1b1c8d83d3
Turning on `enable_noundef_analysis` flag allows better codegen by removing freeze instructions. I modified clang by renaming `enable_noundef_analysis` flag to `disable-noundef-analysis` and turning it off by default. Test updates are made as a separate patch: D108453 Reviewed By: eugenis Differential Revision: https://reviews.llvm.org/D105169
21 lines
1.1 KiB
C
21 lines
1.1 KiB
C
// RUN: %clang_cc1 -triple hexagon -emit-llvm -target-cpu hexagonv66 -target-feature +hvxv66 -target-feature +hvx-length64b -o - %s | FileCheck %s --check-prefix CHECK-HVX64
|
|
// RUN: %clang_cc1 -triple hexagon -emit-llvm -target-cpu hexagonv66 -target-feature +hvxv66 -target-feature +hvx-length128b -o - %s | FileCheck %s --check-prefix CHECK-HVX128
|
|
|
|
typedef long HVX_Vector __attribute__((__vector_size__(__HVX_LENGTH__)))
|
|
__attribute__((aligned(__HVX_LENGTH__)));
|
|
typedef long HVX_VectorPair __attribute__((__vector_size__(2*__HVX_LENGTH__)))
|
|
__attribute__((aligned(__HVX_LENGTH__)));
|
|
|
|
// CHECK-HVX64: define {{.*}} <16 x i32> @foo(<16 x i32> noundef %a, <32 x i32> noundef %b)
|
|
// CHECK-HVX128: define {{.*}} <32 x i32> @foo(<32 x i32> noundef %a, <64 x i32> noundef %b)
|
|
HVX_Vector foo(HVX_Vector a, HVX_VectorPair b) {
|
|
return a;
|
|
}
|
|
|
|
// CHECK-HVX64: define {{.*}} <32 x i32> @bar(<16 x i32> noundef %a, <32 x i32> noundef %b)
|
|
// CHECK-HVX128: define {{.*}} <64 x i32> @bar(<32 x i32> noundef %a, <64 x i32> noundef %b)
|
|
HVX_VectorPair bar(HVX_Vector a, HVX_VectorPair b) {
|
|
return b;
|
|
}
|
|
|