DD-DuDa / BitDecodingLinks
A GPU-optimized system for efficient long-context LLMs decoding with low-bit KV cache.
☆56Updated 2 weeks ago
Alternatives and similar repositories for BitDecoding
Users that are interested in BitDecoding are comparing it to the libraries listed below
Sorting:
- SpInfer: Leveraging Low-Level Sparsity for Efficient Large Language Model Inference on GPUs☆52Updated 4 months ago
- Implement Flash Attention using Cute.☆94Updated 8 months ago
- ☆98Updated 3 months ago
- ☆61Updated 3 months ago
- ☆81Updated 7 months ago
- A lightweight design for computation-communication overlap.☆155Updated last week
- QuTLASS: CUTLASS-Powered Quantized BLAS for Deep Learning☆67Updated last month
- ⚡️Write HGEMM from scratch using Tensor Cores with WMMA, MMA and CuTe API, Achieve Peak⚡️ Performance.☆100Updated 3 months ago
- Tile-based language built for AI computation across all scales☆46Updated this week
- [DAC'25] Official implement of "HybriMoE: Hybrid CPU-GPU Scheduling and Cache Management for Efficient MoE Inference"☆66Updated 2 months ago
- ☆27Updated 4 months ago
- ☆97Updated 11 months ago
- ☆150Updated last year
- DeeperGEMM: crazy optimized version☆71Updated 3 months ago
- Decoding Attention is specially optimized for MHA, MQA, GQA and MLA using CUDA core for the decoding stage of LLM inference.☆40Updated 2 months ago
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆94Updated last month
- [COLM 2024] SKVQ: Sliding-window Key and Value Cache Quantization for Large Language Models☆24Updated 10 months ago
- Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity☆217Updated last year
- We invite you to visit and follow our new repository at https://github.com/microsoft/TileFusion. TiledCUDA is a highly efficient kernel …☆183Updated 6 months ago
- A Vectorized N:M Format for Unleashing the Power of Sparse Tensor Cores☆52Updated last year
- ☆39Updated last year
- Several optimization methods of half-precision general matrix vector multiplication (HGEMV) using CUDA core.☆63Updated 11 months ago
- ☆53Updated 2 months ago
- 使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention☆74Updated last year
- Optimize GEMM with tensorcore step by step☆32Updated last year
- ☆54Updated last year
- ☆23Updated this week
- High performance Transformer implementation in C++.☆129Updated 7 months ago
- A standalone GEMM kernel for fp16 activation and quantized weight, extracted from FasterTransformer☆94Updated 2 weeks ago
- PyTorch bindings for CUTLASS grouped GEMM.☆109Updated 2 months ago