Scarfski/tracy
1
0
Fork 0
mirror of https://github.com/wolfpld/tracy.git synced 2026-10-16 13:55:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Also track cudaGraphLaunch runtime API for graph correlation

Browse Source 
The repro uses cudaGraphLaunch (runtime API) not cuGraphLaunch (driver
API). Add cudaGraphLaunch_v10000 and its _ptsz variant to
cbidRuntimeTrackers so that graphs launched via the runtime API also
get their CPU call site captured for GPU zone correlation.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Basil Milanich
2026-04-06 11:18:20 -05:00
parent 7bca9dcd90
commit 2e5b076175
1 changed files with 4 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
public/tracy/TracyCUDA.hpp
View File

@@ -786,6 +786,10 @@ namespace tracy
{ CUPTI_RUNTIME_TRACE_CBID_cudaEventQuery_v3020, NON_STREAM_FUNC() },
{ CUPTI_RUNTIME_TRACE_CBID_cudaStreamWaitEvent_v3020, NON_STREAM_FUNC() },
{ CUPTI_RUNTIME_TRACE_CBID_cudaDeviceSynchronize_v3020, NON_STREAM_FUNC() },
// Graph launch: tracked so we can correlate GPU activities back to
// the cudaGraphLaunch call site via CUPTI_ACTIVITY_KIND_GRAPH_TRACE
{ CUPTI_RUNTIME_TRACE_CBID_cudaGraphLaunch_v10000, GET_STREAM_FUNC(cudaGraphLaunch_v10000_params, stream) },
{ CUPTI_RUNTIME_TRACE_CBID_cudaGraphLaunch_ptsz_v10000, GET_STREAM_FUNC(cudaGraphLaunch_v10000_params, stream) },
};
#undef NON_STREAM_FUNC
#undef GET_STREAM_FUNC