llvm/test/CodeGen/ARM/vector-extend-narrow.ll
Tim Northover 37abe8df4a Use VLD1 in NEON extenting-load patterns instead of VLDR.
On some cores it's a bad idea for performance to mix VFP and NEON instructions
and since these patterns are NEON anyway, the NEON load should be used.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@155630 91177308-0d34-0410-b5e6-96231b3b80d8
2012-04-26 08:46:29 +00:00

65 lines
1.5 KiB
LLVM

; RUN: llc -mtriple armv7 %s -o - | FileCheck %s
; CHECK: f:
define float @f(<4 x i16>* nocapture %in) {
; CHECK: vldr
; CHECK: vmovl.u16
%1 = load <4 x i16>* %in
; CHECK: vcvt.f32.u32
%2 = uitofp <4 x i16> %1 to <4 x float>
%3 = extractelement <4 x float> %2, i32 0
%4 = extractelement <4 x float> %2, i32 1
%5 = extractelement <4 x float> %2, i32 2
; CHECK: vadd.f32
%6 = fadd float %3, %4
%7 = fadd float %6, %5
ret float %7
}
; CHECK: g:
define float @g(<4 x i8>* nocapture %in) {
; Note: vld1 here is reasonably important. Mixing VFP and NEON
; instructions is bad on some cores
; CHECK: vld1
; CHECK: vmovl.u8
; CHECK: vmovl.u16
%1 = load <4 x i8>* %in
; CHECK: vcvt.f32.u32
%2 = uitofp <4 x i8> %1 to <4 x float>
%3 = extractelement <4 x float> %2, i32 0
%4 = extractelement <4 x float> %2, i32 1
%5 = extractelement <4 x float> %2, i32 2
; CHECK: vadd.f32
%6 = fadd float %3, %4
%7 = fadd float %6, %5
ret float %7
}
; CHECK: h:
define <4 x i8> @h(<4 x float> %v) {
; CHECK: vcvt.{{[us]}}32.f32
; CHECK: vmovn.i32
%1 = fptoui <4 x float> %v to <4 x i8>
ret <4 x i8> %1
}
; CHECK: i:
define <4 x i8> @i(<4 x i8>* %x) {
; Note: vld1 here is reasonably important. Mixing VFP and NEON
; instructions is bad on some cores
; CHECK: vld1
; CHECK: vmovl.s8
; CHECK: vmovl.s16
; CHECK: vrecpe
; CHECK: vrecps
; CHECK: vmul
; CHECK: vmovn
%1 = load <4 x i8>* %x, align 4
%2 = sdiv <4 x i8> zeroinitializer, %1
ret <4 x i8> %2
}