high performance computing on graphics processing units: hgpu.org

hgpu.org » nVidia

GPUVM: GPU-driven Unified Virtual Memory

Nurlan Nazaraliyev, Elaheh Sadredini, Nael Abu-Ghazaleh

View

Download (PDF)

Tags: Computer science, CUDA, Memory, nVidia, Operating systems, Performance, Tesla V100

November 17, 2024 by hgpu

Context Parallelism for Scalable Million-Token Inference

Amy (Jie) Yang, Jingyi Yang, Aya Ibrahim, Xinfeng Xie, Bangsheng Tang, Grigory Sizov, Jeremy Reizenstein, Jongsoo Park, Jianyu Huang

View

Download (PDF)

Tags: Artificial intelligence, Benchmarking, Computer science, LLM, nVidia, nVidia H100, Performance

November 17, 2024 by hgpu

Kokkidio: Fast, expressive, portable code, based on Kokkos and Eigen

Lennart Steffen, Reinhard Hinkelmann

View

Download (PDF)

Source codes

Tags: AMD Radeon Instinct MI100, ATI, Computer science, Intel, Intel Data Center GPU Max 1550, Linear Algebra, nVidia, nVidia A100, nVidia GeForce GTX Titan V, Package, performance portability

November 17, 2024 by hgpu

Improving Parallel Program Performance Through DSL-Driven Code Generation with LLM Optimizers

Anjiang Wei, Allen Nie, Thiago S. F. X. Teixeira, Rohan Yadav, Wonchan Lee, Ke Wang, Alex Aiken

View

Download (PDF)

Tags: Code generation, Computer science, DSL, LLM, Matrix multiplication, nVidia, Optimization, Tesla P100

November 17, 2024 by hgpu

LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators

Krishna Teja Chitty-Venkata, Siddhisanket Raskar, Bharat Kale, Farah Ferdaus, Aditya Tanikanti, Ken Raffenetti, Valerie Taylor, Murali Emani, Venkatram Vishwanath

View

Download (PDF)

Source codes

Tags: AI, AMD Radeon Instinct MI250, AMD Radeon Instinct MI300X, Artificial intelligence, ATI, Benchmarking, Computer science, CUDA, LLM, Machine learning, nVidia, nVidia A100, nVidia GH200, nVidia H100, OpenCL, Performance

November 10, 2024 by hgpu

On a Simplified Approach to Achieve Parallel Performance and Portability Across CPU and GPU Architectures

Nathaniel Morgan, Caleb Yenusah, Adrian Diaz, Daniel Dunning,Jacob Moore,Erin Heilman, Calvin Roth, Evan Lieberman, Steven Walton, Sarah Brown, Daniel Holladay, Marko Knezevic, Gavin Whetstone, Zachary Baker, Robert Robey

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, Fortran, HIP, nVidia, Package, performance portability, Pthreads

November 10, 2024 by hgpu

Over-synchronization in GPU Programs

Ajay Nayak, Arkaprava Basu

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, nVidia, nVidia A100, nVidia GeForce RTX 3090, Package, Performance, PTX

November 10, 2024 by hgpu

Profile Util library: A quick and easy way to get MPI, OpenMP and GPU runtime information

Pascal Jahan Elahi

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, MPI, nVidia, nVidia GH200, OpenMP, Package, Performance

November 10, 2024 by hgpu

NEO: Saving GPU Memory Crisis with CPU Offloading for Online LLM Inference

Xuanlin Jiang, Yang Zhou, Shiyi Cao, Ion Stoica, Minlan Yu

View

Download (PDF)

Tags: Artificial intelligence, Computer science, CUDA, LLM, Memory, nVidia, nVidia A10, nVidia H100, Tesla T4

November 10, 2024 by hgpu

Data-Driven Dynamic Autotuning: Optimizing Autotuning Overhead with Prior Tuning Data

Jaroslav Oľha

View

Download (PDF)

Tags: AMD Radeon RX Vega 56, ATI, Auto-Tuning, Computer science, Heterogeneous systems, nVidia, nVidia GeForce GTX 1070, nVidia GeForce GTX 680, nVidia GeForce GTX 750, nVidia GeForce RTX 2080 Ti, OpenCL, Performance, Tesla K20, Thesis

November 3, 2024 by hgpu

LLload: An Easy-to-Use HPC Utilization Tool

Chansup Byun, Albert Reuther, Julie Mullen, LaToya Anderson, William Arcand, Bill Bergeron, David Bestor, Alexander Bonn, Daniel Burrill, Vijay Gadepally, Michael Houle, Matthew Hubbell, Hayden Jananthan, Michael Jones, Piotr Luszczek, Peter Michaleas, Lauren Milechin, Guillermo Morales, Andrew Prout, Antonio Rosa, Charles Yee, Jeremy Kepner

View

Download (PDF)

Tags: Computer science, CUDA, HPC, nVidia, Performance

November 3, 2024 by hgpu

Is the GPU Half-Empty or Half-Full? Practical Scheduling Techniques for LLMs

Ferdi Kossmann, Bruce Fontaine, Daya Khudia, Michael Cafarella, Samuel Madden

View

Download (PDF)

Tags: Computer science, LLM, NLP, nVidia, nVidia A100, nVidia H100, Task scheduling

November 3, 2024 by hgpu

* * *

high performance computing on graphics processing units: hgpu.org

GPUVM: GPU-driven Unified Virtual Memory

Context Parallelism for Scalable Million-Token Inference

Kokkidio: Fast, expressive, portable code, based on Kokkos and Eigen

Improving Parallel Program Performance Through DSL-Driven Code Generation with LLM Optimizers

LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators

On a Simplified Approach to Achieve Parallel Performance and Portability Across CPU and GPU Architectures

Over-synchronization in GPU Programs

Profile Util library: A quick and easy way to get MPI, OpenMP and GPU runtime information

NEO: Saving GPU Memory Crisis with CPU Offloading for Online LLM Inference

Data-Driven Dynamic Autotuning: Optimizing Autotuning Overhead with Prior Tuning Data

LLload: An Easy-to-Use HPC Utilization Tool

Is the GPU Half-Empty or Half-Full? Practical Scheduling Techniques for LLMs

Recent source codes

Kokkidio: Kokkos+Eigen - write fast, readable multi-platform code

LLaMA-Inference-Bench

MATAR

Over-synchronization in GPU Programs

profile_util: simple C++ library to report memory, timing, thread affinity

MaC: Mamba CPU performance predictor

Mixed-precision finite element kernels

SHMT: The Simultaneous and Heterogenous Multithreading project

Superpipeline: A Universal Approach for Reducing GPU Memory Usage in Large Models

EnergyUCB-Bandit

Most viewed papers (last 30 days)