6]](s64)
; GFX9-UNSAFE-LABEL: name: test_fdiv_s64_constant_negative_one_rcp
@@ -2524,17 +2372,9 @@ body: |
; GFX10: [[FMA2:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMA1]], [[C1]]
; GFX10: [[INT3:%[0-9]+]]:_(s64), [[INT4:%[0-9]+]]:_(s1) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.scale), [[C]](s64), [[COPY]](s64), 1
; GFX10: [[FMA3:%[0-9]+]]:_(s64) = G_FMA [[FMA1]], [[FMA2]], [[FMA1]]
- ; GFX10: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[INT3]](s64)
- ; GFX10: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[FMA3]](s64)
- ; GFX10: [[MUL:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV2]]
- ; GFX10: [[MUL1:%[0-9]+]]:_(s32) = G_MUL [[UV1]], [[UV2]]
- ; GFX10: [[MUL2:%[0-9]+]]:_(s32) = G_MUL [[UV]], [[UV3]]
- ; GFX10: [[UMULH:%[0-9]+]]:_(s32) = G_UMULH [[UV]], [[UV2]]
- ; GFX10: [[ADD:%[0-9]+]]:_(s32) = G_ADD [[MUL1]], [[MUL2]]
- ; GFX10: [[ADD1:%[0-9]+]]:_(s32) = G_ADD [[ADD]], [[UMULH]]
- ; GFX10: [[MV:%[0-9]+]]:_(s64) = G_MERGE_VALUES [[MUL]](s32), [[ADD1]](s32)
- ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[MV]], [[INT3]]
- ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[MV]](s64), [[INT4]](s1)
+ ; GFX10: [[FMUL:%[0-9]+]]:_(s64) = G_FMUL [[INT3]], [[FMA3]]
+ ; GFX10: [[FMA4:%[0-9]+]]:_(s64) = G_FMA [[FNEG]], [[FMUL]], [[INT3]]
+ ; GFX10: [[INT5:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fmas), [[FMA4]](s64), [[FMA3]](s64), [[FMUL]](s64), [[INT4]](s1)
; GFX10: [[INT6:%[0-9]+]]:_(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.div.fixup), [[INT5]](s64), [[COPY]](s64), [[C]](s64)
; GFX10: $vgpr0_vgpr1 = COPY [[INT6]](s64)
%0:_(s64) = G_FCONSTANT double -1.0
More information about the llvm-commits
mailing list