From-scratch Rust+CUDA inference engine, bit-exact by construction — NVFP4, MoE, MTP speculative decoding, tuned against measured limits of one RTX 5090 Laptop (sm_120a).
-
Updated
Jul 28, 2026 - Rust
#
gpu-kernels
Here are 77 public repositories matching this topic...
[DEPRECATED] Moved to ROCm/rocm-systems repo
-
Updated
May 28, 2026 - Python
AgentKernelArena provides an end-to-end siloed-benchmarking environment where different LLM-powered agents—such as Cursor Agent, Claude Code, Codex, SWE-agent, and GEAK—can be evaluated side-by-side on the same GPU kernel tasks, using objective and reproducible metrics.
-
Updated
Jul 29, 2026 - Python
Online CUDA Occupancy Calculator
-
Updated
Oct 12, 2021 - CoffeeScript
Agents, and RL environment, for optimizing GPU kernels on AMD ROCm using LLM agents. Benchmarks LLM serving workloads end-to-end, profiles bottleneck kernels, optimizes them via Claude Code or Codex, and scores on compilation, correctness, and speedup.
-
Updated
Jul 28, 2026 - Python
(Spring 2017) Assignment 2: GPU Executor
-
Updated
Apr 27, 2017 - Python
Open source skill library for AI coding agents to write, optimize, and debug high performance compute kernels across CUDA, Triton, and quantized workloads.
cuda high-performance-computing triton quantization rocm gpu-kernels prompt-engineering llm-agents ai-coding kernel-optimization
-
Updated
Jun 21, 2026 - TypeScript
Runs a single CUDA/OpenCL kernel, taking its source from a file and arguments from the command-line
gpu opencl cuda runner gpgpu multi-language performance-analysis profiling performance-testing debugging-tool gpu-kernels gpu-kernel-performance
-
Updated
Jun 10, 2026 - C++
GPU Drano Static Analysis for GPU programs.
-
Updated
Nov 16, 2018 - C++
Minimal FlashAttention in CUDA C++/CuTe: readable WMMA/CuTe kernels, no NxN workspace, up to 4.5x faster than naive PyTorch
cuda attention cutlass cute gpu-kernels pytorch-extension tensor-cores llm flash-attention flashattention wmma
-
Updated
Jun 2, 2026 - Cuda
The official implementation for paper "AdaExplore: Failure-Driven Adaptation and Diversity-Preserving Search for Efficient Kernel Generation"
-
Updated
Jul 12, 2026 - Python
Prototype for a SPIR-V assembler and dissasembler. It provides a composable Java interface for generating SPIR-V code at runtime.
-
Updated
Oct 31, 2025 - Java
Repo containing artifacts for Neurips 2025 tutorial- How to Build Agents to Generate Kernels for Faster LLMs (and Other Models!)
-
Updated
May 11, 2026 - Jupyter Notebook
Apple Silicon (Metal) backend for OpenAI Triton: write standard @triton.jit kernels on your Mac GPU. The same source runs on NVIDIA (fp32 verified bit-identical), so you develop kernel logic locally and rent a GPU only for the perf pass.
-
Updated
Jul 15, 2026 - Python
From-scratch reimplementation of DeepSeek's Native Sparse Attention (arXiv:2502.11089) in Triton + CUDA Hopper WGMMA. 7.07x faster than FlashAttention-3 at 64k context. Five-model training fleet, perplexity sweep, LongBench v2, MoBA comparison.
cuda pytorch transformer triton hopper attention-mechanism nsa gpu-kernels sparse-attention llm long-context flash-attention deepseek native-sparse-attention wgmma
-
Updated
Jun 21, 2026 - Python
Native Blackwell (sm_100) tcgen05 training backward for the gated-linear-recurrence family (GDN-2/GLA/KDA/SSD), plus a contract-grade verifier that falsifies published GPU kernels. Six open Mamba-3 Triton kernels including the first open MIMO backward; GRPO trainer graded on contract correctness. state-spaces/mamba#904
cuda triton cutlass mamba gpu-kernels state-space-models blackwell mamba-3 tcgen05 kernel-verification
-
Updated
Jul 25, 2026 - Python
Agent-queryable ROCm kernel optimization knowledge base for AMD Instinct MI300/gfx942 and MI350/MI355X/gfx950, packaged for Codex CLI and Claude Code with merged-PR provenance, real-silicon validation, and a maintainer-controlled pull-request evidence pipeline.
-
Updated
Jul 27, 2026 - Python
An optimized, high-performance NF4 (NormalFloat 4-bit) dequantization Triton GPU kernel achieving up to 1.41x speedup over bitsandbytes.
-
Updated
Jul 14, 2026 - Python
ORDA research Triton kernel for logit-level forward-KL distillation with a fused Cross Entropy substrate.
-
Updated
Jul 5, 2026 - Python
LLM inference kernels from scratch in Triton: KV cache, FlashAttention, PagedAttention, RMSNorm, RoPE, SwiGLU, and benchmarks.
cuda pytorch triton gpu-kernels machine-learning-systems inference-optimization kv-cache llm-inference pagedattention flashattention
-
Updated
Jun 18, 2026 - Python
Improve this page
Add a description, image, and links to the gpu-kernels topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the gpu-kernels topic, visit your repo's landing page and select "manage topics."