[RISCV][TTI] Discount slide cost if ri.vinsert/ri.vextract are available #142036
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
If we have the ri.vinsert/vextract instructions from xrivosvisni, we can do an element insert or extract without needing a vslide or a vector temporary register. Adjust the TTI cost to reflect this.
|
@llvm/pr-subscribers-backend-risc-v Author: Philip Reames (preames) ChangesIf we have the ri.vinsert/vextract instructions from xrivosvisni, we can do an element insert or extract without needing a vslide or a vector temporary register. Adjust the TTI cost to reflect this. Patch is 84.26 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/142036.diff 3 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 860d787111ce4..ff822dec232a9 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -2297,9 +2297,12 @@ InstructionCost RISCVTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
Index = Index % M1Max;
}
- // We could extract/insert the first element without vslidedown/vslideup.
if (Index == 0)
+ // We can extract/insert the first element without vslidedown/vslideup.
SlideCost = 0;
+ else if (ST->hasVendorXRivosVisni() && isUInt<5>(Index) &&
+ Val->getScalarType()->isIntegerTy())
+ SlideCost = 0; // With ri.vinsert/ri.vextract there is no slide needed
else if (Opcode == Instruction::InsertElement)
SlideCost = 1; // With a constant index, we do not need to use addi.
}
diff --git a/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll b/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
index 2859d9fa15168..bca1624d279e7 100644
--- a/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
@@ -1,6 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+v,+f,+d,+zfh,+zvfh < %s | FileCheck %s --check-prefixes=RV32V
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh < %s | FileCheck %s --check-prefixes=RV64V
+; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh,+experimental-xrivosvisni < %s | FileCheck %s --check-prefixes=VISNI
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+zve64x < %s | FileCheck %s --check-prefixes=RV32ZVE64X
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+zve64x < %s | FileCheck %s --check-prefixes=RV64ZVE64X
; Check that we don't crash querying costs when vectors are not enabled.
@@ -337,6 +338,171 @@ define void @extractelement_int(i32 %x) {
; RV64V-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %nxv16i64_x = extractelement <vscale x 16 x i64> undef, i32 %x
; RV64V-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
+; VISNI-LABEL: 'extractelement_int'
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2i1_0 = extractelement <2 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = extractelement <4 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8i1_0 = extractelement <8 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16i1_0 = extractelement <16 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32i1_0 = extractelement <32 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv2i1_0 = extractelement <vscale x 2 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv4i1_0 = extractelement <vscale x 4 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv8i1_0 = extractelement <vscale x 8 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %nxv16i1_0 = extractelement <vscale x 16 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %nxv32i1_0 = extractelement <vscale x 32 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = extractelement <2 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = extractelement <4 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = extractelement <8 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = extractelement <16 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = extractelement <32 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = extractelement <64 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v128i8_0 = extractelement <128 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i8_0 = extractelement <vscale x 2 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i8_0 = extractelement <vscale x 4 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8_0 = extractelement <vscale x 8 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8_0 = extractelement <vscale x 16 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i8_0 = extractelement <vscale x 32 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i8_0 = extractelement <vscale x 64 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv128i8_0 = extractelement <vscale x 128 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = extractelement <2 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = extractelement <4 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = extractelement <8 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = extractelement <16 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = extractelement <32 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i16_0 = extractelement <64 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i16_0 = extractelement <vscale x 2 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16_0 = extractelement <vscale x 4 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16_0 = extractelement <vscale x 8 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i16_0 = extractelement <vscale x 16 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i16_0 = extractelement <vscale x 32 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i16_0 = extractelement <vscale x 64 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = extractelement <2 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = extractelement <4 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = extractelement <8 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = extractelement <16 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i32_0 = extractelement <32 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32_0 = extractelement <vscale x 2 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32_0 = extractelement <vscale x 4 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i32_0 = extractelement <vscale x 8 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i32_0 = extractelement <vscale x 16 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i32_0 = extractelement <vscale x 32 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = extractelement <2 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = extractelement <4 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = extractelement <8 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i64_0 = extractelement <16 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64_0 = extractelement <vscale x 2 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i64_0 = extractelement <vscale x 4 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i64_0 = extractelement <vscale x 8 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i64_0 = extractelement <vscale x 16 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2i1_1 = extractelement <2 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v4i1_1 = extractelement <4 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8i1_1 = extractelement <8 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16i1_1 = extractelement <16 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32i1_1 = extractelement <32 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv2i1_1 = extractelement <vscale x 2 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv4i1_1 = extractelement <vscale x 4 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv8i1_1 = extractelement <vscale x 8 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %nxv16i1_1 = extractelement <vscale x 16 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %nxv32i1_1 = extractelement <vscale x 32 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i8_1 = extractelement <2 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i8_1 = extractelement <4 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8_1 = extractelement <8 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8_1 = extractelement <16 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i8_1 = extractelement <32 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i8_1 = extractelement <64 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v128i8_1 = extractelement <128 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i8_1 = extractelement <vscale x 2 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i8_1 = extractelement <vscale x 4 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8_1 = extractelement <vscale x 8 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8_1 = extractelement <vscale x 16 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i8_1 = extractelement <vscale x 32 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i8_1 = extractelement <vscale x 64 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv128i8_1 = extractelement <vscale x 128 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = extractelement <2 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16_1 = extractelement <4 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16_1 = extractelement <8 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i16_1 = extractelement <16 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i16_1 = extractelement <32 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i16_1 = extractelement <64 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i16_1 = extractelement <vscale x 2 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16_1 = extractelement <vscale x 4 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16_1 = extractelement <vscale x 8 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i16_1 = extractelement <vscale x 16 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i16_1 = extractelement <vscale x 32 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i16_1 = extractelement <vscale x 64 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = extractelement <2 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32_1 = extractelement <4 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i32_1 = extractelement <8 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i32_1 = extractelement <16 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i32_1 = extractelement <32 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32_1 = extractelement <vscale x 2 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32_1 = extractelement <vscale x 4 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i32_1 = extractelement <vscale x 8 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i32_1 = extractelement <vscale x 16 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i32_1 = extractelement <vscale x 32 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = extractelement <2 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i64_1 = extractelement <4 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i64_1 = extractelement <8 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i64_1 = extractelement <16 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64_1 = extractelement <vscale x 2 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i64_1 = extractelement <vscale x 4 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i64_1 = extractelement <vscale x 8 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i64_1 = extractelement <vscale x 16 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2i1_x = extractelement <2 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4i1_x = extractelement <4 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8i1_x = extractelement <8 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v16i1_x = extractelement <16 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v32i1_x = extractelement <32 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv2i1_x = extractelement <vscale x 2 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv4i1_x = extractelement <vscale x 4 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv8i1_x = extractelement <vscale x 8 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %nxv16i1_x = extractelement <vscale x 16 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %nxv32i1_x = extractelement <vscale x 32 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2i8_x = extractelement <2 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i8_x = extractelement <4 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i8_x = extractelement <8 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i8_x = extractelement <16 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i8_x = extractelement <32 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v64i8_x = extractelement <64 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v128i8_x = extractelement <128 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv2i8_x = extractelement <vscale x 2 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i8_x = extractelement <vscale x 4 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8i8_x = extractelement <vscale x 8 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16i8_x = extractelement <vscale x 16 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv32i8_x = extractelement <vscale x 32 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv64i8_x = extractelement <vscale x 64 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %nxv128i8_x = extractelement <vscale x 128 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2i16_x = extractelement <2 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i16_x = extractelement <4 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i16_x = extractelement <8 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i16_x = extractelement <16 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i16_x = extractelement <32 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v64i16_x = extractelement <64 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv2i16_x = extractelement <vscale x 2 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i16_x = extractelement <vscale x 4 x i16> undef, i32 %...
[truncated]
|
|
@llvm/pr-subscribers-llvm-analysis Author: Philip Reames (preames) ChangesIf we have the ri.vinsert/vextract instructions from xrivosvisni, we can do an element insert or extract without needing a vslide or a vector temporary register. Adjust the TTI cost to reflect this. Patch is 84.26 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/142036.diff 3 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 860d787111ce4..ff822dec232a9 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -2297,9 +2297,12 @@ InstructionCost RISCVTTIImpl::getVectorInstrCost(unsigned Opcode, Type *Val,
Index = Index % M1Max;
}
- // We could extract/insert the first element without vslidedown/vslideup.
if (Index == 0)
+ // We can extract/insert the first element without vslidedown/vslideup.
SlideCost = 0;
+ else if (ST->hasVendorXRivosVisni() && isUInt<5>(Index) &&
+ Val->getScalarType()->isIntegerTy())
+ SlideCost = 0; // With ri.vinsert/ri.vextract there is no slide needed
else if (Opcode == Instruction::InsertElement)
SlideCost = 1; // With a constant index, we do not need to use addi.
}
diff --git a/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll b/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
index 2859d9fa15168..bca1624d279e7 100644
--- a/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll
@@ -1,6 +1,7 @@
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+v,+f,+d,+zfh,+zvfh < %s | FileCheck %s --check-prefixes=RV32V
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh < %s | FileCheck %s --check-prefixes=RV64V
+; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh,+experimental-xrivosvisni < %s | FileCheck %s --check-prefixes=VISNI
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+zve64x < %s | FileCheck %s --check-prefixes=RV32ZVE64X
; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+zve64x < %s | FileCheck %s --check-prefixes=RV64ZVE64X
; Check that we don't crash querying costs when vectors are not enabled.
@@ -337,6 +338,171 @@ define void @extractelement_int(i32 %x) {
; RV64V-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %nxv16i64_x = extractelement <vscale x 16 x i64> undef, i32 %x
; RV64V-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
+; VISNI-LABEL: 'extractelement_int'
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2i1_0 = extractelement <2 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v4i1_0 = extractelement <4 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8i1_0 = extractelement <8 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16i1_0 = extractelement <16 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32i1_0 = extractelement <32 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv2i1_0 = extractelement <vscale x 2 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv4i1_0 = extractelement <vscale x 4 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv8i1_0 = extractelement <vscale x 8 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %nxv16i1_0 = extractelement <vscale x 16 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %nxv32i1_0 = extractelement <vscale x 32 x i1> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i8_0 = extractelement <2 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i8_0 = extractelement <4 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = extractelement <8 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = extractelement <16 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i8_0 = extractelement <32 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i8_0 = extractelement <64 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v128i8_0 = extractelement <128 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i8_0 = extractelement <vscale x 2 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i8_0 = extractelement <vscale x 4 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8_0 = extractelement <vscale x 8 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8_0 = extractelement <vscale x 16 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i8_0 = extractelement <vscale x 32 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i8_0 = extractelement <vscale x 64 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv128i8_0 = extractelement <vscale x 128 x i8> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i16_0 = extractelement <2 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = extractelement <4 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = extractelement <8 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i16_0 = extractelement <16 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i16_0 = extractelement <32 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i16_0 = extractelement <64 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i16_0 = extractelement <vscale x 2 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16_0 = extractelement <vscale x 4 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16_0 = extractelement <vscale x 8 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i16_0 = extractelement <vscale x 16 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i16_0 = extractelement <vscale x 32 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i16_0 = extractelement <vscale x 64 x i16> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = extractelement <2 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = extractelement <4 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i32_0 = extractelement <8 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i32_0 = extractelement <16 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i32_0 = extractelement <32 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32_0 = extractelement <vscale x 2 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32_0 = extractelement <vscale x 4 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i32_0 = extractelement <vscale x 8 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i32_0 = extractelement <vscale x 16 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i32_0 = extractelement <vscale x 32 x i32> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = extractelement <2 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i64_0 = extractelement <4 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i64_0 = extractelement <8 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i64_0 = extractelement <16 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64_0 = extractelement <vscale x 2 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i64_0 = extractelement <vscale x 4 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i64_0 = extractelement <vscale x 8 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i64_0 = extractelement <vscale x 16 x i64> undef, i32 0
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2i1_1 = extractelement <2 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v4i1_1 = extractelement <4 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v8i1_1 = extractelement <8 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16i1_1 = extractelement <16 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32i1_1 = extractelement <32 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv2i1_1 = extractelement <vscale x 2 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv4i1_1 = extractelement <vscale x 4 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %nxv8i1_1 = extractelement <vscale x 8 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %nxv16i1_1 = extractelement <vscale x 16 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %nxv32i1_1 = extractelement <vscale x 32 x i1> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i8_1 = extractelement <2 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i8_1 = extractelement <4 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i8_1 = extractelement <8 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i8_1 = extractelement <16 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i8_1 = extractelement <32 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i8_1 = extractelement <64 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v128i8_1 = extractelement <128 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i8_1 = extractelement <vscale x 2 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i8_1 = extractelement <vscale x 4 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i8_1 = extractelement <vscale x 8 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i8_1 = extractelement <vscale x 16 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i8_1 = extractelement <vscale x 32 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i8_1 = extractelement <vscale x 64 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv128i8_1 = extractelement <vscale x 128 x i8> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i16_1 = extractelement <2 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i16_1 = extractelement <4 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i16_1 = extractelement <8 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i16_1 = extractelement <16 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i16_1 = extractelement <32 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v64i16_1 = extractelement <64 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i16_1 = extractelement <vscale x 2 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i16_1 = extractelement <vscale x 4 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i16_1 = extractelement <vscale x 8 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i16_1 = extractelement <vscale x 16 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i16_1 = extractelement <vscale x 32 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv64i16_1 = extractelement <vscale x 64 x i16> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = extractelement <2 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i32_1 = extractelement <4 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i32_1 = extractelement <8 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i32_1 = extractelement <16 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v32i32_1 = extractelement <32 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i32_1 = extractelement <vscale x 2 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i32_1 = extractelement <vscale x 4 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i32_1 = extractelement <vscale x 8 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i32_1 = extractelement <vscale x 16 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv32i32_1 = extractelement <vscale x 32 x i32> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = extractelement <2 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4i64_1 = extractelement <4 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8i64_1 = extractelement <8 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16i64_1 = extractelement <16 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv2i64_1 = extractelement <vscale x 2 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv4i64_1 = extractelement <vscale x 4 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv8i64_1 = extractelement <vscale x 8 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %nxv16i64_1 = extractelement <vscale x 16 x i64> undef, i32 1
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2i1_x = extractelement <2 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4i1_x = extractelement <4 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8i1_x = extractelement <8 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v16i1_x = extractelement <16 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v32i1_x = extractelement <32 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv2i1_x = extractelement <vscale x 2 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv4i1_x = extractelement <vscale x 4 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %nxv8i1_x = extractelement <vscale x 8 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %nxv16i1_x = extractelement <vscale x 16 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %nxv32i1_x = extractelement <vscale x 32 x i1> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2i8_x = extractelement <2 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i8_x = extractelement <4 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i8_x = extractelement <8 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i8_x = extractelement <16 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i8_x = extractelement <32 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v64i8_x = extractelement <64 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v128i8_x = extractelement <128 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv2i8_x = extractelement <vscale x 2 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i8_x = extractelement <vscale x 4 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv8i8_x = extractelement <vscale x 8 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv16i8_x = extractelement <vscale x 16 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv32i8_x = extractelement <vscale x 32 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv64i8_x = extractelement <vscale x 64 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %nxv128i8_x = extractelement <vscale x 128 x i8> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2i16_x = extractelement <2 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4i16_x = extractelement <4 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8i16_x = extractelement <8 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16i16_x = extractelement <16 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v32i16_x = extractelement <32 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v64i16_x = extractelement <64 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv2i16_x = extractelement <vscale x 2 x i16> undef, i32 %x
+; VISNI-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %nxv4i16_x = extractelement <vscale x 4 x i16> undef, i32 %...
[truncated]
|
You can test this locally with the following command:git diff -U0 --pickaxe-regex -S '([^a-zA-Z0-9#_-]undef[^a-zA-Z0-9_-]|UndefValue::get)' 'HEAD~1' HEAD llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp llvm/test/Analysis/CostModel/RISCV/rvv-extractelement.ll llvm/test/Analysis/CostModel/RISCV/rvv-insertelement.llThe following files introduce new uses of undef:
Undef is now deprecated and should only be used in the rare cases where no replacement is possible. For example, a load of uninitialized memory yields In tests, avoid using For example, this is considered a bad practice: define void @fn() {
...
br i1 undef, ...
}Please use the following instead: define void @fn(i1 %cond) {
...
br i1 %cond, ...
}Please refer to the Undefined Behavior Manual for more information. |
lukel97
approved these changes
May 30, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
If we have the ri.vinsert/vextract instructions from xrivosvisni, we can do an element insert or extract without needing a vslide or a vector temporary register. Adjust the TTI cost to reflect this.