1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
|
; 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(ptr %a0, ptr %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, ptr %a0, i32 %v4
%v7 = load <128 x i8>, ptr %v5, 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, ptr %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" }
|
