Infrastructure●●Solid
Continuous Nvidia CUDA PC Sampling Profiler
Production-ready CUDA profiling when NSight only works in development.
Big Brain
gnurizen
1469d ago
Back to browse
GitHub Repository
Graphsignal Profiler
207 starsPython
CUDA Profiler for Production Inference
by npgraph·Jun 23, 2026·6 points·0 comments
AI Analysis
●●SolidSolve My ProblemSlick
LLM inference profiling with per-token timing, but Arize and Langfuse already own this space.
Strengths
- •vLLM integration via CLI wrapper means zero code changes to existing deployments
- •Per-step LLM generation tracing with token throughput breakdowns is genuinely useful
- •CUDA 12/13 support keeps pace with current GPU hardware
Weaknesses
- •Cloud-only with API key requirement limits self-hosted and air-gapped deployments
- •LLM observability is crowded with Arize, Langfuse, Datadog already well-funded
Category
Target Audience
ML engineers running production LLM inference
Similar To
Arize · Langfuse · PyTorch Profiler
Similar Projects
Infrastructure●●Solid
Go LLM inference with a Vulkan GPU back end that beats Ollama's CUDA
28% faster Vulkan-to-CUDA on Qwen, but llm.c and llama.cpp already own inference.
WizardryBig BrainNiche Gem
computerex
103mo ago
Education●●Solid
Tiny-vLLM – high performance LLM inference engine in C++ and CUDA
Build vLLM from scratch with PagedAttention kernels when llama.cpp already exists.
Big BrainNiche Gem
yu3zhou4
2051826d ago
AI/ML●Mid
Reduce TTFT via Streaming to an LLM
Academic paper on TTFT optimization with no implementation to evaluate.
Big Brain
rajveerb
102mo ago
AI/ML●Mid
I reduced LLM inference GPU calls by 94% using semantic routing
94% GPU reduction claim needs verifiable benchmarks to stand out.
Bold BetShip It
kanacki
2122d ago
Infrastructure●●Solid
Piqc – GPU waste scanner for LLM inference clusters
Read-only GPU waste scanner finds 20-40% cluster spend waste without agents or sidecars.
Solve My ProblemSlick
paralleliq
3022d ago