1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=i386-apple-darwin -mattr=+sse2 | FileCheck %s --check-prefix=X32
3 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+sse2 | FileCheck %s --check-prefix=X64
5 define i32 @t(i8* %ref_frame_ptr, i32 %ref_frame_stride, i32 %idxX, i32 %idxY) nounwind {
7 ; X32: ## %bb.0: ## %entry
8 ; X32-NEXT: movl {{[0-9]+}}(%esp), %eax
9 ; X32-NEXT: movl {{[0-9]+}}(%esp), %ecx
10 ; X32-NEXT: imull {{[0-9]+}}(%esp), %ecx
11 ; X32-NEXT: addl {{[0-9]+}}(%esp), %ecx
12 ; X32-NEXT: movl (%eax,%ecx), %eax
16 ; X64: ## %bb.0: ## %entry
17 ; X64-NEXT: imull %ecx, %esi
18 ; X64-NEXT: addl %edx, %esi
19 ; X64-NEXT: movslq %esi, %rax
20 ; X64-NEXT: movl (%rdi,%rax), %eax
21 ; X64-NEXT: movq %rax, %xmm0
22 ; X64-NEXT: movd %xmm0, %eax
25 %tmp7 = mul i32 %idxY, %ref_frame_stride ; <i32> [#uses=2]
26 %tmp9 = add i32 %tmp7, %idxX ; <i32> [#uses=1]
27 %tmp11 = getelementptr i8, i8* %ref_frame_ptr, i32 %tmp9 ; <i8*> [#uses=1]
28 %tmp1112 = bitcast i8* %tmp11 to i32* ; <i32*> [#uses=1]
29 %tmp13 = load i32, i32* %tmp1112, align 4 ; <i32> [#uses=1]
30 %tmp18 = add i32 %idxX, 4 ; <i32> [#uses=1]
31 %tmp20.sum = add i32 %tmp18, %tmp7 ; <i32> [#uses=1]
32 %tmp21 = getelementptr i8, i8* %ref_frame_ptr, i32 %tmp20.sum ; <i8*> [#uses=1]
33 %tmp2122 = bitcast i8* %tmp21 to i16* ; <i16*> [#uses=1]
34 %tmp23 = load i16, i16* %tmp2122, align 2 ; <i16> [#uses=1]
35 %tmp2425 = zext i16 %tmp23 to i64 ; <i64> [#uses=1]
36 %tmp26 = shl i64 %tmp2425, 32 ; <i64> [#uses=1]
37 %tmp2728 = zext i32 %tmp13 to i64 ; <i64> [#uses=1]
38 %tmp29 = or i64 %tmp26, %tmp2728 ; <i64> [#uses=1]
39 %tmp3454 = bitcast i64 %tmp29 to double ; <double> [#uses=1]
40 %tmp35 = insertelement <2 x double> undef, double %tmp3454, i32 0 ; <<2 x double>> [#uses=1]
41 %tmp36 = insertelement <2 x double> %tmp35, double 0.000000e+00, i32 1 ; <<2 x double>> [#uses=1]
42 %tmp42 = bitcast <2 x double> %tmp36 to <8 x i16> ; <<8 x i16>> [#uses=1]
43 %tmp43 = shufflevector <8 x i16> %tmp42, <8 x i16> undef, <8 x i32> < i32 0, i32 1, i32 1, i32 2, i32 4, i32 5, i32 6, i32 7 > ; <<8 x i16>> [#uses=1]
44 %tmp47 = bitcast <8 x i16> %tmp43 to <4 x i32> ; <<4 x i32>> [#uses=1]
45 %tmp48 = extractelement <4 x i32> %tmp47, i32 0 ; <i32> [#uses=1]