Repeerc / flash-attention-v2-RDNA3-minimal
a simple Flash Attention v2 implementation with ROCM (RDNA3 GPU, roc wmma), mainly used for stable diffusion(ComfyUI) in Windows ZLUDA environments.
☆23Updated 2 months ago
Related projects ⓘ
Alternatives and complementary repositories for flash-attention-v2-RDNA3-minimal
- [WIP] Context parallel attention that works with torch.compile☆49Updated this week
- Development repository for the Triton language and compiler☆93Updated this week
- (WIP) Parallel inference for black-forest-labs' FLUX model.☆11Updated this week
- Standalone Flash Attention v2 kernel without libtorch dependency☆98Updated 2 months ago
- Model Compression Toolbox for Large Language Models and Diffusion Models☆231Updated last week
- Fast and memory-efficient exact attention☆139Updated this week
- A high-throughput and memory-efficient inference and serving engine for LLMs☆45Updated this week
- High-speed GEMV kernels, at most 2.7x speedup compared to pytorch baseline.☆90Updated 4 months ago
- AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (N…☆11Updated 4 months ago
- A standalone GEMM kernel for fp16 activation and quantized weight, extracted from FasterTransformer☆85Updated 8 months ago
- ☆67Updated last week
- rocWMMA☆92Updated this week
- ☆35Updated 2 weeks ago
- QQQ is an innovative and hardware-optimized W4A8 quantization solution for LLMs.☆89Updated last month
- ☆152Updated last week
- ☆47Updated 2 months ago
- Ahead of Time (AOT) Triton Math Library☆41Updated this week
- SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models☆327Updated this week
- Composable Kernel: Performance Portable Programming Model for Machine Learning Tensor Operators☆313Updated this week
- A parallelism VAE avoids OOM for high resolution image generation☆40Updated last month
- llama INT4 cuda inference with AWQ☆48Updated 4 months ago
- ☆123Updated 11 months ago
- 使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention☆52Updated 3 months ago
- [ECCV24] MixDQ: Memory-Efficient Few-Step Text-to-Image Diffusion Models with Metric-Decoupled Mixed Precision Quantization☆31Updated 2 months ago
- ☆79Updated 2 months ago
- ☆48Updated this week
- Official PyTorch implementation of FlatQuant: Flatness Matters for LLM Quantization☆63Updated last week
- A CUDA kernel for NHWC GroupNorm for PyTorch☆15Updated this week
- TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.☆156Updated this week
- ☆32Updated last month