weishengying / tiny-flash-attention
使用 cutlass 实现 flash-attention 精简版,具有教学意义
☆29Updated last month
Related projects: ⓘ
- 使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention☆46Updated last month
- Performance of the C++ interface of flash attention and flash attention v2 in large language model (LLM) inference scenarios.☆20Updated last week
- ☆32Updated 3 months ago
- ☆77Updated last year
- A standalone GEMM kernel for fp16 activation and quantized weight, extracted from FasterTransformer☆82Updated 6 months ago
- play gemm with tvm☆81Updated last year
- Several optimization methods of half-precision general matrix vector multiplication (HGEMV) using CUDA core.☆40Updated last week
- ☆70Updated 6 months ago
- ☆140Updated 4 months ago
- ☆90Updated 6 months ago
- ☆133Updated 2 months ago
- ☆67Updated last week
- Standalone Flash Attention v2 kernel without libtorch dependency☆93Updated last week
- ☆56Updated this week
- flash attention tutorial written in python, triton, cuda, cutlass☆159Updated 3 months ago
- ☆15Updated last week
- ☆92Updated 3 years ago
- ☆48Updated 2 years ago
- TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.☆114Updated last week
- Summary of some awesome work for optimizing LLM inference☆26Updated this week
- ☆23Updated this week
- Tutorials for writing high-performance GPU operators in AI frameworks.☆118Updated last year
- ☆18Updated 5 months ago
- ☆100Updated 5 months ago
- ☆17Updated 5 months ago
- Optimize GEMM with tensorcore step by step☆11Updated 9 months ago
- code reading for tvm☆69Updated 2 years ago
- CUDA 6大并行计算模式 代码与笔记☆57Updated 4 years ago
- ☆134Updated last year
- This is a demo how to write a high performance convolution run on apple silicon☆52Updated 2 years ago