1 ; RUN: llc -march=hexagon -mcpu=hexagonv5 -enable-pipeliner < %s -pipeliner-experimental-cg=true | FileCheck %s
2 ; RUN: llc -march=hexagon -mcpu=hexagonv5 -O2 < %s -pipeliner-experimental-cg=true | FileCheck %s
3 ; RUN: llc -march=hexagon -mcpu=hexagonv5 -O3 < %s -pipeliner-experimental-cg=true | FileCheck %s
4 ; RUN: llc -march=hexagon -mcpu=hexagonv5 -enable-pipeliner < %s -pipeliner-experimental-cg=true -early-live-intervals -verify-machineinstrs | FileCheck %s
6 ; Check that we pipeline a vectorized dot product in a single packet.
15 @a = common global [5000 x i32] zeroinitializer, align 8
16 @b = common global [5000 x i32] zeroinitializer, align 8
18 define i32 @vecMultGlobal() {
20 br label %polly.loop_body
23 %0 = extractelement <2 x i32> %addp_vec, i32 0
24 %1 = extractelement <2 x i32> %addp_vec, i32 1
25 %add_sum = add i32 %0, %1
29 %polly.loopiv13 = phi i32 [ 0, %entry ], [ %polly.next_loopiv, %polly.loop_body ]
30 %reduction.012 = phi <2 x i32> [ zeroinitializer, %entry ], [ %addp_vec, %polly.loop_body ]
31 %polly.next_loopiv = add nsw i32 %polly.loopiv13, 2
32 %p_arrayidx1 = getelementptr [5000 x i32], ptr @b, i32 0, i32 %polly.loopiv13
33 %p_arrayidx = getelementptr [5000 x i32], ptr @a, i32 0, i32 %polly.loopiv13
34 %_p_vec_full = load <2 x i32>, ptr %p_arrayidx1, align 8
35 %_p_vec_full8 = load <2 x i32>, ptr %p_arrayidx, align 8
36 %mulp_vec = mul <2 x i32> %_p_vec_full8, %_p_vec_full
37 %addp_vec = add <2 x i32> %mulp_vec, %reduction.012
38 %2 = icmp slt i32 %polly.next_loopiv, 5000
39 br i1 %2, label %polly.loop_body, label %polly.loop_after