high performance computing on graphics processing units: hgpu.org

Programming

hgpu.org » Programming

Memory Efficient Mixed-Precision Optimizers

Basile Lewandowski, Atli Kosson

View

Download (PDF)

Tags: Computer science, CUDA, Machine learning, Mixed precision, nVidia, nVidia A100, nVidia V100

October 1, 2023 by hgpu

OpenMP Kernel Language Extensions for Performance Portable GPU Codes

Shilei Tian, Tom Scogland, Barbara Chapman, Johannes Doerfert

View

Download (PDF)

Tags: AMD Radeon Instinct MI250, ATI, Benchmarking, Compilers, Computer science, CUDA, HIP, nVidia, nVidia A100, OpenMP, performance portability

October 1, 2023 by hgpu

Beehive SPIR-V Toolkit: A Composable and Functional API for Runtime SPIR-V Code Generation

Juan Fumero Alfonso, György Rethy, Athanasios Stratikopoulos, Nikolaos Foutris, Christos-Efthymios Kotselidis

View

Download (PDF)

Source codes

Tags: Code generation, Computer science, Heterogeneous systems, Java, OpenCL, Package

October 1, 2023 by hgpu

Novel Parallelization Strategies for High-Performance DNN Training on HPC Systems

Arpan Jain

View

Download (PDF)

Tags: Artificial intelligence, Computer science, CUDA, Data parallelism, Deep learning, HPC, Machine learning, MPI, Neural networks, nVidia, Speech recognition, Tesla K80, Tesla V100, Thesis

October 1, 2023 by hgpu

Evaluating the performance portability of SYCL across CPUs and GPUs on bandwidth-bound applications

Istvan Z Reguly

View

Download (PDF)

Tags: AMD Radeon Instinct MI250X, ATI, Benchmarking, cfd, Computer science, CUDA, Fluid dynamics, HIP, Intel, Intel Ponte Vecchio Max 1100, nVidia, nVidia A100, OpenCL, performance portability, SYCL

September 24, 2023 by hgpu

Comparing Performance and Portability between CUDA and SYCL for Protein Database Search on NVIDIA, AMD, and Intel GPUs

Manuel Costanzo, Enzo Rucci, Carlos García Sánchez, Marcelo Naiouf, Manuel Prieto-Matías

View

Download (PDF)

Tags: AMD Radeon RX 6700 XT, ATI, Bioinformatics, Biology, Computer science, CUDA, Intel, Intel UHD 630, nVidia, nVidia GeForce GTX 1080, nVidia GeForce GTX 980, nVidia GeForce RTX 2070, nVidia GeForce RTX 3070, nVidia GeForce RTX 3090, oneAPI, Programming Languages, Smith-Waterman algorithm, SYCL, Tesla V100

September 24, 2023 by hgpu

Comparing Llama-2 and GPT-3 LLMs for HPC kernels generation

Pedro Valero-Lara, Alexis Huante, Mustafa Al Lail, William F. Godoy, Keita Teranishi, Prasanna Balaprakash, Jeffrey S. Vetter

View

Download (PDF)

Source codes

Tags: Code generation, Computer science, CUDA, Deep learning, Fortran, HIP, HPC, nVidia, OpenACC, OpenMP, Package, Python

September 17, 2023 by hgpu

Many Cores, Many Models: GPU Programming Model vs. Vendor Compatibility Overview

Andreas Herten

View

Download (PDF)

Tags: Computer science, CUDA, Fortran, HIP, HPC, nVidia, Overview, programming model, SYCL

September 17, 2023 by hgpu

Improving the Efficiency of OpenCL Kernels through Pipes

Mostafa Eghbali Zarch, Michela Becchi

View

Download (PDF)

Tags: Computer science, FPGA, OpenCL, Performance, performance portability

September 17, 2023 by hgpu

host device — Generic programming in Cuda

Thomas Mejstrik

View

Download (PDF)

Tags: Computer science, CUDA, nVidia, Programming Languages

September 17, 2023 by hgpu

Unified Shared Memory: Friend or Foe?

Juan Fumero Alfonso, Florin-Gabriel Blanaru, Athanasios Stratikopoulos, Steve Dohrmann, Sandhya Viswanathan, Christos-Efthymios Kotselidis

View

Download (PDF)

Tags: Code generation, Computer science, CUDA, FPGA, Heterogeneous systems, Java, Memory, nVidia, nVidia GeForce RTX 3070, OpenCL

September 17, 2023 by hgpu

Leveraging Memory Copy Overlap for Efficient Sparse Matrix-Vector Multiplication on GPUs

Guangsen Zeng, Yi Zou

View

Download (PDF)

Tags: Computer science, CUDA, Machine learning, nVidia, nVidia A40, Sparse matrix

September 6, 2023 by hgpu

Agentic Code Optimization via Compiler-LLM Cooperation

MegaTrain: Full Precision Training of 100B+ Parameter Large Language Models on a Single GPU

Device Virtual Machine (DVM)

DVM: Real-Time Kernel Generation for Dynamic AI Models

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

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

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

* * *

high performance computing on graphics processing units: hgpu.org

Programming

Memory Efficient Mixed-Precision Optimizers

OpenMP Kernel Language Extensions for Performance Portable GPU Codes

Beehive SPIR-V Toolkit: A Composable and Functional API for Runtime SPIR-V Code Generation

Novel Parallelization Strategies for High-Performance DNN Training on HPC Systems

Evaluating the performance portability of SYCL across CPUs and GPUs on bandwidth-bound applications

Comparing Performance and Portability between CUDA and SYCL for Protein Database Search on NVIDIA, AMD, and Intel GPUs

Comparing Llama-2 and GPT-3 LLMs for HPC kernels generation

Many Cores, Many Models: GPU Programming Model vs. Vendor Compatibility Overview

Improving the Efficiency of OpenCL Kernels through Pipes

host device — Generic programming in Cuda

Unified Shared Memory: Friend or Foe?

Leveraging Memory Copy Overlap for Efficient Sparse Matrix-Vector Multiplication on GPUs

Recent source codes

Agentic Code Optimization via Compiler-LLM Cooperation

MegaTrain: Full Precision Training of 100B+ Parameter Large Language Models on a Single GPU

Device Virtual Machine (DVM)

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

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

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

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

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

Most viewed papers (last 30 days)