storytold/cloud-worker
2
0
Fork 0
mirror of https://github.com/storytold/cloud-worker.git synced 2026-10-17 00:23:49 +00:00
Code Issues Packages Projects Releases Wiki Activity

concurrency experiment measured: co-scheduling 13% slower than sequential — one worker per GPU confirmed

Browse Source 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Hanashi
2026-08-06 14:27:32 -04:00
parent 24134b9653
commit fbc68bb749
2 changed files with 23 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/BENCHMARKS.md
+12 -1
View File
@@ -77,7 +77,18 @@ both cards land then too.)
See `docs/CONCURRENCY.md` for the model. Experiment: two resident instances,
sequential pair vs concurrent pair of identical warm t2v jobs.
RESULTS_PENDING
Measured (pruned int8, t2v 864×480 5 s, both instances warm):
| Mode | Total for 2 videos | Per-video effective |
|-------------------------------|--------------------|---------------------|
| Sequential (one instance) | **128 s** | 64 s |
| Concurrent (two instances) | **145 s** | 72.5 s |
Concurrent was **13% slower** for the same work — each co-scheduled job ran
142 s vs 63 s alone (2.27×). There is no pipeline-overlap win; cache and
scheduler contention make co-scheduling strictly worse. **One worker per GPU,
queue in front, scale horizontally.** (This also removes any argument for
MPS/MIG partitioning for throughput.)
## Takeaways