6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-11-15 20:06:46 +00:00
Code Issues Projects Releases Wiki Activity
9467b3e0ac
llvm-6502/test/CodeGen/ARM/neon_div.ll

49 lines
1.1 KiB
LLVM
Raw Normal View History

ARM scheduling fix: compute predicated implicit use properly. Minor drive by fix to cleanup latency computation. Calling getOperandLatency with a deliberately incorrect operand index does not give you the latency you want. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158959 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-22 02:50:31 +00:00
; RUN: llc < %s -march=arm -mattr=+neon -pre-RA-sched=source -disable-post-ra | FileCheck %s
Implement sdiv & udiv for <4 x i16> and <8 x i8> NEON vector types. This avoids moving each element to the integer register file and calling __divsi3 etc. on it. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125402 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-11 20:53:29 +00:00
define <8 x i8> @sdivi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK: vrecpe.f32
;CHECK: vmovn.i32
ARM scheduling fix: compute predicated implicit use properly. Minor drive by fix to cleanup latency computation. Calling getOperandLatency with a deliberately incorrect operand index does not give you the latency you want. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158959 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-22 02:50:31 +00:00
;CHECK: vrecpe.f32
Implement sdiv & udiv for <4 x i16> and <8 x i8> NEON vector types. This avoids moving each element to the integer register file and calling __divsi3 etc. on it. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125402 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-11 20:53:29 +00:00
;CHECK: vmovn.i32
;CHECK: vmovn.i16
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
%tmp3 = sdiv <8 x i8> %tmp1, %tmp2
ret <8 x i8> %tmp3
}
define <8 x i8> @udivi8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
;CHECK: vrecpe.f32
;CHECK: vrecps.f32
ARM scheduling fix: compute predicated implicit use properly. Minor drive by fix to cleanup latency computation. Calling getOperandLatency with a deliberately incorrect operand index does not give you the latency you want. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158959 91177308-0d34-0410-b5e6-96231b3b80d8
2012-06-22 02:50:31 +00:00
;CHECK: vmovn.i32
Implement sdiv & udiv for <4 x i16> and <8 x i8> NEON vector types. This avoids moving each element to the integer register file and calling __divsi3 etc. on it. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@125402 91177308-0d34-0410-b5e6-96231b3b80d8
2011-02-11 20:53:29 +00:00
;CHECK: vrecpe.f32
;CHECK: vrecps.f32
;CHECK: vmovn.i32
;CHECK: vqmovun.s16
%tmp1 = load <8 x i8>* %A
%tmp2 = load <8 x i8>* %B
%tmp3 = udiv <8 x i8> %tmp1, %tmp2
ret <8 x i8> %tmp3
}
define <4 x i16> @sdivi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK: vrecpe.f32
;CHECK: vrecps.f32
;CHECK: vmovn.i32
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
%tmp3 = sdiv <4 x i16> %tmp1, %tmp2
ret <4 x i16> %tmp3
}
define <4 x i16> @udivi16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
;CHECK: vrecpe.f32
;CHECK: vrecps.f32
;CHECK: vrecps.f32
;CHECK: vmovn.i32
%tmp1 = load <4 x i16>* %A
%tmp2 = load <4 x i16>* %B
%tmp3 = udiv <4 x i16> %tmp1, %tmp2
ret <4 x i16> %tmp3
}
Reference in New Issue Copy Permalink