microsoft / ark
A GPU-driven system framework for scalable AI applications
☆114Updated 2 months ago
Alternatives and similar repositories for ark:
Users that are interested in ark are comparing it to the libraries listed below
- MSCCL++: A GPU-driven communication stack for scalable AI applications☆331Updated this week
- FlexFlow Serve: Low-Latency, High-Performance LLM Serving☆34Updated this week
- Microsoft Collective Communication Library☆65Updated 4 months ago
- DeepSeek-V3/R1 inference performance simulator☆106Updated 3 weeks ago
- An experimental parallel training platform☆54Updated last year
- We invite you to visit and follow our new repository at https://github.com/microsoft/TileFusion. TiledCUDA is a highly efficient kernel …☆181Updated 2 months ago
- Automated Parallelization System and Infrastructure for Multiple Ecosystems☆79Updated 4 months ago
- NCCL Profiling Kit☆129Updated 9 months ago
- An extension library of WMMA API (Tensor Core API)☆95Updated 9 months ago
- ☆76Updated 5 months ago
- Perplexity GPU Kernels☆204Updated last week
- ☆47Updated last week
- [OSDI'24] Serving LLM-based Applications Efficiently with Semantic Variable☆153Updated 6 months ago
- Paella: Low-latency Model Serving with Virtualized GPU Scheduling☆58Updated 11 months ago
- Microsoft Collective Communication Library☆342Updated last year
- ☆68Updated 3 weeks ago
- ☆26Updated last year
- Thunder Research Group's Collective Communication Library☆36Updated 11 months ago
- ☆78Updated 2 years ago
- High performance Transformer implementation in C++.☆116Updated 3 months ago
- ☆66Updated 3 months ago
- Synthesizer for optimal collective communication algorithms☆105Updated last year
- RCCL Performance Benchmark Tests☆63Updated last week
- ☆23Updated 2 months ago
- Artifact of OSDI '24 paper, ”Llumnix: Dynamic Scheduling for Large Language Model Serving“☆61Updated 10 months ago
- nnScaler: Compiling DNN models for Parallel Training☆106Updated 2 months ago
- ☆97Updated last month
- PArametrized Recommendation and Ai Model benchmark is a repository for development of numerous uBenchmarks as well as end to end nets for…☆134Updated last week
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆80Updated last week
- Dynamic Memory Management for Serving LLMs without PagedAttention☆347Updated 3 weeks ago