aomp
aomp copied to clipboard
ROCm
AOMP is an open source Clang/LLVM based compiler with added support for the OpenMP® API on Radeon™ GPUs. Use this repository for releases, issues, documentation, packaging, and examples.
…) write to the same global memory location without any synchronization
Is it possible to build "fat binary" which is compatible with multiple GPU architectures? I'm able to build binary with OpenMP target offload for single GPU architecture (such as AOMP_GPU=gfx900),...
red_bug_51 is racy in that it uses multiple teams, each updating counts1 and counts2 mapped scalars, without using atomics or other mechanisms to prevent data race. This PR removes teams.
Updated build script to reflect latest changes and updated some variables in f18_common_vars modified: build_project.sh modified: f18.sh modified: f18_common_vars
Scalars are implicitly passed to a target region as firstprivate
Please note that the test has an unnecessary target region in it. Without that, clang will not emit the register_requires call to the openmp runtime that sets up the flags...
E.g. https://github.com/ROCm-Developer-Tools/llvm-project/pull/93 It's line noise. I don't have permissions to delete it.
On my apu laptop when graphic memory is set low, the memory allocation failure caused a deadlock in the device plugin. ``` [/home/estewart/git/aomp11/amd-llvm-project/openmp/libomptarget/plugins/hsa/impl/data.cpp:99] atmi_malloc failed: HSA_STATUS_ERROR_INVALID_ALLOCATION ``` backtrace ``` __lll_lock_wait...
I have built latest AOMP (SHA: e2f40a73975ccc0f453e8767a8016e6deb849782) with the amd-stg-open branch. However it is unable to enumerate the HIP GPU device though rocminfo shows both cpu and gpu. I have...
