high performance computing on graphics processing units: hgpu.org

Changxin Ke, Rui Zhang, Shuo Wang, Li Ding, Guangli Li, Yuanbo Wen, Shuoming Zhang, Ruiyuan Xu, Jin Qin, Jiaming Guo, Chenxi Wang, Ling Li, Qi Guo, Yunji Chen

View

Download (PDF)

Source codes

Tags: Code generation, Computer science, CUDA, HPC, LLM, Machine learning, nVidia, nVidia A100, nVidia GeForce RTX 4090, Package

July 13, 2025 by hgpu

KIS-S: A GPU-Aware Kubernetes Inference Simulator with RL-Based Auto-Scaling

Guilin Zhang, Wulan Guo, Ziqi Tan, Qiang Guan, Hailong Jiang

View

Download (PDF)

Source codes

Tags: Computer science, nVidia, nVidia GeForce RTX 3080, Package, Task scheduling

July 13, 2025 by hgpu

Demystifying NCCL: An In-depth Analysis of GPU Communication Protocols and Algorithms

Zhiyi Hu, Siyuan Shen, Tommaso Bonato, Sylvain Jeaugey, Cedell Alexander, Eric Spada, Jeff Hammond, Torsten Hoefler

View

Download (PDF)

Tags: Computer science, CUDA, GPU cluster, Network communication, nVidia, nVidia GH200

July 13, 2025 by hgpu

AutoKernel: Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels

AutoKernel: Autonomous GPU Kernel Optimization via Iterative Agent-Driven Search

Triton-Sanitizer: A Fast and Device-Agnostic Memory Sanitizer for Triton with Rich Diagnostic Context

LLM.Q: Quantized LLM training in pure CUDA/C++

LLMQ: Efficient Lower-Precision LLM Training for Consumer GPUs

SOL-ExecBench: Speed-of-Light Benchmarking for Real-World GPU Kernels Against Hardware Limits

True 4-Bit Quantized CNN Training on CPU

True 4-Bit Quantized Convolutional Neural Network Training on CPU: Achieving Full-Precision Parity

cuFuzz: A GPU-oriented coverage-guided fuzzer for userland CUDA application

Hunting CUDA Bugs at Scale with cuFuzz

KernelSkill: A Multi-Agent Framework for GPU Kernel Optimization

MSKernelBench & CUDAMaster

Making LLMs Optimize Multi-Scenario CUDA Kernels Like Experts

EvoScientist: Harness Vibe Research with Self-evolving AI Scientists

EvoScientist: Towards Multi-Agent Evolving AI Scientists for End-to-End Scientific Discovery

RepoLaunch: Automating Build and Test Pipeline of Code Repositories on ANY Language and ANY Platform

RepoLaunch: Automating Build & Test Pipeline of Code Repositories on ANY Language and ANY Platform

See all packages

* * *

high performance computing on graphics processing units: hgpu.org

ConTraPh: Contrastive Learning for Parallelization and Performance Optimization

SIGMo: High-Throughput Batched Subgraph Isomorphism on GPUs for Molecular Matching

DGEMM without FP64 Arithmetic – using FP64 Emulation and FP8 Tensor Cores with Ozaki Scheme

OpenDwarfs 2025: Modernizing the OpenDwarfs Benchmark Suite for Heterogeneous Computing

Performance Portable Gradient Computations Using Source Transformation

Kevin: Multi-Turn RL for Generating CUDA Kernels

Dissecting the NVIDIA Blackwell Architecture with Microbenchmarks

Specx: a C++ task-based runtime system for heterogeneous distributed architectures

Pre-Training LLMs on a budget: A comparison of three optimizers

Mutual-Supervised Learning for Sequential-to-Parallel Code Translation

KIS-S: A GPU-Aware Kubernetes Inference Simulator with RL-Based Auto-Scaling

Demystifying NCCL: An In-depth Analysis of GPU Communication Protocols and Algorithms

Recent source codes

AutoKernel: Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels

Triton-Sanitizer: A Fast and Device-Agnostic Memory Sanitizer for Triton with Rich Diagnostic Context

LLM.Q: Quantized LLM training in pure CUDA/C++

SOL-ExecBench: Speed-of-Light Benchmarking for Real-World GPU Kernels Against Hardware Limits

True 4-Bit Quantized CNN Training on CPU

cuFuzz: A GPU-oriented coverage-guided fuzzer for userland CUDA application

KernelSkill: A Multi-Agent Framework for GPU Kernel Optimization

MSKernelBench & CUDAMaster

EvoScientist: Harness Vibe Research with Self-evolving AI Scientists

RepoLaunch: Automating Build and Test Pipeline of Code Repositories on ANY Language and ANY Platform

Most viewed papers (last 30 days)