From 63d787e5d41e27db7369fb61ef8e07d4e5a45f40 Mon Sep 17 00:00:00 2001 From: Sander de Smalen Date: Fri, 12 Feb 2021 07:40:52 +0000 Subject: [PATCH] [CostModel] An extending load to illegal type is not free. COST(zext (<4 x i32> load(...) to <4 x i64>)) != 0 when <4 x i64> is an illegal result type that requires splitting of the operation. Reviewed By: dmgreen Differential Revision: https://reviews.llvm.org/D96250 --- llvm/include/llvm/CodeGen/BasicTTIImpl.h | 5 +++-- llvm/test/Analysis/CostModel/AArch64/cast.ll | 6 ++++++ llvm/test/Analysis/CostModel/ARM/cast_ldst.ll | 12 ++++++------ 3 files changed, 15 insertions(+), 8 deletions(-) diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h index 75e897c..6596da1 100644 --- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h +++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h @@ -785,13 +785,14 @@ public: return 0; // If this is a zext/sext of a load, return 0 if the corresponding - // extending load exists on target. + // extending load exists on target and the result type is legal. if (CCH == TTI::CastContextHint::Normal) { EVT ExtVT = EVT::getEVT(Dst); EVT LoadVT = EVT::getEVT(Src); unsigned LType = ((Opcode == Instruction::ZExt) ? ISD::ZEXTLOAD : ISD::SEXTLOAD); - if (TLI->isLoadExtLegal(LType, ExtVT, LoadVT)) + if (DstLT.first == SrcLT.first && + TLI->isLoadExtLegal(LType, ExtVT, LoadVT)) return 0; } break; diff --git a/llvm/test/Analysis/CostModel/AArch64/cast.ll b/llvm/test/Analysis/CostModel/AArch64/cast.ll index 2255f84..d55c2bc 100644 --- a/llvm/test/Analysis/CostModel/AArch64/cast.ll +++ b/llvm/test/Analysis/CostModel/AArch64/cast.ll @@ -707,6 +707,7 @@ define i32 @load_extends() { ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %loadv2i16 = load <2 x i16>, <2 x i16>* undef ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %loadv4i16 = load <4 x i16>, <4 x i16>* undef ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %loadv2i32 = load <2 x i32>, <2 x i32>* undef +; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %loadv4i32 = load <4 x i32>, <4 x i32>* undef ; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %r0 = sext i8 %loadi8 to i16 ; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %r1 = zext i8 %loadi8 to i16 ; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %r2 = sext i8 %loadi8 to i32 @@ -731,6 +732,8 @@ define i32 @load_extends() { ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v9 = zext <2 x i16> %loadv2i16 to <2 x i64> ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v10 = sext <2 x i32> %loadv2i32 to <2 x i64> ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v11 = zext <2 x i32> %loadv2i32 to <2 x i64> +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v12 = sext <4 x i32> %loadv4i32 to <4 x i64> +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v13 = zext <4 x i32> %loadv4i32 to <4 x i64> ; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef ; %loadi8 = load i8, i8* undef @@ -742,6 +745,7 @@ define i32 @load_extends() { %loadv2i16 = load <2 x i16>, <2 x i16>* undef %loadv4i16 = load <4 x i16>, <4 x i16>* undef %loadv2i32 = load <2 x i32>, <2 x i32>* undef + %loadv4i32 = load <4 x i32>, <4 x i32>* undef %r0 = sext i8 %loadi8 to i16 %r1 = zext i8 %loadi8 to i16 @@ -768,6 +772,8 @@ define i32 @load_extends() { %v9 = zext <2 x i16> %loadv2i16 to <2 x i64> %v10 = sext <2 x i32> %loadv2i32 to <2 x i64> %v11 = zext <2 x i32> %loadv2i32 to <2 x i64> + %v12 = sext <4 x i32> %loadv4i32 to <4 x i64> + %v13 = zext <4 x i32> %loadv4i32 to <4 x i64> ret i32 undef } diff --git a/llvm/test/Analysis/CostModel/ARM/cast_ldst.ll b/llvm/test/Analysis/CostModel/ARM/cast_ldst.ll index fd4e1a6..4406e27 100644 --- a/llvm/test/Analysis/CostModel/ARM/cast_ldst.ll +++ b/llvm/test/Analysis/CostModel/ARM/cast_ldst.ll @@ -45,8 +45,8 @@ define i32 @load_extends() { ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4816u = zext <4 x i8> %loadv4i8 to <4 x i16> ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4832s = sext <4 x i8> %loadv4i8 to <4 x i32> ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4832u = zext <4 x i8> %loadv4i8 to <4 x i32> -; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> -; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816s = sext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816u = zext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-NEON-RECIP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8832s = sext <8 x i8> %loadv8i8 to <8 x i32> @@ -313,8 +313,8 @@ define i32 @load_extends() { ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4816u = zext <4 x i8> %loadv4i8 to <4 x i16> ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4832s = sext <4 x i8> %loadv4i8 to <4 x i32> ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4832u = zext <4 x i8> %loadv4i8 to <4 x i32> -; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> -; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816s = sext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816u = zext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-V8R-RECIP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8832s = sext <8 x i8> %loadv8i8 to <8 x i32> @@ -581,8 +581,8 @@ define i32 @load_extends() { ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4816u = zext <4 x i8> %loadv4i8 to <4 x i16> ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4832s = sext <4 x i8> %loadv4i8 to <4 x i32> ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4832u = zext <4 x i8> %loadv4i8 to <4 x i32> -; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> -; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4864s = sext <4 x i8> %loadv4i8 to <4 x i64> +; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4864u = zext <4 x i8> %loadv4i8 to <4 x i64> ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816s = sext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8816u = zext <8 x i8> %loadv8i8 to <8 x i16> ; CHECK-V8R-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8832s = sext <8 x i8> %loadv8i8 to <8 x i32> -- 2.7.4