330d8983d2
In a nutshell, this moves our libomptarget code to populate the offload subproject. With this commit, users need to enable the new LLVM/Offload subproject as a runtime in their cmake configuration. No further changes are expected for downstream code. Tests and other components still depend on OpenMP and have also not been renamed. The results below are for a build in which OpenMP and Offload are enabled runtimes. In addition to the pure `git mv`, we needed to adjust some CMake files. Nothing is intended to change semantics. ``` ninja check-offload ``` Works with the X86 and AMDGPU offload tests ``` ninja check-openmp ``` Still works but doesn't build offload tests anymore. ``` ls install/lib ``` Shows all expected libraries, incl. - `libomptarget.devicertl.a` - `libomptarget-nvptx-sm_90.bc` - `libomptarget.rtl.amdgpu.so` -> `libomptarget.rtl.amdgpu.so.18git` - `libomptarget.so` -> `libomptarget.so.18git` Fixes: https://github.com/llvm/llvm-project/issues/75124 --------- Co-authored-by: Saiyedul Islam <Saiyedul.Islam@amd.com>
54 lines
1.9 KiB
C
54 lines
1.9 KiB
C
// Use the generic state machine. On some architectures, other threads in the
|
|
// main thread's warp must avoid barrier instructions.
|
|
//
|
|
// RUN: %libomptarget-compile-run-and-check-generic
|
|
|
|
// SPMDize. There is no main thread, so there's no issue.
|
|
//
|
|
// RUN: %libomptarget-compile-generic -O1 -Rpass=openmp-opt > %t.spmd 2>&1
|
|
// RUN: %fcheck-nvptx64-nvidia-cuda -check-prefix=SPMD -input-file=%t.spmd
|
|
// RUN: %fcheck-amdgcn-amd-amdhsa -check-prefix=SPMD -input-file=%t.spmd
|
|
// RUN: %libomptarget-run-generic 2>&1 | %fcheck-generic
|
|
//
|
|
// SPMD: Transformed generic-mode kernel to SPMD-mode.
|
|
|
|
// Use the custom state machine, which must avoid the same barrier problem as
|
|
// the generic state machine.
|
|
//
|
|
// RUN: %libomptarget-compile-generic -O1 -Rpass=openmp-opt \
|
|
// RUN: -mllvm -openmp-opt-disable-spmdization > %t.custom 2>&1
|
|
// RUN: %fcheck-nvptx64-nvidia-cuda -check-prefix=CUSTOM -input-file=%t.custom
|
|
// RUN: %fcheck-amdgcn-amd-amdhsa -check-prefix=CUSTOM -input-file=%t.custom
|
|
// RUN: %libomptarget-run-generic 2>&1 | %fcheck-generic
|
|
//
|
|
// Repeat with reduction clause, which has managed to break the custom state
|
|
// machine in the past.
|
|
//
|
|
// RUN: %libomptarget-compile-generic -O1 -Rpass=openmp-opt -DADD_REDUCTION \
|
|
// RUN: -mllvm -openmp-opt-disable-spmdization > %t.custom 2>&1
|
|
// RUN: %fcheck-nvptx64-nvidia-cuda -check-prefix=CUSTOM -input-file=%t.custom
|
|
// RUN: %fcheck-amdgcn-amd-amdhsa -check-prefix=CUSTOM -input-file=%t.custom
|
|
// RUN: %libomptarget-run-generic 2>&1 | %fcheck-generic
|
|
//
|
|
// CUSTOM: Rewriting generic-mode kernel with a customized state machine.
|
|
|
|
#if ADD_REDUCTION
|
|
#define REDUCTION(...) reduction(__VA_ARGS__)
|
|
#else
|
|
#define REDUCTION(...)
|
|
#endif
|
|
|
|
#include <stdio.h>
|
|
int main() {
|
|
int x = 0, y = 1;
|
|
#pragma omp target teams num_teams(1) map(tofrom : x, y) REDUCTION(+ : x)
|
|
{
|
|
x += 5;
|
|
#pragma omp parallel
|
|
y = 6;
|
|
}
|
|
// CHECK: 5, 6
|
|
printf("%d, %d\n", x, y);
|
|
return 0;
|
|
}
|