mirror of
https://github.com/wolfpld/tracy.git
synced 2026-08-31 23:48:26 +00:00
6c6999bf01
Minimal reproducer showing that CUDA Graph-launched kernels produce 0 GPU zones in Tracy. The repro creates a simple graph (2 kernels + 1 memcpy), launches it 10 times, and expects ~30 GPU zones. Without the fallback patch, all activity records are dropped by matchError(). Tested on NVIDIA H100, CUDA 13.1.
36 lines
862 B
Markdown
36 lines
862 B
Markdown
# Tracy CUDA Graph GPU Zone Repro
|
|
|
|
Demonstrates that unpatched Tracy fails to show GPU zones for kernels
|
|
launched via CUDA Graphs (`cudaGraphLaunch`).
|
|
|
|
## Root cause
|
|
|
|
When kernels are launched through CUDA Graphs, CUPTI delivers
|
|
`CONCURRENT_KERNEL` and `MEMCPY` activity records but no corresponding
|
|
API callback fires for the individual kernel launches. Tracy's
|
|
`matchActivityToAPICall()` always fails, and `matchError()` silently
|
|
drops every GPU zone.
|
|
|
|
## Build and run
|
|
|
|
```bash
|
|
make
|
|
./repro
|
|
```
|
|
|
|
## What to expect
|
|
|
|
| Tracy version | GPU zones shown |
|
|
|---|---|
|
|
| Unpatched | 0 |
|
|
| Patched (cuda-graph-gpu-zones.patch) | ~30 (10 launches x 3 ops) |
|
|
|
|
## The graph structure
|
|
|
|
Each graph launch contains:
|
|
1. `vector_add` kernel (c = a + b)
|
|
2. Device-to-device memcpy
|
|
3. `vector_add` kernel (c = a + c)
|
|
|
|
The graph is launched 10 times, so 30 GPU operations total.
|