forked from OSchip/llvm-project
50 lines
1.6 KiB
LLVM
50 lines
1.6 KiB
LLVM
; RUN: llc -march=hexagon < %s | FileCheck %s
|
|
; CHECK: r{{[0-9]+}}:{{[0-9]+}} ^= pmpyw(r{{[0-9]+}},r{{[0-9]+}})
|
|
|
|
; Function Attrs: nounwind
|
|
define i32 @f0(i32 %a0, i32 %a1, i32 %a2, i32 %a3) #0 {
|
|
b0:
|
|
%v0 = alloca i32, align 4
|
|
%v1 = alloca i32, align 4
|
|
%v2 = alloca i32, align 4
|
|
%v3 = alloca i32, align 4
|
|
%v4 = alloca i64, align 8
|
|
%v5 = alloca i64, align 8
|
|
store i32 %a0, i32* %v0, align 4
|
|
store i32 %a1, i32* %v1, align 4
|
|
store i32 %a2, i32* %v2, align 4
|
|
store i32 %a3, i32* %v3, align 4
|
|
%v6 = load i32, i32* %v0, align 4
|
|
%v7 = load i32, i32* %v1, align 4
|
|
%v8 = call i64 @llvm.hexagon.M4.pmpyw(i32 %v6, i32 %v7)
|
|
store i64 %v8, i64* %v5, align 8
|
|
%v9 = load i64, i64* %v5, align 8
|
|
store i64 %v9, i64* %v4, align 8
|
|
%v10 = load i64, i64* %v5, align 8
|
|
%v11 = load i32, i32* %v3, align 4
|
|
%v12 = load i64, i64* %v5, align 8
|
|
%v13 = lshr i64 %v12, 32
|
|
%v14 = trunc i64 %v13 to i32
|
|
%v15 = call i64 @llvm.hexagon.M4.pmpyw.acc(i64 %v10, i32 %v11, i32 %v14)
|
|
store i64 %v15, i64* %v5, align 8
|
|
%v16 = load i64, i64* %v4, align 8
|
|
%v17 = load i64, i64* %v5, align 8
|
|
%v18 = lshr i64 %v17, 32
|
|
%v19 = trunc i64 %v18 to i32
|
|
%v20 = load i32, i32* %v2, align 4
|
|
%v21 = call i64 @llvm.hexagon.M4.pmpyw.acc(i64 %v16, i32 %v19, i32 %v20)
|
|
store i64 %v21, i64* %v4, align 8
|
|
%v22 = load i64, i64* %v4, align 8
|
|
%v23 = trunc i64 %v22 to i32
|
|
ret i32 %v23
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @llvm.hexagon.M4.pmpyw(i32, i32) #1
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @llvm.hexagon.M4.pmpyw.acc(i64, i32, i32) #1
|
|
|
|
attributes #0 = { nounwind }
|
|
attributes #1 = { nounwind readnone }
|