[PATCH] D124786: [AArch64] Add extra reverse costs.

Mon May 2 10:53:08 PDT 2022

dmgreen created this revision.
dmgreen added reviewers: sdesmalen, samtebbs, jaykang10, labrinea.
Herald added subscribers: hiraditya, kristof.beyls.
Herald added a project: All.
dmgreen requested review of this revision.
Herald added a project: LLVM.

This adds some extra costs for reverse shuffles under AArch64, filling in the i16/f16/i8 gaps in the cost model.


https://reviews.llvm.org/D124786

Files:
  llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
  llvm/test/Analysis/CostModel/AArch64/getIntrinsicInstrCost-vector-reverse.ll
  llvm/test/Analysis/CostModel/AArch64/shuffle-reverse.ll


Index: llvm/test/Analysis/CostModel/AArch64/shuffle-reverse.ll
===================================================================

--- llvm/test/Analysis/CostModel/AArch64/shuffle-reverse.ll
+++ llvm/test/Analysis/CostModel/AArch64/shuffle-reverse.ll
@@ -9,22 +9,22 @@
 define void @reverse() {
 ; CHECK-LABEL: 'reverse'
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i8 = shufflevector <2 x i8> undef, <2 x i8> undef, <2 x i32> <i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v4i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v16i8 = shufflevector <16 x i8> undef, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i8 = shufflevector <4 x i8> undef, <4 x i8> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8 = shufflevector <8 x i8> undef, <8 x i8> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i8 = shufflevector <16 x i8> undef, <16 x i8> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16 = shufflevector <2 x i16> undef, <2 x i16> undef, <2 x i32> <i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16 = shufflevector <4 x i16> undef, <4 x i16> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16i16 = shufflevector <16 x i16> undef, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i16 = shufflevector <8 x i16> undef, <8 x i16> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i16 = shufflevector <16 x i16> undef, <16 x i16> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32 = shufflevector <2 x i32> undef, <2 x i32> undef, <2 x i32> <i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i32 = shufflevector <4 x i32> undef, <4 x i32> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i32 = shufflevector <8 x i32> undef, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64 = shufflevector <2 x i64> undef, <2 x i64> undef, <2 x i32> <i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i64 = shufflevector <4 x i64> undef, <4 x i64> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v2f16 = shufflevector <2 x half> undef, <2 x half> undef, <2 x i32> <i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = shufflevector <2 x half> undef, <2 x half> undef, <2 x i32> <i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = shufflevector <4 x half> undef, <4 x half> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f16 = shufflevector <8 x half> undef, <8 x half> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16f16 = shufflevector <16 x half> undef, <16 x half> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8f16 = shufflevector <8 x half> undef, <8 x half> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16f16 = shufflevector <16 x half> undef, <16 x half> undef, <16 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2f32 = shufflevector <2 x float> undef, <2 x float> undef, <2 x i32> <i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4f32 = shufflevector <4 x float> undef, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8f32 = shufflevector <8 x float> undef, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
Index: llvm/test/Analysis/CostModel/AArch64/getIntrinsicInstrCost-vector-reverse.ll
===================================================================
--- llvm/test/Analysis/CostModel/AArch64/getIntrinsicInstrCost-vector-reverse.ll
+++ llvm/test/Analysis/CostModel/AArch64/getIntrinsicInstrCost-vector-reverse.ll
@@ -5,16 +5,16 @@
 
 define void @vector_reverse() #0{
 ; CHECK-LABEL: 'vector_reverse'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %1 = call <16 x i8> @llvm.experimental.vector.reverse.v16i8(<16 x i8> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %2 = call <32 x i8> @llvm.experimental.vector.reverse.v32i8(<32 x i8> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %3 = call <8 x i16> @llvm.experimental.vector.reverse.v8i16(<8 x i16> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %4 = call <16 x i16> @llvm.experimental.vector.reverse.v16i16(<16 x i16> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %1 = call <16 x i8> @llvm.experimental.vector.reverse.v16i8(<16 x i8> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %2 = call <32 x i8> @llvm.experimental.vector.reverse.v32i8(<32 x i8> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %3 = call <8 x i16> @llvm.experimental.vector.reverse.v8i16(<8 x i16> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %4 = call <16 x i16> @llvm.experimental.vector.reverse.v16i16(<16 x i16> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %5 = call <4 x i32> @llvm.experimental.vector.reverse.v4i32(<4 x i32> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %6 = call <8 x i32> @llvm.experimental.vector.reverse.v8i32(<8 x i32> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %7 = call <2 x i64> @llvm.experimental.vector.reverse.v2i64(<2 x i64> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %8 = call <4 x i64> @llvm.experimental.vector.reverse.v4i64(<4 x i64> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %9 = call <8 x half> @llvm.experimental.vector.reverse.v8f16(<8 x half> undef)
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %10 = call <16 x half> @llvm.experimental.vector.reverse.v16f16(<16 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %9 = call <8 x half> @llvm.experimental.vector.reverse.v8f16(<8 x half> undef)
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %10 = call <16 x half> @llvm.experimental.vector.reverse.v16f16(<16 x half> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %11 = call <4 x float> @llvm.experimental.vector.reverse.v4f32(<4 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %12 = call <8 x float> @llvm.experimental.vector.reverse.v8f32(<8 x float> undef)
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %13 = call <2 x double> @llvm.experimental.vector.reverse.v2f64(<2 x double> undef)
Index: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -2831,6 +2831,12 @@
       { TTI::SK_Reverse, MVT::v2f32, 1 }, // mov.
       { TTI::SK_Reverse, MVT::v4f32, 2 }, // REV64; EXT
       { TTI::SK_Reverse, MVT::v2f64, 1 }, // mov.
+      { TTI::SK_Reverse, MVT::v8f16, 2 }, // REV64; EXT
+      { TTI::SK_Reverse, MVT::v8i16, 2 }, // REV64; EXT
+      { TTI::SK_Reverse, MVT::v16i8, 2 }, // REV64; EXT
+      { TTI::SK_Reverse, MVT::v4f16, 1 }, // REV64
+      { TTI::SK_Reverse, MVT::v4i16, 1 }, // REV64
+      { TTI::SK_Reverse, MVT::v8i8, 1 }, // REV64
       // Broadcast shuffle kinds for scalable vectors
       { TTI::SK_Broadcast, MVT::nxv16i8,  1 },
       { TTI::SK_Broadcast, MVT::nxv8i16,  1 },


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D124786.426454.patch
Type: text/x-patch
Size: 10007 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220502/293ad575/attachment.bin>
