high performance computing on graphics processing units: hgpu.org

hgpu.org » nVidia GB200

Hardware-Aware FP4 FlashAttention-4

Robert Hu

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, nVidia, nVidia B300, nVidia GB200, Package, PTX

September 14, 2026 by hgpu

A Contract-Grade Verifier for LLM-Generated GPU Kernels, and a Native Blackwell Backward for the Gated-Linear-Recurrence Family

Rishi Shah, Rishav Shrestha

View

Download (PDF)

Tags: Computer science, LLM, nVidia, nVidia B200, nVidia GB200, Triton

August 17, 2026 by hgpu

Practical FP4 Training for Large-Scale MoE Models on Hopper GPUs

Wuyue Zhang, Chongdong Huang, Chunbo You, Cheng Gu, Fengjuan Wang, Mou Sun

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, LLM, nVidia, nVidia GB200, Package, Precision

March 8, 2026 by hgpu

PTXBench: an open-source environment for kernel agents with architecture-specific CUDA/PTX

PTXBench: Benchmark and Adapt LLMs for GPU Kernel Optimization with Architecture-specific PTX

RealisticTritonBench: A Benchmark for Triton-Kernel Generation in Real-World AI Frameworks

TensorCorePTX: PTX Tensor Core GEMM Exploration

Hand-Written PTX Tensor-Core GEMM Kernels: A Multi-Precision Study on NVIDIA L4

FlashPDE: Triton-fused PDE stencil kernels for physics-informed neural networks

FlashPDE: A Drop-In Fused Triton Operator Library for Neural PDE Solvers

PortLBM: Portable GPU-accelerated Lattice Boltzmann Method Simulations

PortLBM: A Portable Lattice Boltzmann Tool Leveraging SYCL on AMD, NVIDIA, and Intel GPUs

Harness Engineering for LLM-Driven GPU Kernel Generation

CANNBot Skills

Compiler-Grounded Hierarchical Diagnosis for LLM-Based Triton Kernel Optimization

See all packages

* * *

high performance computing on graphics processing units: hgpu.org

Hardware-Aware FP4 FlashAttention-4

A Contract-Grade Verifier for LLM-Generated GPU Kernels, and a Native Blackwell Backward for the Gated-Linear-Recurrence Family

Practical FP4 Training for Large-Scale MoE Models on Hopper GPUs

Recent source codes

FP4 FlashAttention forward, backward, and TorchTitan reproduction code

EinsummableV3 - graph-based computation with tensor operations support

MaxKernel: Intelligent TPU Kernel Engineering Agents

PTXBench: an open-source environment for kernel agents with architecture-specific CUDA/PTX

RealisticTritonBench: A Benchmark for Triton-Kernel Generation in Real-World AI Frameworks

TensorCorePTX: PTX Tensor Core GEMM Exploration

FlashPDE: Triton-fused PDE stencil kernels for physics-informed neural networks

PortLBM: Portable GPU-accelerated Lattice Boltzmann Method Simulations

Harness Engineering for LLM-Driven GPU Kernel Generation

CANNBot Skills

Most viewed papers (last 30 days)