[llvm] r195881 - R600: Expand vector FABS

Tom Stellard thomas.stellard at amd.com
Wed Nov 27 13:23:40 PST 2013


Author: tstellar
Date: Wed Nov 27 15:23:39 2013
New Revision: 195881

URL: http://llvm.org/viewvc/llvm-project?rev=195881&view=rev
Log:
R600: Expand vector FABS

NOTE: This is a candidate for the 3.4 branch.

Modified:
    llvm/trunk/lib/Target/R600/AMDGPUISelLowering.cpp
    llvm/trunk/test/CodeGen/R600/fabs.ll

Modified: llvm/trunk/lib/Target/R600/AMDGPUISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/R600/AMDGPUISelLowering.cpp?rev=195881&r1=195880&r2=195881&view=diff
==============================================================================
--- llvm/trunk/lib/Target/R600/AMDGPUISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/R600/AMDGPUISelLowering.cpp Wed Nov 27 15:23:39 2013
@@ -179,6 +179,7 @@ AMDGPUTargetLowering::AMDGPUTargetLoweri
 
   for (unsigned int x = 0; x < NumFloatTypes; ++x) {
     MVT::SimpleValueType VT = FloatTypes[x];
+    setOperationAction(ISD::FABS, VT, Expand);
     setOperationAction(ISD::FADD, VT, Expand);
     setOperationAction(ISD::FDIV, VT, Expand);
     setOperationAction(ISD::FFLOOR, VT, Expand);

Modified: llvm/trunk/test/CodeGen/R600/fabs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/R600/fabs.ll?rev=195881&r1=195880&r2=195881&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/R600/fabs.ll (original)
+++ llvm/trunk/test/CodeGen/R600/fabs.ll Wed Nov 27 15:23:39 2013
@@ -5,10 +5,10 @@
 ; (fabs (f32 bitcast (i32 a))) => (f32 bitcast (and (i32 a), 0x7FFFFFFF))
 ; unless isFabsFree returns true
 
-; R600-CHECK: @fabs_free
+; R600-CHECK-LABEL: @fabs_free
 ; R600-CHECK-NOT: AND
 ; R600-CHECK: |PV.{{[XYZW]}}|
-; SI-CHECK: @fabs_free
+; SI-CHECK-LABEL: @fabs_free
 ; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
 
 define void @fabs_free(float addrspace(1)* %out, i32 %in) {
@@ -19,4 +19,36 @@ entry:
   ret void
 }
 
+; R600-CHECK-LABEL: @fabs_v2
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; SI-CHECK-LABEL: @fabs_v2
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+define void @fabs_v2(<2 x float> addrspace(1)* %out, <2 x float> %in) {
+entry:
+  %0 = call <2 x float> @llvm.fabs.v2f32(<2 x float> %in)
+  store <2 x float> %0, <2 x float> addrspace(1)* %out
+  ret void
+}
+
+; R600-CHECK-LABEL: @fabs_v4
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; R600-CHECK: |{{(PV|T[0-9])\.[XYZW]}}|
+; SI-CHECK-LABEL: @fabs_v4
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+; SI-CHECK: V_ADD_F32_e64 v{{[0-9]}}, s{{[0-9]}}, 0, 1, 0, 0, 0
+define void @fabs_v4(<4 x float> addrspace(1)* %out, <4 x float> %in) {
+entry:
+  %0 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %in)
+  store <4 x float> %0, <4 x float> addrspace(1)* %out
+  ret void
+}
+
 declare float @fabs(float ) readnone
+declare <2 x float> @llvm.fabs.v2f32(<2 x float> ) readnone
+declare <4 x float> @llvm.fabs.v4f32(<4 x float> ) readnone





More information about the llvm-commits mailing list