build(cuda): enable GGML_CUDA_GRAPHS on CUDA builds#26
Merged
Conversation
ggml leaves GGML_CUDA_GRAPHS off by default. Turning it on lets the CUDA backend capture and replay the compute graph, a small but free speedup (about 1% measured on a GB10: 1.4% on tdt-1.1b, 0.4% on tdt-0.6b-v3, and never negative across interleaved runs). Enable it whenever we forward CUDA so every CUDA build (docker, release, local) inherits it. The runtime kill-switch GGML_CUDA_DISABLE_GRAPHS=1 still disables it for A/B testing. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
mudler
approved these changes
Jun 12, 2026
adyranov
pushed a commit
to adyranov/parakeet.cpp
that referenced
this pull request
Jun 13, 2026
ggml leaves GGML_CUDA_GRAPHS off by default. Turning it on lets the CUDA backend capture and replay the compute graph, a small but free speedup (about 1% measured on a GB10: 1.4% on tdt-1.1b, 0.4% on tdt-0.6b-v3, and never negative across interleaved runs). Enable it whenever we forward CUDA so every CUDA build (docker, release, local) inherits it. The runtime kill-switch GGML_CUDA_DISABLE_GRAPHS=1 still disables it for A/B testing. Co-authored-by: Ettore Di Giacinto <mudler@localai.io> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Enable
GGML_CUDA_GRAPHSwhenever parakeet.cpp forwards CUDA (PARAKEET_GGML_CUDA=ON). ggml leaves this off by default.Why
With CUDA graphs on, the CUDA backend captures and replays the compute graph, a small but free speedup. Measured on a GB10 (interleaved, best-of, same 180s clip):
Never negative across runs. The gain is capped because parakeet rebuilds each graph in a fresh
ggml_contextper call, which defeats ggml's cross-call graph replay (keyed on the first node pointer); the encoder also runs once per request. Lifting that is a separate, larger change. This PR just takes the free win.Notes
CMakeLists.txt, so docker/release/local CUDA builds all inherit it.GGML_CUDA_DISABLE_GRAPHS=1still works for A/B testing.-DGGML_CUDA_GRAPHS=ONon the GB10 and benchmarking; this change just automates that flag.🤖 Generated with Claude Code