llvm-6502/test/CodeGen/X86/vec_insert-4.ll

; RUN: llc < %s -march=x86 -mcpu=yonah | grep 1084227584 | count 1

; ModuleID = '<stdin>'
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
target triple = "i686-apple-darwin9.2.2"

define <8 x float> @f(<8 x float> %a, i32 %b) nounwind  {
entry:
	%vecins = insertelement <8 x float> %a, float 5.000000e+00, i32 %b		; <<4 x float>> [#uses=1]
	ret <8 x float> %vecins
}
Eliminate more uses of llvm-as and llvm-dis. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81290 91177308-0d34-0410-b5e6-96231b3b80d8 2009-09-08 23:54:48 +00:00			`; RUN: llc < %s -march=x86 -mcpu=yonah \| grep 1084227584 \| count 1`
Add a testcase for the recent "handle variable vector insert elt in mem" patch git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@50303 91177308-0d34-0410-b5e6-96231b3b80d8 2008-04-25 21:26:59 +00:00
			`; ModuleID = '<stdin>'`
			`target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"`
			`target triple = "i686-apple-darwin9.2.2"`

			`define <8 x float> @f(<8 x float> %a, i32 %b) nounwind {`
			`entry:`
			`%vecins = insertelement <8 x float> %a, float 5.000000e+00, i32 %b ; <<4 x float>> [#uses=1]`
			`ret <8 x float> %vecins`
			`}`