A plug-and-play compiler that delivers free-lunch optimizations for both inference and training.
☆324Jul 31, 2026Updated this week
Alternatives and similar repositories for MagiCompiler
Users that are interested in MagiCompiler are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Training☆893Updated this week
- FA4-based Relative Attention Kernel developed by TML and Colfax☆17Jul 17, 2026Updated 2 weeks ago
- ☆2,110Apr 11, 2026Updated 3 months ago
- ☆84Feb 5, 2026Updated 5 months ago
- NVSHMEM‑Tutorial: Build a DeepEP‑like GPU Buffer☆194Feb 11, 2026Updated 5 months ago
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- A Triton-only attention backend for vLLM☆27Jul 14, 2026Updated 2 weeks ago
- Code repository for the SOSP'25 paper DCP: Addressing Input Dynamism In Long-Context Training via Dynamic Context Parallelism.☆22Nov 28, 2025Updated 8 months ago
- CUDA kernels for linear attention variants, written in CuTe DSL and CUTLASS C++.☆535Updated this week
- VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo☆2,123Updated this week
- Distributed Compiler based on Triton for Parallel Systems☆1,507Jul 20, 2026Updated 2 weeks ago
- flex-block-attn: an efficient block sparse attention computation library☆129Dec 26, 2025Updated 7 months ago
- Tile-based language built for AI computation across all scales☆183Updated this week
- NVIDIA cuTile learn☆169Dec 9, 2025Updated 7 months ago
- A Quirky Assortment of CuTe Kernels☆1,084Updated this week
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference☆684May 21, 2026Updated 2 months ago
- [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.☆1,021Feb 25, 2026Updated 5 months ago
- ☆57Updated this week
- ML kernels and benchmarking infrastructure written in TIRx☆78Updated this week
- NCU-driven iterative optimization workflow for CUDA/CUTLASS/Triton/CuTe DSL kernels.☆24Apr 10, 2026Updated 3 months ago
- mKernel: fast multi-node, multi-GPU fused kernels☆260Updated this week
- Official implementation of paper "VMoBA: Mixture-of-Block Attention for Video Diffusion Models"☆64Jul 1, 2025Updated last year
- A lightweight design for computation-communication overlap.☆243Jan 20, 2026Updated 6 months ago
- FLA but cuTile☆27Apr 17, 2026Updated 3 months ago
- GPU virtual machines on DigitalOcean Gradient AI • AdGet to production fast with high-performance AMD and NVIDIA GPUs you can spin up in seconds. The definition of operational simplicity.
- ☆37Aug 7, 2025Updated 11 months ago
- Mirage Persistent Kernel: Compiling LLMs into a MegaKernel☆2,404Jul 28, 2026Updated last week
- Accelerating MoE with IO and Tile-aware Optimizations☆736Jul 4, 2026Updated last month
- A benchmark of real-world DL kernel problems☆269Jul 15, 2026Updated 2 weeks ago
- 🎬 3.7× faster video generation E2E 🖼️ 1.6× faster image generation E2E ⚡ ColumnSparseAttn 9.3× vs FlashAttn‑3 💨 ColumnSparseGEMM 2.5× …☆111Sep 8, 2025Updated 10 months ago
- [ICML2025, NeurIPS2025 Spotlight] Sparse VideoGen 1 & 2: Accelerating Video Diffusion Transformers with Sparse Attention☆699Jul 4, 2026Updated last month
- High performance RMSNorm Implement by using SM Core Storage(Registers and Shared Memory)☆30Jan 22, 2026Updated 6 months ago
- Accelerate LLM preference tuning via prefix sharing with a single line of code☆52Jul 4, 2025Updated last year
- ☆32Jul 2, 2025Updated last year
- Managed Kubernetes at scale on DigitalOcean • AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- A fast communication-overlapping library for tensor/expert parallelism on GPUs.☆1,352Aug 28, 2025Updated 11 months ago
- High-performance LLM operator library built on TileLang.☆167Updated this week
- Low overhead tracing library and trace visualizer for pipelined CUDA kernels☆135Updated this week
- FlashTile is a CUDA Tile IR compiler that is compatible with NVIDIA's tileiras, targeting SM70 through SM121 NVIDIA GPUs.☆61Feb 6, 2026Updated 5 months ago
- An experimental communicating attention kernel based on DeepEP.☆34Jul 29, 2025Updated last year
- A unified inference and post-training framework for accelerated video generation.☆3,911Updated this week
- Benchmark tests supporting the TiledCUDA library.☆19Nov 19, 2024Updated last year