mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 18:54:02 +01:00
db3c93e416
This will remove suboptimal branching from the generated ll/sc loops. The extra simplification pass affects a lot of testcases, which have been modified to accommodate this change: either by modifying the test to become immune to the CFG simplification, or (less preferablt) by adding option -hexagon-initial-cfg-clenaup=0. llvm-svn: 338774
47 lines
2.4 KiB
LLVM
47 lines
2.4 KiB
LLVM
; RUN: llc -march=hexagon -hexagon-initial-cfg-cleanup=0 < %s | FileCheck %s
|
|
; CHECK: = vmem(r{{[0-9]+}}++#1)
|
|
|
|
target triple = "hexagon-unknown--elf"
|
|
|
|
declare <32 x i32> @llvm.hexagon.V6.hi.128B(<64 x i32>) #0
|
|
declare <64 x i32> @llvm.hexagon.V6.vcombine.128B(<32 x i32>, <32 x i32>) #0
|
|
declare <64 x i32> @llvm.hexagon.V6.vzb.128B(<32 x i32>) #0
|
|
declare <32 x i32> @llvm.hexagon.V6.vsathub.128B(<32 x i32>, <32 x i32>) #0
|
|
declare <64 x i32> @llvm.hexagon.V6.vaddh.dv.128B(<64 x i32>, <64 x i32>) #0
|
|
declare <64 x i32> @llvm.hexagon.V6.vadduhsat.dv.128B(<64 x i32>, <64 x i32>) #0
|
|
declare <32 x i32> @llvm.hexagon.V6.vabsdiffuh.128B(<32 x i32>, <32 x i32>) #0
|
|
|
|
define void @f0(i8* %a0, <32 x i32>* %a1) #1 {
|
|
b0:
|
|
br label %b1
|
|
|
|
b1: ; preds = %b2, %b1
|
|
%v0 = phi <128 x i8> [ %v7, %b1 ], [ undef, %b0 ]
|
|
%v1 = phi i32 [ %v19, %b1 ], [ 0, %b0 ]
|
|
%v2 = add nsw i32 %v1, undef
|
|
%v3 = shl i32 %v2, 7
|
|
%v4 = add nsw i32 %v3, 128
|
|
%v5 = getelementptr inbounds i8, i8* %a0, i32 %v4
|
|
%v6 = bitcast i8* %v5 to <128 x i8>*
|
|
%v7 = load <128 x i8>, <128 x i8>* %v6, align 128
|
|
%v8 = bitcast <128 x i8> %v0 to <32 x i32>
|
|
%v9 = tail call <32 x i32> @llvm.hexagon.V6.valignbi.128B(<32 x i32> undef, <32 x i32> %v8, i32 1)
|
|
%v10 = tail call <64 x i32> @llvm.hexagon.V6.vzb.128B(<32 x i32> %v9) #1
|
|
%v11 = tail call <64 x i32> @llvm.hexagon.V6.vadduhsat.dv.128B(<64 x i32> undef, <64 x i32> %v10) #1
|
|
%v12 = tail call <64 x i32> @llvm.hexagon.V6.vadduhsat.dv.128B(<64 x i32> %v11, <64 x i32> undef) #1
|
|
%v13 = tail call <32 x i32> @llvm.hexagon.V6.hi.128B(<64 x i32> %v12)
|
|
%v14 = tail call <32 x i32> @llvm.hexagon.V6.vabsdiffuh.128B(<32 x i32> undef, <32 x i32> %v13) #1
|
|
%v15 = tail call <64 x i32> @llvm.hexagon.V6.vcombine.128B(<32 x i32> %v14, <32 x i32> undef)
|
|
%v16 = tail call <64 x i32> @llvm.hexagon.V6.vaddh.dv.128B(<64 x i32> undef, <64 x i32> %v15) #1
|
|
%v17 = tail call <32 x i32> @llvm.hexagon.V6.hi.128B(<64 x i32> %v16) #1
|
|
%v18 = tail call <32 x i32> @llvm.hexagon.V6.vsathub.128B(<32 x i32> %v17, <32 x i32> undef) #1
|
|
store <32 x i32> %v18, <32 x i32>* %a1, align 128
|
|
%v19 = add nuw nsw i32 %v1, 1
|
|
br label %b1
|
|
}
|
|
|
|
declare <32 x i32> @llvm.hexagon.V6.valignbi.128B(<32 x i32>, <32 x i32>, i32) #0
|
|
|
|
attributes #0 = { nounwind readnone }
|
|
attributes #1 = { nounwind "target-cpu"="hexagonv60" "target-features"="+hvxv60,+hvx-length128b" }
|