RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-01-31 20:51:52 +01:00
Code Issues Projects Releases Wiki Activity

AMDGPU: TTI: Make insertelement free.

Browse Source 
We don't want to have a cost to scalarizing operations.

llvm-svn: 264364
This commit is contained in:
Matt Arsenault 2016-03-25 00:14:11 +00:00
parent 57ae7f9db2
commit ffa532b58f
2 changed files with 42 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
View File

@ -99,6 +99,11 @@ int AMDGPUTTIImpl::getVectorInstrCost(unsigned Opcode, Type *ValTy,
unsigned Index) {
switch (Opcode) {
case Instruction::ExtractElement:
case Instruction::InsertElement:
// Extracts are just reads of a subregister, so are free. Inserts are
// considered free because we don't want to have any cost for scalarizing
// operations, and we don't have to copy into a different register class.
// Dynamic indexing isn't free and is best avoided.
return Index == ~0u ? 2 : 0;
default:

37
test/Analysis/CostModel/AMDGPU/insertelement.ll Normal file
View File

@ -0,0 +1,37 @@
; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa < %s | FileCheck %s
; CHECK: 'insertelement_v2i32'
; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i32>
define void @insertelement_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %vaddr) {
%vec = load <2 x i32>, <2 x i32> addrspace(1)* %vaddr
%insert = insertelement <2 x i32> %vec, i32 1, i32 123
store <2 x i32> %insert, <2 x i32> addrspace(1)* %out
ret void
}
; CHECK: 'insertelement_v2i64'
; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i64>
define void @insertelement_v2i64(<2 x i64> addrspace(1)* %out, <2 x i64> addrspace(1)* %vaddr) {
%vec = load <2 x i64>, <2 x i64> addrspace(1)* %vaddr
%insert = insertelement <2 x i64> %vec, i64 1, i64 123
store <2 x i64> %insert, <2 x i64> addrspace(1)* %out
ret void
}
; CHECK: 'insertelement_v2i16'
; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i16>
define void @insertelement_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) {
%vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr
%insert = insertelement <2 x i16> %vec, i16 1, i16 123
store <2 x i16> %insert, <2 x i16> addrspace(1)* %out
ret void
}
; CHECK: 'insertelement_v2i8'
; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i8>
define void @insertelement_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> addrspace(1)* %vaddr) {
%vec = load <2 x i8>, <2 x i8> addrspace(1)* %vaddr
%insert = insertelement <2 x i8> %vec, i8 1, i8 123
store <2 x i8> %insert, <2 x i8> addrspace(1)* %out
ret void
}