AMDGPU: Fold fneg into round instructions

llvm-svn: 293127

3 files changed, 99 insertions, 11 deletions

diff --git a/llvm/test/CodeGen/AMDGPU/fneg-combines.ll b/llvm/test/CodeGen/AMDGPU/fneg-combines.ll
index 040f1eceed5..37115e795e8 100644
--- a/llvm/test/CodeGen/AMDGPU/fneg-combines.ll
+++ b/llvm/test/CodeGen/AMDGPU/fneg-combines.ll
@@ -1327,7 +1327,91 @@ define void @v_fneg_amdgcn_sin_f32(float addrspace(1)* %out, float addrspace(1)*
   %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i64 %tid.ext
   %a = load volatile float, float addrspace(1)* %a.gep
   %sin = call float @llvm.amdgcn.sin.f32(float %a)
-  %fneg = fsub float -0.000000e+00, %sin
+  %fneg = fsub float -0.0, %sin
+  store float %fneg, float addrspace(1)* %out.gep
+  ret void
+}
+
+; --------------------------------------------------------------------------------
+; ftrunc tests
+; --------------------------------------------------------------------------------
+
+; GCN-LABEL: {{^}}v_fneg_trunc_f32:
+; GCN: {{buffer|flat}}_load_dword [[A:v[0-9]+]]
+; GCN: v_trunc_f32_e64 [[RESULT:v[0-9]+]], -[[A]]
+; GCN: buffer_store_dword [[RESULT]]
+define void @v_fneg_trunc_f32(float addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %tid.ext = sext i32 %tid to i64
+  %a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
+  %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i64 %tid.ext
+  %a = load volatile float, float addrspace(1)* %a.gep
+  %trunc = call float @llvm.trunc.f32(float %a)
+  %fneg = fsub float -0.0, %trunc
+  store float %fneg, float addrspace(1)* %out.gep
+  ret void
+}
+
+; --------------------------------------------------------------------------------
+; fround tests
+; --------------------------------------------------------------------------------
+
+; GCN-LABEL: {{^}}v_fneg_round_f32:
+; GCN: {{buffer|flat}}_load_dword [[A:v[0-9]+]]
+; GCN: v_trunc_f32_e32
+; GCN: v_subrev_f32_e32
+; GCN: v_cndmask_b32
+; GCN-NSZ: v_sub_f32_e64 [[RESULT:v[0-9]+]], -v{{[0-9]+}}, v{{[0-9]+}}
+; GCN-SAFE: v_xor_b32_e32 [[RESULT:v[0-9]+]], 0x80000000, v{{[0-9]+}}
+; GCN: buffer_store_dword [[RESULT]]
+define void @v_fneg_round_f32(float addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %tid.ext = sext i32 %tid to i64
+  %a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
+  %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i64 %tid.ext
+  %a = load volatile float, float addrspace(1)* %a.gep
+  %round = call float @llvm.round.f32(float %a)
+  %fneg = fsub float -0.0, %round
+  store float %fneg, float addrspace(1)* %out.gep
+  ret void
+}
+
+; --------------------------------------------------------------------------------
+; rint tests
+; --------------------------------------------------------------------------------
+
+; GCN-LABEL: {{^}}v_fneg_rint_f32:
+; GCN: {{buffer|flat}}_load_dword [[A:v[0-9]+]]
+; GCN: v_rndne_f32_e64 [[RESULT:v[0-9]+]], -[[A]]
+; GCN: buffer_store_dword [[RESULT]]
+define void @v_fneg_rint_f32(float addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %tid.ext = sext i32 %tid to i64
+  %a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
+  %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i64 %tid.ext
+  %a = load volatile float, float addrspace(1)* %a.gep
+  %rint = call float @llvm.rint.f32(float %a)
+  %fneg = fsub float -0.0, %rint
+  store float %fneg, float addrspace(1)* %out.gep
+  ret void
+}
+
+; --------------------------------------------------------------------------------
+; nearbyint tests
+; --------------------------------------------------------------------------------
+
+; GCN-LABEL: {{^}}v_fneg_nearbyint_f32:
+; GCN: {{buffer|flat}}_load_dword [[A:v[0-9]+]]
+; GCN: v_rndne_f32_e64 [[RESULT:v[0-9]+]], -[[A]]
+; GCN: buffer_store_dword [[RESULT]]
+define void @v_fneg_nearbyint_f32(float addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %tid.ext = sext i32 %tid to i64
+  %a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
+  %out.gep = getelementptr inbounds float, float addrspace(1)* %out, i64 %tid.ext
+  %a = load volatile float, float addrspace(1)* %a.gep
+  %nearbyint = call float @llvm.nearbyint.f32(float %a)
+  %fneg = fsub float -0.0, %nearbyint
   store float %fneg, float addrspace(1)* %out.gep
   ret void
 }
@@ -1336,6 +1420,10 @@ declare i32 @llvm.amdgcn.workitem.id.x() #1
 declare float @llvm.fma.f32(float, float, float) #1
 declare float @llvm.fmuladd.f32(float, float, float) #1
 declare float @llvm.sin.f32(float) #1
+declare float @llvm.trunc.f32(float) #1
+declare float @llvm.round.f32(float) #1
+declare float @llvm.rint.f32(float) #1
+declare float @llvm.nearbyint.f32(float) #1
 
 declare float @llvm.amdgcn.sin.f32(float) #1
 declare float @llvm.amdgcn.rcp.f32(float) #1
diff --git a/llvm/test/CodeGen/AMDGPU/frem.ll b/llvm/test/CodeGen/AMDGPU/frem.ll
index 039623c0219..97533c418c9 100644
--- a/llvm/test/CodeGen/AMDGPU/frem.ll
+++ b/llvm/test/CodeGen/AMDGPU/frem.ll
@@ -1,6 +1,6 @@
-; RUN: llc -march=amdgcn -enable-misched < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=FUNC %s
-; RUN: llc -march=amdgcn -mcpu=bonaire -enable-misched < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
-; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -enable-misched < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
+; RUN: llc -march=amdgcn -verify-machineinstrs  < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=FUNC %s
+; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
+; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
 
 ; FUNC-LABEL: {{^}}frem_f32:
 ; GCN-DAG: buffer_load_dword [[X:v[0-9]+]], {{.*$}}
@@ -12,8 +12,8 @@
 ; GCN: v_mul_f32_e32
 ; GCN: v_div_fmas_f32
 ; GCN: v_div_fixup_f32
-; GCN: v_trunc_f32_e32
-; GCN: v_mad_f32
+; GCN: v_trunc_f32_e64 v{{[0-9]+}}, -v{{[0-9]+}}
+; GCN: v_mac_f32_e32
 ; GCN: s_endpgm
 define void @frem_f32(float addrspace(1)* %out, float addrspace(1)* %in1,
                       float addrspace(1)* %in2) #0 {
@@ -28,11 +28,11 @@ define void @frem_f32(float addrspace(1)* %out, float addrspace(1)* %in1,
 ; FUNC-LABEL: {{^}}unsafe_frem_f32:
 ; GCN: buffer_load_dword [[Y:v[0-9]+]], {{.*}} offset:16
 ; GCN: buffer_load_dword [[X:v[0-9]+]], {{.*}}
-; GCN: v_rcp_f32_e32 [[INVY:v[0-9]+]], [[Y]]
+; GCN: v_rcp_f32_e64 [[INVY:v[0-9]+]], -[[Y]]
 ; GCN: v_mul_f32_e32 [[DIV:v[0-9]+]], [[INVY]], [[X]]
 ; GCN: v_trunc_f32_e32 [[TRUNC:v[0-9]+]], [[DIV]]
-; GCN: v_mad_f32 [[RESULT:v[0-9]+]], -[[TRUNC]], [[Y]], [[X]]
-; GCN: buffer_store_dword [[RESULT]]
+; GCN: v_mac_f32_e32 [[X]], [[Y]], [[TRUNC]]
+; GCN: buffer_store_dword [[X]]
 ; GCN: s_endpgm
 define void @unsafe_frem_f32(float addrspace(1)* %out, float addrspace(1)* %in1,
                              float addrspace(1)* %in2) #1 {
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.class.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.class.f16.ll
index 011a0fdbd21..3798c46677f 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.class.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.class.f16.ll
@@ -46,8 +46,8 @@ entry:
 ; GCN-LABEL: {{^}}class_f16_fneg
 ; GCN: s_load_dword s[[SA_F16:[0-9]+]]
 ; GCN: s_load_dword s[[SB_I32:[0-9]+]]
-; VI:  v_trunc_f16_e32 v[[VA_F16:[0-9]+]], s[[SA_F16]]
-; VI:  v_cmp_class_f16_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], -v[[VA_F16]], s[[SB_I32]]
+; VI:  v_trunc_f16_e64 v[[VA_F16:[0-9]+]], -s[[SA_F16]]
+; VI:  v_cmp_class_f16_e64 [[CMP:s\[[0-9]+:[0-9]+\]]], v[[VA_F16]], s[[SB_I32]]
 ; VI:  v_cndmask_b32_e64 v[[VR_I32:[0-9]+]], 0, -1, [[CMP]]
 ; GCN: buffer_store_dword v[[VR_I32]]
 ; GCN: s_endpgm