llvm-6502/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll
Tim Northover 7b837d8c75 ARM64: initial backend import
This adds a second implementation of the AArch64 architecture to LLVM,
accessible in parallel via the "arm64" triple. The plan over the
coming weeks & months is to merge the two into a single backend,
during which time thorough code review should naturally occur.

Everything will be easier with the target in-tree though, hence this
commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@205090 91177308-0d34-0410-b5e6-96231b3b80d8
2014-03-29 10:18:08 +00:00

136 lines
5.1 KiB
LLVM

; RUN: opt -S -instcombine < %s | FileCheck %s
define <4 x i32> @mulByZero(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> %x, <4 x i16> zeroinitializer) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> zeroinitializer
}
define <4 x i32> @mulByOne(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> %x, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: %a = sext <4 x i16> %x to <4 x i32>
; CHECK-NEXT: ret <4 x i32> %a
}
define <4 x i32> @constantMul() nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 3, i16 3, i16 3, i16 3>, <4 x i16> <i16 2, i16 2, i16 2, i16 2>) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 6, i32 6, i32 6, i32 6>
}
define <4 x i32> @constantMulS() nounwind readnone ssp {
entry:
%b = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 -1, i16 -1, i16 -1, i16 -1>, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 -1, i32 -1, i32 -1, i32 -1>
}
define <4 x i32> @constantMulU() nounwind readnone ssp {
entry:
%b = tail call <4 x i32> @llvm.arm.neon.vmullu.v4i32(<4 x i16> <i16 -1, i16 -1, i16 -1, i16 -1>, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>
}
define <4 x i32> @complex1(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) nounwind
%b = add <4 x i32> zeroinitializer, %a
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) [[NUW:#[0-9]+]]
; CHECK-NEXT: ret <4 x i32> %a
}
define <4 x i32> @complex2(<4 x i32> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> <i16 3, i16 3, i16 3, i16 3>, <4 x i16> <i16 2, i16 2, i16 2, i16 2>) nounwind
%b = add <4 x i32> %x, %a
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: %b = add <4 x i32> %x, <i32 6, i32 6, i32 6, i32 6>
; CHECK-NEXT: ret <4 x i32> %b
}
declare <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vmullu.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
; ARM64 variants - <rdar://problem/12349617>
define <4 x i32> @mulByZeroARM64(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> %x, <4 x i16> zeroinitializer) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> zeroinitializer
}
define <4 x i32> @mulByOneARM64(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> %x, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: %a = sext <4 x i16> %x to <4 x i32>
; CHECK-NEXT: ret <4 x i32> %a
}
define <4 x i32> @constantMulARM64() nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> <i16 3, i16 3, i16 3, i16 3>, <4 x i16> <i16 2, i16 2, i16 2, i16 2>) nounwind
ret <4 x i32> %a
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 6, i32 6, i32 6, i32 6>
}
define <4 x i32> @constantMulSARM64() nounwind readnone ssp {
entry:
%b = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> <i16 -1, i16 -1, i16 -1, i16 -1>, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 -1, i32 -1, i32 -1, i32 -1>
}
define <4 x i32> @constantMulUARM64() nounwind readnone ssp {
entry:
%b = tail call <4 x i32> @llvm.arm64.neon.umull.v4i32(<4 x i16> <i16 -1, i16 -1, i16 -1, i16 -1>, <4 x i16> <i16 1, i16 1, i16 1, i16 1>) nounwind
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: ret <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>
}
define <4 x i32> @complex1ARM64(<4 x i16> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) nounwind
%b = add <4 x i32> zeroinitializer, %a
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> <i16 2, i16 2, i16 2, i16 2>, <4 x i16> %x) [[NUW:#[0-9]+]]
; CHECK-NEXT: ret <4 x i32> %a
}
define <4 x i32> @complex2ARM64(<4 x i32> %x) nounwind readnone ssp {
entry:
%a = tail call <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16> <i16 3, i16 3, i16 3, i16 3>, <4 x i16> <i16 2, i16 2, i16 2, i16 2>) nounwind
%b = add <4 x i32> %x, %a
ret <4 x i32> %b
; CHECK: entry:
; CHECK-NEXT: %b = add <4 x i32> %x, <i32 6, i32 6, i32 6, i32 6>
; CHECK-NEXT: ret <4 x i32> %b
}
declare <4 x i32> @llvm.arm64.neon.smull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm64.neon.umull.v4i32(<4 x i16>, <4 x i16>) nounwind readnone
; CHECK: attributes #0 = { nounwind readnone ssp }
; CHECK: attributes #1 = { nounwind readnone }
; CHECK: attributes [[NUW]] = { nounwind }