HGPU group
LightNews LightNews
banner
hgpu.bsky.social
HGPU group
@hgpu.bsky.social
87 followers 10 following 250 posts
High performance computing on graphics processing units (GPU): AMD, Nvidia, Intel, CUDA, OpenCL, OpenGL, HPC
Posts Replies Media Videos
hgpu.bsky.social
tritonBLAS: Triton-based Analytical Approach for GEMM Kernel Parameter Selection

#Triton #BLAS #GEMM #AMD #ROCm #HPC #Performance #Package

hgpu.org?p=30441
tritonBLAS: Triton-based Analytical Approach for GEMM Kernel Parameter Selection
We present tritonBLAS, a fast and deterministic analytical model that uses architectural parameters like the cache hierarchy, and relative code and data placement to generate performant GPU GEMM ke…
hgpu.org
December 7, 2025 at 9:32 PM
hgpu.bsky.social
Targeted Testing of Compiler Optimizations via Grammar-Level Composition Styles

#LLVM #Compilers

hgpu.org?p=30440
Targeted Testing of Compiler Optimizations via Grammar-Level Composition Styles
Ensuring the correctness of compiler optimizations is critical, but existing fuzzers struggle to test optimizations effectively. First, most fuzzers use optimization pipelines (heuristics-based, fi…
hgpu.org
December 7, 2025 at 9:32 PM
hgpu.bsky.social
Decoupled Triton: A Block-Level Decoupled Language for Writing and Exploring Efficient Machine-Learning Kernels

#Triton #Compilers #MachineLearning #ML #Thesis

hgpu.org?p=30439
Decoupled Triton: A Block-Level Decoupled Language for Writing and Exploring Efficient Machine-Learning Kernels
Machine-learning (ML) applications frequently utilize high-performance ML kernels to execute tensor operations like matrix product and softmax. An ML kernel can be decomposed into two components: t…
hgpu.org
December 7, 2025 at 9:31 PM
hgpu.bsky.social
hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware

#FPGA #HLS #MachineLearning #ML #DeepLearning #DL #Package

hgpu.org?p=30438
hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware
We present hls4ml, a free and open-source platform that translates machine learning (ML) models from modern deep learning frameworks into high-level synthesis (HLS) code that can be integrated into…
hgpu.org
December 7, 2025 at 9:30 PM
hgpu.bsky.social
Microbenchmarking NVIDIA’s Blackwell Architecture: An in-depth Architectural Analysis

#PTX #CUDA #Benchmarking #Blackwell #HPC

hgpu.org?p=30437
Microbenchmarking NVIDIA’s Blackwell Architecture: An in-depth Architectural Analysis
As GPU architectures rapidly evolve to meet the overcoming demands of exascale computing and machine learning, the performance implications of architectural innovations remain poorly understood acr…
hgpu.org
December 7, 2025 at 9:30 PM
hgpu.bsky.social
QiMeng-Kernel: Macro-Thinking Micro-Coding Paradigm for LLM-Based High-Performance GPU Kernel Generation

#Triton #CUDA #AI #CodeGeneration #LLM

hgpu.org?p=30413
QiMeng-Kernel: Macro-Thinking Micro-Coding Paradigm for LLM-Based High-Performance GPU Kernel Generation
Developing high-performance GPU kernels is critical for AI and scientific computing, but remains challenging due to its reliance on expert crafting and poor portability. While LLMs offer promise fo…
hgpu.org
November 30, 2025 at 7:12 PM
hgpu.bsky.social
KernelBand: Boosting LLM-based Kernel Optimization with a Hierarchical and Hardware-aware Multi-armed Bandit

#Triton #CUDA #LLM #CodeGeneration

hgpu.org?p=30412
KernelBand: Boosting LLM-based Kernel Optimization with a Hierarchical and Hardware-aware Multi-armed Bandit
High quality kernels are critical for reducing training and inference costs of Large Language Models (LLMs), yet they traditionally require significant expertise in hardware architecture and softwa…
hgpu.org
November 30, 2025 at 7:11 PM
hgpu.bsky.social
GPU-Initiated Networking for NCCL

#CUDA #NCCL

hgpu.org?p=30411
GPU-Initiated Networking for NCCL
Modern AI workloads, especially Mixture-of-Experts (MoE) architectures, increasingly demand low-latency, fine-grained GPU-to-GPU communication with device-side control. Traditional GPU communicatio…
hgpu.org
November 30, 2025 at 7:11 PM
hgpu.bsky.social
NVIDIA Nemotron Parse 1.1

#Nvidia #OCR #Package

hgpu.org?p=30410
NVIDIA Nemotron Parse 1.1
We introduce Nemotron-Parse-1.1, a lightweight document parsing and OCR model that advances the capabilities of its predecessor, Nemoretriever-Parse-1.0. Nemotron-Parse-1.1 delivers improved capabi…
hgpu.org
November 30, 2025 at 7:09 PM
hgpu.bsky.social
ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels

#CUDA #AI #Package

hgpu.org?p=30409
ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels
Inter-GPU communication has become a major bottleneck for modern AI workloads as models scale and improvements in hardware compute throughput outpace improvements in interconnect bandwidth. Existin…
hgpu.org
November 30, 2025 at 7:07 PM
hgpu.bsky.social
Iris: First-Class Multi-GPU Programming Experience in Triton

#Triton #HIP #CUDA #Package

hgpu.org?p=30375
Iris: First-Class Multi-GPU Programming Experience in Triton
Multi-GPU programming traditionally requires developers to navigate complex trade-offs between performance and programmability. High-performance implementations typically rely on low-level HIP/CUDA…
hgpu.org
November 23, 2025 at 5:54 PM
hgpu.bsky.social
ProofWright: Towards Agentic Formal Verification of CUDA

#CUDA #LLM #CodeGeneration

hgpu.org?p=30374
ProofWright: Towards Agentic Formal Verification of CUDA
Large Language Models (LLMs) are increasingly used to automatically generate optimized CUDA kernels, substantially improving developer productivity. However, despite rapid generation, these kernels…
hgpu.org
November 23, 2025 at 5:54 PM
hgpu.bsky.social
AIvailable: A Software-Defined Architecture for LLM-as-a-Service on Heterogeneous and Legacy GPUs

#ROCm #CUDA #LLM

hgpu.org?p=30373
AIvailable: A Software-Defined Architecture for LLM-as-a-Service on Heterogeneous and Legacy GPUs
The rise of Large Language Models (LLM) has increased the need for scalable, high-performance inference systems, yet most existing frameworks assume homogeneous, resource-rich hardware, often unrea…
hgpu.org
November 23, 2025 at 5:53 PM
hgpu.bsky.social
Inside VOLT: Designing an Open-Source GPU Compiler

#OpenCL #CUDA #Compilers

hgpu.org?p=30372
Inside VOLT: Designing an Open-Source GPU Compiler
Recent efforts in open-source GPU research are opening new avenues in a domain that has long been tightly coupled with a few commercial vendors. Emerging open GPU architectures define SIMT function…
hgpu.org
November 23, 2025 at 5:51 PM
hgpu.bsky.social
The Anatomy of a Triton Attention Kernel

#Triton #HIP #CUDA #LLM #Performance

hgpu.org?p=30371
The Anatomy of a Triton Attention Kernel
A long-standing goal in both industry and academia is to develop an LLM inference platform that is portable across hardware architectures, eliminates the need for low-level hand-tuning, and still d…
hgpu.org
November 23, 2025 at 5:51 PM
hgpu.bsky.social
An MLIR pipeline for offloading Fortran to FPGAs via OpenMP

#OpenMP #FPGA #Fortran #Package

hgpu.org?p=30356
An MLIR pipeline for offloading Fortran to FPGAs via OpenMP
With the slowing of Moore’s Law, heterogeneous computing platforms such as Field Programmable Gate Arrays (FPGAs) have gained increasing interest for accelerating HPC workloads. In this work …
hgpu.org
November 16, 2025 at 3:02 PM
hgpu.bsky.social
HipKittens: Fast and Furious AMD Kernels

#AMD #Performance #Package

hgpu.org?p=30355
HipKittens: Fast and Furious AMD Kernels
AMD GPUs offer state-of-the-art compute and memory bandwidth; however, peak performance AMD kernels are written in raw assembly. To address the difficulty of mapping AI algorithms to hardware, rece…
hgpu.org
November 16, 2025 at 3:01 PM
hgpu.bsky.social
PRAGMA: A Profiling-Reasoned Multi-Agent Framework for Automatic Kernel Optimization

#CUDA #LLM #CodeGeneration

hgpu.org?p=30354
PRAGMA: A Profiling-Reasoned Multi-Agent Framework for Automatic Kernel Optimization
Designing high-performance kernels requires expert-level tuning and a deep understanding of hardware characteristics. Recent advances in large language models (LLMs) have enabled automated kernel g…
hgpu.org
November 16, 2025 at 3:00 PM
hgpu.bsky.social
A High-Throughput GPU Framework for Adaptive Lossless Compression of Floating-Point Data

#CUDA #Compression #Package

hgpu.org?p=30353
A High-Throughput GPU Framework for Adaptive Lossless Compression of Floating-Point Data
The torrential influx of floating-point data from domains like IoT and HPC necessitates high-performance lossless compression to mitigate storage costs while preserving absolute data fidelity. Leve…
hgpu.org
November 16, 2025 at 2:59 PM
hgpu.bsky.social
MT4G: A Tool for Reliable Auto-Discovery of NVIDIA and AMD GPU Compute and Memory Topologies

#CUDA #PTX #HIP #Benchmarking #Package

hgpu.org?p=30352
MT4G: A Tool for Reliable Auto-Discovery of NVIDIA and AMD GPU Compute and Memory Topologies
Understanding GPU topology is essential for performance-related tasks in HPC or AI. Yet, unlike for CPUs with tools like hwloc, GPU information is hard to come by, incomplete, and vendor-specific. …
hgpu.org
November 16, 2025 at 2:58 PM
hgpu.bsky.social
CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization

#CUDA #CodeGeneration #Performance #Package

hgpu.org?p=30343
CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization
Developing efficient CUDA kernels is increasingly critical for AI applications such as large-scale LLM training. However, manual kernel design is both costly and time-consuming, motivating automati…
hgpu.org
November 9, 2025 at 4:29 PM
hgpu.bsky.social
Characterizing the Performance of Parallel Data-Compression Algorithms across Compilers and GPUs

#CUDA #HIP #Compression #Package

hgpu.org?p=30342
Characterizing the Performance of Parallel Data-Compression Algorithms across Compilers and GPUs
Different compilers can generate code with notably different performance characteristics – even on the same system. Today, GPU developers have three popular options for compiling CUDA or HIP …
hgpu.org
November 9, 2025 at 4:28 PM
hgpu.bsky.social
FP8-Flow-MoE: A Casting-Free FP8 Recipe without Double Quantization Error

#FP8 #Precision

hgpu.org?p=30341
FP8-Flow-MoE: A Casting-Free FP8 Recipe without Double Quantization Error
Training large Mixture-of-Experts (MoE) models remains computationally prohibitive due to their extreme compute and memory demands. Although low-precision training promises to accelerate computatio…
hgpu.org
November 9, 2025 at 4:28 PM
hgpu.bsky.social
AMD MI300X GPU Performance Analysis

#AMD #HIP #Benchmarking #Performance

hgpu.org?p=30340
AMD MI300X GPU Performance Analysis
The rapid growth of large language models (LLMs) has driven the need for high-performance, scalable GPU hardware capable of efficiently serving models with hundreds of billions of parameters. While…
hgpu.org
November 9, 2025 at 4:27 PM
hgpu.bsky.social
RDMA Point-to-Point Communication for LLM Systems

#CUDA #RDMA #LLM #Package

hgpu.org?p=30339
RDMA Point-to-Point Communication for LLM Systems
Emerging Large Language Model (LLM) system patterns, such as disaggregated inference, Mixture-of-Experts (MoE) routing, and asynchronous reinforcement fine-tuning, require flexible point-to-point c…
hgpu.org
November 9, 2025 at 4:26 PM