DefTruth / ffpa-attn-mma
📚FFPA: Yet another Faster Flash Prefill Attention with O(1)⚡️SRAM complexity for headdim > 256, 1.8x~3x↑🎉faster than SDPA EA.
☆106Updated this week
Alternatives and similar repositories for ffpa-attn-mma:
Users that are interested in ffpa-attn-mma are comparing it to the libraries listed below
- Implement Flash Attention using Cute.☆69Updated 2 months ago
- ⚡️Write HGEMM from scratch using Tensor Cores with WMMA, MMA and CuTe API, Achieve Peak⚡️ Performance.☆52Updated 2 weeks ago
- ☆67Updated 2 months ago
- Standalone Flash Attention v2 kernel without libtorch dependency☆104Updated 5 months ago
- ☆81Updated 5 months ago
- 使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention☆53Updated 6 months ago
- We invite you to visit and follow our new repository at https://github.com/microsoft/TileFusion. TiledCUDA is a highly efficient kernel …☆175Updated 3 weeks ago
- Decoding Attention is specially optimized for multi head attention (MHA) using CUDA core for the decoding stage of LLM inference.☆29Updated 3 months ago
- ☆101Updated 6 months ago
- flash attention tutorial written in python, triton, cuda, cutlass☆260Updated last month
- ☆59Updated 3 months ago
- ☆61Updated 3 weeks ago
- QQQ is an innovative and hardware-optimized W4A8 quantization solution for LLMs.☆101Updated this week
- Performance of the C++ interface of flash attention and flash attention v2 in large language model (LLM) inference scenarios.☆35Updated 5 months ago
- ☆36Updated last month
- A sparse attention kernel supporting mix sparse patterns☆133Updated last week
- A standalone GEMM kernel for fp16 activation and quantized weight, extracted from FasterTransformer☆88Updated 11 months ago
- Quantized Attention on GPU☆34Updated 2 months ago
- Examples of CUDA implementations by Cutlass CuTe☆138Updated 2 weeks ago
- High-speed GEMV kernels, at most 2.7x speedup compared to pytorch baseline.☆97Updated 7 months ago
- TileFusion is a highly efficient kernel template library designed to elevate the level of abstraction in CUDA C for processing tiles.☆56Updated this week
- ☆44Updated last month
- PyTorch bindings for CUTLASS grouped GEMM.☆64Updated 3 months ago
- Several optimization methods of half-precision general matrix vector multiplication (HGEMV) using CUDA core.☆55Updated 5 months ago
- Puzzles for learning Triton, play it with minimal environment configuration!☆229Updated 2 months ago
- GPTQ inference TVM kernel☆38Updated 9 months ago
- Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.☆89Updated this week
- Odysseus: Playground of LLM Sequence Parallelism☆64Updated 8 months ago
- An efficient GPU support for LLM inference with x-bit quantization (e.g. FP6,FP5).☆234Updated 3 months ago
- Dynamic Memory Management for Serving LLMs without PagedAttention☆290Updated this week