maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/OpenMP/distribute_simd_reduction_c...

218 lines
11 KiB
C++
Raw Normal View History

[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-64
// RUN: %clang_cc1 -DCHECK -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix CHECK --check-prefix CHECK-32
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCHECK -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DCHECK -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY0 %s
// SIMD-ONLY0-NOT: {{__kmpc|__tgt}}
[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DLAMBDA -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix LAMBDA --check-prefix LAMBDA-64
[OPENMP] Support for -fopenmp-simd option with compilation of simd loops only. Added support for -fopenmp-simd option that allows compilation of simd-based constructs without emission of OpenMP runtime calls. llvm-svn: 321560
2017-12-30 02:07:07 +08:00
// RUN: %clang_cc1 -DLAMBDA -verify -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck --check-prefix SIMD-ONLY1 %s
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s
// RUN: %clang_cc1 -DLAMBDA -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck --check-prefix SIMD-ONLY1 %s
// SIMD-ONLY1-NOT: {{__kmpc|__tgt}}
[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// expected-no-diagnostics
#ifndef HEADER
#define HEADER
template <typename T>
T tmain() {
T t_var = T();
T vec[] = {1, 2};
#pragma omp target
#pragma omp teams
#pragma omp distribute simd reduction(+: t_var)
for (int i = 0; i < 2; ++i) {
t_var += (T) i;
}
return T();
}
int main() {
static int sivar;
#ifdef LAMBDA
// LAMBDA-LABEL: @main
// LAMBDA: call void [[OUTER_LAMBDA:@.+]](
[&]() {
// LAMBDA: define{{.*}} internal{{.*}} void [[OUTER_LAMBDA]](
// LAMBDA: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// LAMBDA: call void @[[LOFFL1:.+]](
// LAMBDA: ret
#pragma omp target
#pragma omp teams
#pragma omp distribute simd reduction(+: sivar)
for (int i = 0; i < 2; ++i) {
// LAMBDA: define{{.*}} internal{{.*}} void @[[LOFFL1]](i{{64|32}} [[SIVAR_ARG:%.+]])
// LAMBDA: [[SIVAR_ADDR:%.+]] = alloca i{{.+}},
// LAMBDA: store{{.+}} [[SIVAR_ARG]], {{.+}} [[SIVAR_ADDR]],
// LAMBDA: [[SIVAR_CONV:%.+]] = bitcast{{.+}} [[SIVAR_ADDR]] to
// LAMBDA: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 1, {{.+}} @[[LOUTL1:.+]] to {{.+}}, {{.+}} [[SIVAR_CONV]])
// LAMBDA: ret void
// LAMBDA: define internal void @[[LOUTL1]]({{.+}}, {{.+}}, {{.+}} [[SIVAR_ARG:%.+]])
// Skip global and bound tid vars
// LAMBDA: {{.+}} = alloca i32*,
// LAMBDA: {{.+}} = alloca i32*,
// LAMBDA: [[SIVAR_ADDR:%.+]] = alloca i{{.+}}*,
// LAMBDA: alloca i{{.+}},
// LAMBDA: alloca i{{.+}},
// LAMBDA: alloca i{{.+}},
// LAMBDA: alloca i{{.+}},
// LAMBDA: alloca i{{.+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// LAMBDA: alloca i{{.+}},
[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// LAMBDA: [[SIVAR_PRIV:%.+]] = alloca i{{.+}},
// LAMBDA: store{{.+}} [[SIVAR_ARG]], {{.+}} [[SIVAR_ADDR]],
// LAMBDA: [[SIVAR_REF:%.+]] = load{{.+}}, {{.+}} [[SIVAR_ADDR]]
// LAMBDA: store{{.+}} 0, {{.+}} [[SIVAR_PRIV]],
// LAMBDA: call void @__kmpc_for_static_init_4(
// LAMBDA: store{{.+}}, {{.+}} [[SIVAR_PRIV]],
// LAMBDA: call void [[INNER_LAMBDA:@.+]](
// LAMBDA: call void @__kmpc_for_static_fini(
// LAMBDA: [[LAST_ITER:%.+]] = load i32, i32* %
// LAMBDA: [[IS_LAST:%.+]] = icmp ne i32 [[LAST_ITER]], 0
// LAMBDA: br i1 [[IS_LAST]], label %[[THEN:.+]], label %[[DONE:.+]]
// LAMBDA: [[THEN]]
// LAMBDA: store i32 2, i32* %
// LAMBDA: br label %[[DONE]]
// LAMBDA: [[DONE]]
// LAMBDA: [[SIVAR_ORIG_VAL:%.+]] = load i32, i32* [[SIVAR_REF]],
// LAMBDA: [[SIVAR_PRIV_VAL:%.+]] = load i32, i32* [[SIVAR_PRIV]],
// LAMBDA: [[ADD:%.+]] = add nsw i32 [[SIVAR_ORIG_VAL]], [[SIVAR_PRIV_VAL]]
// LAMBDA: store i32 [[ADD]], i32* [[SIVAR_REF]],
// LAMBDA: ret void
sivar += i;
[&]() {
// LAMBDA: define {{.+}} void [[INNER_LAMBDA]](%{{.+}}* [[ARG_PTR:%.+]])
// LAMBDA: store %{{.+}}* [[ARG_PTR]], %{{.+}}** [[ARG_PTR_REF:%.+]],
sivar += 4;
// LAMBDA: [[ARG_PTR:%.+]] = load %{{.+}}*, %{{.+}}** [[ARG_PTR_REF]]
// LAMBDA: [[SIVAR_PTR_REF:%.+]] = getelementptr inbounds %{{.+}}, %{{.+}}* [[ARG_PTR]], i{{[0-9]+}} 0, i{{[0-9]+}} 0
// LAMBDA: [[SIVAR_REF:%.+]] = load i{{[0-9]+}}*, i{{[0-9]+}}** [[SIVAR_PTR_REF]]
// LAMBDA: [[SIVAR_VAL:%.+]] = load i{{[0-9]+}}, i{{[0-9]+}}* [[SIVAR_REF]]
// LAMBDA: [[SIVAR_INC:%.+]] = add{{.+}} [[SIVAR_VAL]], 4
// LAMBDA: store i{{[0-9]+}} [[SIVAR_INC]], i{{[0-9]+}}* [[SIVAR_REF]]
}();
}
}();
return 0;
#else
#pragma omp target
#pragma omp teams
#pragma omp distribute simd reduction(+: sivar)
for (int i = 0; i < 2; ++i) {
sivar += i;
}
return tmain<int>();
#endif
}
// CHECK: define {{.*}}i{{[0-9]+}} @main()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1, i8** %{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64* {{.+}}@{{[^,]+}}, i32 0, i32 0), i32 0, i32 0)
// CHECK: call void @[[OFFL1:.+]](i{{64|32}} %{{.+}})
// CHECK: {{%.+}} = call{{.*}} i32 @[[TMAIN_INT:.+]]()
// CHECK: ret
// CHECK: define{{.*}} void @[[OFFL1]](i{{64|32}} [[SIVAR_ARG:%.+]])
// CHECK: [[SIVAR_ADDR:%.+]] = alloca i{{.+}},
// CHECK: store{{.+}} [[SIVAR_ARG]], {{.+}} [[SIVAR_ADDR]],
// CHECK-64: [[SIVAR_CONV:%.+]] = bitcast{{.+}} [[SIVAR_ADDR]] to
// CHECK-64: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 1, {{.+}} @[[OUTL1:.+]] to {{.+}}, {{.+}} [[SIVAR_CONV]])
// CHECK-32: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 1, {{.+}} @[[OUTL1:.+]] to {{.+}}, {{.+}} [[SIVAR_ADDR]])
// CHECK: ret void
// CHECK: define internal void @[[OUTL1]]({{.+}}, {{.+}}, {{.+}} [[SIVAR_ARG:%.+]])
// Skip global and bound tid vars
// CHECK: {{.+}} = alloca i32*,
// CHECK: {{.+}} = alloca i32*,
// CHECK: [[SIVAR_ADDR:%.+]] = alloca i{{.+}}*,
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// CHECK: alloca i{{.+}},
[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// CHECK: [[SIVAR_PRIV:%.+]] = alloca i{{.+}},
// CHECK: store{{.+}} [[SIVAR_ARG]], {{.+}} [[SIVAR_ADDR]],
// CHECK: [[SIVAR_REF:%.+]] = load{{.+}}, {{.+}} [[SIVAR_ADDR]]
// CHECK: store{{.+}} 0, {{.+}} [[SIVAR_PRIV]],
// CHECK: call void @__kmpc_for_static_init_4(
// CHECK: store{{.+}}, {{.+}} [[SIVAR_PRIV]],
// CHECK: call void @__kmpc_for_static_fini(
// CHECK: [[LAST_ITER:%.+]] = load i32, i32* %
// CHECK: [[IS_LAST:%.+]] = icmp ne i32 [[LAST_ITER]], 0
// CHECK: br i1 [[IS_LAST]], label %[[THEN:.+]], label %[[DONE:.+]]
// CHECK: [[THEN]]
// CHECK: store i32 2, i32* %
// CHECK: br label %[[DONE]]
// CHECK: [[DONE]]
// CHECK: [[SIVAR_ORIG_VAL:%.+]] = load i32, i32* [[SIVAR_REF]],
// CHECK: [[SIVAR_PRIV_VAL:%.+]] = load i32, i32* [[SIVAR_PRIV]],
// CHECK: [[ADD:%.+]] = add nsw i32 [[SIVAR_ORIG_VAL]], [[SIVAR_PRIV_VAL]]
// CHECK: store i32 [[ADD]], i32* [[SIVAR_REF]],
// CHECK: ret void
// CHECK: define{{.*}} i{{[0-9]+}} @[[TMAIN_INT]]()
// CHECK: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 1,
// CHECK: call void @[[TOFFL1:.+]]({{.+}})
// CHECK: ret
// CHECK: define{{.*}} void @[[TOFFL1]](i{{64|32}} [[TVAR_ARG:%.+]])
// CHECK: [[TVAR_ADDR:%.+]] = alloca i{{.+}},
// CHECK: store{{.+}} [[TVAR_ARG]], {{.+}} [[TVAR_ADDR]],
// CHECK-64: [[TVAR_CONV:%.+]] = bitcast{{.+}} [[TVAR_ADDR]] to
// CHECK-64: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 1, {{.+}} @[[TOUTL1:.+]] to {{.+}}, {{.+}} [[TVAR_CONV]])
// CHECK-32: call void {{.+}} @__kmpc_fork_teams({{.+}}, i32 1, {{.+}} @[[TOUTL1:.+]] to {{.+}}, {{.+}} [[TVAR_ADDR]])
// CHECK: ret void
// CHECK: define internal void @[[TOUTL1]]({{.+}}, {{.+}}, {{.+}} [[TVAR_ARG:%.+]])
// Skip global and bound tid vars
// CHECK: {{.+}} = alloca i32*,
// CHECK: {{.+}} = alloca i32*,
// CHECK: [[TVAR_ADDR:%.+]] = alloca i{{.+}}*,
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
[OPENMP] Fix PR35486: crash when collapsing loops with dependent iteration spaces. Though it is incorrect from point of view of OpenMP standard to have dependent iteration space in OpenMP loops, compiler should not crash. Patch fixes this problem. llvm-svn: 319700
2017-12-05 05:30:42 +08:00
// CHECK: alloca i{{.+}},
[OPENMP] Codegen for `distribute simd` directive. Initial codegen support for `distribute simd` directive. llvm-svn: 319661
2017-12-04 23:38:33 +08:00
// CHECK: alloca i{{.+}},
// CHECK: alloca i{{.+}},
// CHECK: [[TVAR_PRIV:%.+]] = alloca i{{.+}},
// CHECK: store{{.+}} [[TVAR_ARG]], {{.+}} [[TVAR_ADDR]],
// CHECK: [[TVAR_REF:%.+]] = load{{.+}}, {{.+}} [[TVAR_ADDR]]
// CHECK: store{{.+}} 0, {{.+}} [[TVAR_PRIV]],
// CHECK: call void @__kmpc_for_static_init_4(
// CHECK: store{{.+}}, {{.+}} [[TVAR_PRIV]],
// CHECK: call void @__kmpc_for_static_fini(
// CHECK: [[LAST_ITER:%.+]] = load i32, i32* %
// CHECK: [[IS_LAST:%.+]] = icmp ne i32 [[LAST_ITER]], 0
// CHECK: br i1 [[IS_LAST]], label %[[THEN:.+]], label %[[DONE:.+]]
// CHECK: [[THEN]]
// CHECK: store i32 2, i32* %
// CHECK: br label %[[DONE]]
// CHECK: [[DONE]]
// CHECK: [[TVAR_ORIG_VAL:%.+]] = load i32, i32* [[TVAR_REF]],
// CHECK: [[TVAR_PRIV_VAL:%.+]] = load i32, i32* [[TVAR_PRIV]],
// CHECK: [[ADD:%.+]] = add nsw i32 [[TVAR_ORIG_VAL]], [[TVAR_PRIV_VAL]]
// CHECK: store i32 [[ADD]], i32* [[TVAR_REF]],
// CHECK: ret void
// CHECK: !{!"llvm.loop.vectorize.enable", i1 true}
#endif