| Hongbin Zheng | 6417255 | 2012-05-06 10:22:43 +0000 | [diff] [blame] | 1 | ; RUN: opt %loadPolly -basicaa -polly-codegen %vector-opt -S %s | FileCheck %s |
| Tobias Grosser | 4cb5461 | 2012-04-12 10:46:55 +0000 | [diff] [blame] | 2 | target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" |
| 3 | target triple = "x86_64-unknown-linux-gnu" |
| 4 | |
| 5 | @A = common global [1024 x float] zeroinitializer, align 16 |
| 6 | @B = common global [1024 x float] zeroinitializer, align 16 |
| 7 | |
| 8 | declare float @foo(float) readnone |
| 9 | |
| 10 | define void @simple_vec_call() nounwind { |
| 11 | entry: |
| 12 | br label %body |
| 13 | |
| 14 | body: |
| 15 | %indvar = phi i64 [ 0, %entry ], [ %indvar_next, %body ] |
| 16 | %scevgep = getelementptr [1024 x float]* @B, i64 0, i64 %indvar |
| 17 | %value = load float* getelementptr inbounds ([1024 x float]* @A, i64 0, i64 0), align 16 |
| 18 | %result = tail call float @foo(float %value) nounwind |
| 19 | store float %result, float* %scevgep, align 4 |
| 20 | %indvar_next = add i64 %indvar, 1 |
| 21 | %exitcond = icmp eq i64 %indvar_next, 4 |
| 22 | br i1 %exitcond, label %return, label %body |
| 23 | |
| 24 | return: |
| 25 | ret void |
| 26 | } |
| 27 | |
| 28 | ; CHECK: %value_p_splat_one = load <1 x float>* bitcast ([1024 x float]* @A to <1 x float>*), align 8 |
| 29 | ; CHECK: %value_p_splat = shufflevector <1 x float> %value_p_splat_one, <1 x float> %value_p_splat_one, <4 x i32> zeroinitializer |
| 30 | ; CHECK: %0 = extractelement <4 x float> %value_p_splat, i32 0 |
| 31 | ; CHECK: %1 = extractelement <4 x float> %value_p_splat, i32 1 |
| 32 | ; CHECK: %2 = extractelement <4 x float> %value_p_splat, i32 2 |
| 33 | ; CHECK: %3 = extractelement <4 x float> %value_p_splat, i32 3 |
| Tobias Grosser | 7c3061a | 2012-04-24 11:08:07 +0000 | [diff] [blame] | 34 | ; CHECK: [[RES1:%[a-zA-Z0-9_]+]] = tail call float @foo(float %0) nounwind |
| 35 | ; CHECK: [[RES2:%[a-zA-Z0-9_]+]] = tail call float @foo(float %1) nounwind |
| 36 | ; CHECK: [[RES3:%[a-zA-Z0-9_]+]] = tail call float @foo(float %2) nounwind |
| 37 | ; CHECK: [[RES4:%[a-zA-Z0-9_]+]] = tail call float @foo(float %3) nounwind |
| 38 | ; CHECK: %4 = insertelement <4 x float> undef, float [[RES1]], i32 0 |
| 39 | ; CHECK: %5 = insertelement <4 x float> %4, float [[RES2]], i32 1 |
| 40 | ; CHECK: %6 = insertelement <4 x float> %5, float [[RES3]], i32 2 |
| 41 | ; CHECK: %7 = insertelement <4 x float> %6, float [[RES4]], i32 3 |
| 42 | ; CHECK: store <4 x float> %7 |