AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update Barrier_All and Sync_All APIs (#72)

Browse Source 
* Fix deadlock in `rocshmem_ctx_wg_barrier_all` API in IPC conduit by adding per-context pSync buffers and context IDs
  - Added separate pSync buffers for each device context
  - Resolved deadlock when invoking barrier API (`rocshmem_ctx_wg_barrier_all`) concurrently from multiple contexts

* Update barrier_all functional tests for multi-context support

* Add thread, wavefront, and workgroup-level barrier_all APIs in IPC and RO conduits
  - Implemented barrier_all APIs at thread, wavefront, and workgroup granularity
  - Added support in both IPC and RO conduits
  - Updated functional tests to cover all `barrier_all` APIs

* Add thread, wavefront, and workgroup-level sync_all APIs in IPC and RO conduits
  - Implemented sync_all APIs for thread, wavefront, and workgroup scopes
  - Added support into both IPC and RO conduits
  - Added functional tests to cover all `sync_all` APIs
This commit is contained in:
Avinash Kethineedi
2025-04-02 11:58:55 -05:00
committed by GitHub
parent e16ca7a1e3
commit c652f58cef
22 changed files with 508 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/functional_tests/sync_tester.cpp
-9
View File
@@ -41,15 +41,6 @@ __global__ void SyncTest(int loop, int skip, long long int *start_time,
__syncthreads();
switch (type) {
case SyncAllTestType:
/**
* The function `rocshmem_ctx_wg_sync_all` should be called from only
* one group within the grid to avoid unintended behavior.
*/
if (is_block_zero_in_grid()) {
rocshmem_ctx_wg_sync_all(ctx);
}
break;
case SyncTestType:
rocshmem_ctx_wg_team_sync(ctx, teams[wg_id]);
break;