tile-ai / tilelang
Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels
☆971Updated this week
Alternatives and similar repositories for tilelang:
Users that are interested in tilelang are comparing it to the libraries listed below
- Mirage: Automatically Generating Fast GPU Kernels without Programming in Triton/CUDA☆798Updated this week
- Distributed Triton for Parallel Systems☆415Updated last week
- BitBLAS is a library to support mixed-precision matrix multiplications, especially for quantized LLM deployment.☆579Updated 2 months ago
- Puzzles for learning Triton, play it with minimal environment configuration!☆281Updated 4 months ago
- depyf is a tool to help you understand and adapt to PyTorch compiler torch.compile.☆628Updated 4 months ago
- [MLSys'25] QServe: W4A8KV4 Quantization and System Co-design for Efficient LLM Serving; [MLSys'25] LServe: Efficient Long-sequence LLM Se…☆639Updated last month
- A throughput-oriented high-performance serving framework for LLMs☆794Updated 6 months ago
- FlashInfer: Kernel Library for LLM Serving☆2,659Updated last week
- FlagGems is an operator library for large language models implemented in Triton Language.☆488Updated this week
- Dynamic Memory Management for Serving LLMs without PagedAttention☆347Updated 3 weeks ago
- A Easy-to-understand TensorOp Matmul Tutorial☆340Updated 6 months ago
- flash attention tutorial written in python, triton, cuda, cutlass☆325Updated 3 months ago
- A fast communication-overlapping library for tensor/expert parallelism on GPUs.☆864Updated 2 weeks ago
- Fast CUDA matrix multiplication from scratch☆689Updated last year
- Fastest kernels written from scratch☆223Updated 2 weeks ago
- A collection of memory efficient attention operators implemented in the Triton language.☆262Updated 10 months ago
- A PyTorch Native LLM Training Framework☆783Updated 3 months ago
- Ring attention implementation with flash attention☆734Updated last week
- Zero Bubble Pipeline Parallelism☆382Updated last week
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.☆529Updated last month
- Several optimization methods of half-precision general matrix multiplication (HGEMM) using tensor core with WMMA API and MMA PTX instruct…☆387Updated 7 months ago
- Analyze the inference of Large Language Models (LLMs). Analyze aspects like computation, storage, transmission, and hardware roofline mod…☆438Updated 7 months ago
- Flash Attention in ~100 lines of CUDA (forward pass only)☆779Updated 3 months ago
- USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference☆473Updated this week
- Tile primitives for speedy kernels☆2,259Updated this week
- Disaggregated serving system for Large Language Models (LLMs).☆550Updated last week
- KernelBench: Can LLMs Write GPU Kernels? - Benchmark with Torch -> CUDA problems☆263Updated last week
- LLM KV cache compression made easy☆454Updated 3 weeks ago
- Step-by-step optimization of CUDA SGEMM☆308Updated 3 years ago
- FP16xINT4 LLM inference kernel that can achieve near-ideal ~4x speedups up to medium batchsizes of 16-32 tokens.☆797Updated 7 months ago