perf-torch-sync-free
perf-torch-sync-free is an MLOps & model serving skill for AI agents, published by NVIDIA in the TensorRT-LLM repository on GitHub. Skillspace rates perf-torch-sync-free C (78/100) and it passed the safety scan; a judge model has also reviewed it.
Identify and eliminate host-device synchronizations in PyTorch code. Detects sync points (.item(), .cpu(), boolean indexing, torch.tensor on CUDA), classifies false vs true dependencies, provides sync-free alternatives. Triggers: sync-free, synchronization, .item(), .cpu(), host-device sync, eliminate syncs, CPU stall, non_blocking, set_sync_debug_mode, cudaStreamSynchronize, cudaEventSynchronize, remove syncs, async GPU.
- Grade
- C
- Quality
- 78 / 100
- Safety
- Passed
- GitHub stars
- 14,127
Placed #4 of 6 in the judged category final for MLOps & model serving.
Our two independent classifiers differ on this skill's category, so treat the placement as provisional.
This rating describes the version of the repository we crawled; it may have changed since, so go by its current files.
Our review
- Architecture
- 84 / 100
- Maturity
- Packaged: licensed, with a version or references
- Security scan
- No issues found
- License
- Not declared. Check the repository before reusing it.
- Adoption
- Adopted by 2 other repositories, in 1 collections
- Scope
- Vendor-specific: built around one product. Guidance on eliminating syncs in public PyTorch code.
Judge review
Placed #4 of 6 in the judged category final for MLOps & model serving.
- Trigger
95 95 - Instructions
88 88 - Specificity
92 92 - Safety
80 80 - Disclosure
68 68 - Maintenance
82 82
Sharp false-vs-true dependency framework with real before/after code and an explicit "do not use for" section handing off five adjacent scenarios to sibling skills.
Not checked yet
- No blind task eval yet
- Sibling files (scripts, references) not fetched; security scan covers SKILL.md only
Skillspace indexes and rates this skill; it does not host or endorse it. Run third-party skills at your own risk.