From 51409214d7471328df2e92e0a8f9d05221fe0e6f Mon Sep 17 00:00:00 2001 From: Nate Begeman Date: Wed, 28 Jul 2010 00:21:48 +0000 Subject: Implement a vectorized algorithm for <16 x i8> << <16 x i8> This is about 4x faster and smaller than the existing scalarization. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109566 91177308-0d34-0410-b5e6-96231b3b80d8 --- test/CodeGen/X86/vec_shift4.ll | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'test/CodeGen/X86/vec_shift4.ll') diff --git a/test/CodeGen/X86/vec_shift4.ll b/test/CodeGen/X86/vec_shift4.ll index d8f4e4e..9ef7fbd 100644 --- a/test/CodeGen/X86/vec_shift4.ll +++ b/test/CodeGen/X86/vec_shift4.ll @@ -1,6 +1,6 @@ ; RUN: llc < %s -march=x86 -mattr=+sse41 | FileCheck %s -define <2 x i64> @shl(<4 x i32> %r, <4 x i32> %a) nounwind readnone ssp { +define <2 x i64> @shl1(<4 x i32> %r, <4 x i32> %a) nounwind readnone ssp { entry: ; CHECK-NOT: shll ; CHECK: pslld @@ -12,3 +12,14 @@ entry: %tmp2 = bitcast <4 x i32> %shl to <2 x i64> ; <<2 x i64>> [#uses=1] ret <2 x i64> %tmp2 } + +define <2 x i64> @shl2(<16 x i8> %r, <16 x i8> %a) nounwind readnone ssp { +entry: +; CHECK-NOT: shlb +; CHECK: pblendvb +; CHECK: pblendvb +; CHECK: pblendvb + %shl = shl <16 x i8> %r, %a ; <<16 x i8>> [#uses=1] + %tmp2 = bitcast <16 x i8> %shl to <2 x i64> ; <<2 x i64>> [#uses=1] + ret <2 x i64> %tmp2 +} -- cgit v1.1