1
0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-29 10:32:47 +00:00
llvm-6502/test/CodeGen/ARM64/simd-scalar-to-vector.ll
Tim Northover 3844cadc9a ARM64: make sure first argument to INSERT_SUBVECTOR has right type.
Again, coalescing and other optimisations swiftly made the MachineInstrs
consistent again, but when compiled at -O0 a bad INSERT_SUBREGISTER was
produced.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205423 91177308-0d34-0410-b5e6-96231b3b80d8
2014-04-02 14:38:58 +00:00

23 lines
917 B
LLVM

; RUN: llc < %s -march=arm64 -arm64-neon-syntax=apple | FileCheck %s
; RUN: llc < %s -march=arm64 -arm64-neon-syntax=apple -O0 | FileCheck %s --check-prefix=CHECK-FAST
define <16 x i8> @foo(<16 x i8> %a) nounwind optsize readnone ssp {
; CHECK: uaddlv.16b h0, v0
; CHECK: rshrn.8b v0, v0, #4
; CHECK: dup.16b v0, v0[0]
; CHECK: ret
; CHECK-FAST: uaddlv.16b
; CHECK-FAST: rshrn.8b
; CHECK-FAST: dup.16b
%tmp = tail call i32 @llvm.arm64.neon.uaddlv.i32.v16i8(<16 x i8> %a) nounwind
%tmp1 = trunc i32 %tmp to i16
%tmp2 = insertelement <8 x i16> undef, i16 %tmp1, i32 0
%tmp3 = tail call <8 x i8> @llvm.arm64.neon.rshrn.v8i8(<8 x i16> %tmp2, i32 4)
%tmp4 = shufflevector <8 x i8> %tmp3, <8 x i8> undef, <16 x i32> zeroinitializer
ret <16 x i8> %tmp4
}
declare <8 x i8> @llvm.arm64.neon.rshrn.v8i8(<8 x i16>, i32) nounwind readnone
declare i32 @llvm.arm64.neon.uaddlv.i32.v16i8(<16 x i8>) nounwind readnone