mirage-project / mirage
Mirage: Automatically Generating Fast GPU Kernels without Programming in Triton/CUDA
☆636Updated this week
Related projects ⓘ
Alternatives and complementary repositories for mirage
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆420Updated this week
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.☆483Updated 3 weeks ago
- QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving☆443Updated last week
- Flash Attention in ~100 lines of CUDA (forward pass only)☆626Updated 7 months ago
- FlashInfer: Kernel Library for LLM Serving☆1,452Updated this week
- FlagGems is an operator library for large language models implemented in Triton Language.☆342Updated this week
- depyf is a tool to help you understand and adapt to PyTorch compiler torch.compile.☆500Updated 2 weeks ago
- Ring attention implementation with flash attention☆585Updated last week
- A Easy-to-understand TensorOp Matmul Tutorial☆290Updated 2 months ago
- A fast communication-overlapping library for tensor parallelism on GPUs.☆224Updated 3 weeks ago
- Applied AI experiments and examples for PyTorch☆166Updated 2 weeks ago
- A throughput-oriented high-performance serving framework for LLMs☆636Updated 2 months ago
- This repository contains the experimental PyTorch native float8 training UX☆211Updated 3 months ago
- A collection of memory efficient attention operators implemented in the Triton language.☆219Updated 5 months ago
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆624Updated 2 months ago
- Efficient implementations of state-of-the-art linear attention models in Pytorch and Triton☆1,339Updated this week
- Simple and fast low-bit matmul kernels in CUDA / Triton☆143Updated this week
- flash attention tutorial written in python, triton, cuda, cutlass☆202Updated 5 months ago
- Fast CUDA matrix multiplication from scratch☆479Updated 10 months ago
- ☆152Updated this week
- Tile primitives for speedy kernels☆1,658Updated this week
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆278Updated 4 months ago
- Fast Matrix Multiplications for Lookup Table-Quantized LLMs☆187Updated this week
- Cataloging released Triton kernels.☆134Updated 2 months ago
- Code for Neurips24 paper: QuaRot, an end-to-end 4-bit inference of large language models.☆284Updated 3 months ago
- ☆167Updated 4 months ago
- Step-by-step optimization of CUDA SGEMM☆240Updated 2 years ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆238Updated last week
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆305Updated 3 months ago
- Several optimization methods of half-precision general matrix multiplication (HGEMM) using tensor core with WMMA API and MMA PTX instruct…☆302Updated 2 months ago