antgroup / DeepXTraceLinks
DeepXTrace is a lightweight tool for precisely diagnosing slow ranks in DeepEP-based environments.
☆54Updated last week
Alternatives and similar repositories for DeepXTrace
Users that are interested in DeepXTrace are comparing it to the libraries listed below
Sorting:
- NVSHMEM‑Tutorial: Build a DeepEP‑like GPU Buffer☆130Updated 2 weeks ago
- A lightweight design for computation-communication overlap.☆177Updated 2 weeks ago
- DeeperGEMM: crazy optimized version☆71Updated 4 months ago
- DLSlime: Flexible & Efficient Heterogeneous Transfer Toolkit☆66Updated this week
- A prefill & decode disaggregated LLM serving framework with shared GPU memory and fine-grained compute isolation.☆111Updated 4 months ago
- An experimental communicating attention kernel based on DeepEP.☆34Updated 2 months ago
- ☆64Updated 5 months ago
- ☆46Updated 9 months ago
- ☆57Updated 4 months ago
- Aims to implement dual-port and multi-qp solutions in deepEP ibrc transport☆63Updated 4 months ago
- ☆60Updated 8 months ago
- ⚡️Write HGEMM from scratch using Tensor Cores with WMMA, MMA and CuTe API, Achieve Peak⚡️ Performance.☆118Updated 4 months ago
- Tile-based language built for AI computation across all scales☆61Updated this week
- Pipeline Parallelism Emulation and Visualization☆67Updated 3 months ago
- ☆298Updated last week
- ☆95Updated 6 months ago
- ☆98Updated last year
- Automated Parallelization System and Infrastructure for Multiple Ecosystems☆80Updated 10 months ago
- ☆30Updated 3 months ago
- Debug print operator for cudagraph debugging☆13Updated last year
- kvcached: Elastic KV cache for dynamic GPU sharing and efficient multi-LLM inference.☆94Updated this week
- Efficient Compute-Communication Overlap for Distributed LLM Inference☆57Updated 3 weeks ago
- Microsoft Collective Communication Library☆66Updated 10 months ago
- ☆85Updated 6 months ago
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆97Updated 3 months ago
- High performance Transformer implementation in C++.☆134Updated 8 months ago
- gLLM: Global Balanced Pipeline Parallelism System for Distributed LLM Serving with Token Throttling☆41Updated this week
- NVIDIA NVSHMEM is a parallel programming interface for NVIDIA GPUs based on OpenSHMEM. NVSHMEM can significantly reduce multi-process com…☆324Updated 2 weeks ago
- Stateful LLM Serving☆85Updated 6 months ago
- Compare different hardware platforms via the Roofline Model for LLM inference tasks.☆115Updated last year