high performance computing on graphics processing units: hgpu.org

hgpu.org » NLP

InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU

Heejun Lee, Geon Park, Jaduk Suh, Sung Ju Hwang

View

Download (PDF)

Source codes

Tags: Computer science, LLM, NLP, nVidia, nVidia GeForce RTX 4090, Package

February 16, 2025 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

Efficiently Training 7B LLM with 1 Million Sequence Length on 8 GPUs

Pinxue Zhao, Hailin Zhang, Fangcheng Fu, Xiaonan Nie, Qibin Liu, Fang Yang, Yuanbo Peng, Dian Jiao, Shuaipeng Li, Jinbao Xue, Yangyu Tao, Bin Cui

View

Download (PDF)

Tags: AI, Computer science, CUDA, LLM, Machine learning, NLP, nVidia, nVidia A800

July 28, 2024 by hgpu

A Survey on Hardware Accelerators for Large Language Models

Christoforos Kachris

View

Download (PDF)

Tags: Computer science, Energy-efficient computing, FPGA, Hardware Architecture, NLP, survey

January 21, 2024 by hgpu

Cramming: Training a Language Model on a Single GPU in One Day

Jonas Geiping, Tom Goldstein

View

Download (PDF)

Source codes

Tags: Computer science, Deep learning, NLP, nVidia, nVidia GeForce RTX 2080 Ti, nVidia RTX A4000, nVidia RTX A6000, Package

January 8, 2023 by hgpu

EnergonAI: An Inference System for 10-100 Billion Parameter Transformer Models

Jiangsu Du, Ziming Liu, Jiarui Fang, Shenggui Li, Yongbin Li, Yutong Lu, Yang You

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, Deep learning, Heterogeneous systems, NLP, nVidia, nVidia A100, Package

September 11, 2022 by hgpu

Towards making the most of NLP-based device mapping optimization for OpenCL kernels

Petros Vavaroutsos, Ioannis Oroutzoglou, Dimosthenis Masouros, Dimitrios Soudris

View

Download (PDF)

Tags: AMD Radeon HD 7970, ATI, Computer science, Deep learning, Machine learning, NLP, nVidia, nVidia GeForce GTX 970, OpenCL

September 4, 2022 by hgpu

Long Code for Code Search

Fan Hu, Yanlin Wang, Lun Du, Hongyu Zhang, Shi Han, Dongmei Zhang, Xirong Li

View

Download (PDF)

Tags: Computer science, Deep learning, NLP, nVidia, Software Engineering, Tesla V100

August 28, 2022 by hgpu

MapReduce for Counting Word Frequencies with MPI and GPUs

Nithin Kavi

View

Download (PDF)

Source codes

Tags: Algorithms, Computer science, CUDA, Julia, MapReduce, MPI, NLP, nVidia, Optimization, Package, Tesla V100

June 19, 2022 by hgpu

Lossless Acceleration for Seq2seq Generation with Aggressive Decoding

Tao Ge, Heming Xia, Xin Sun, Si-Qing Chen, Furu Wei

View

Download (PDF)

Source codes

Tags: Computer science, CUDA, Machine learning, NLP, nVidia, Package, Tesla V100

May 29, 2022 by hgpu

The Ecological Footprint of Neural Machine Translation Systems

Dimitar Sherionov, Eva Vanmassenhove

View

Download (PDF)

Source codes

Tags: Artificial intelligence, Computer science, CUDA, Deep learning, LSTM, NLP, nVidia, nVidia GeForce GTX 1080 Ti, Package, Tesla P100

February 13, 2022 by hgpu

Autotuning CUDA: Applying NLP Techniques to LS-CAT

Lars Bjertnes, Jacob O. Tørring, Anne C. Elster

View

Download (PDF)

Tags: Auto-Tuning, Computer science, CUDA, Machine learning, NLP, nVidia

December 5, 2021 by hgpu

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

Hardware Compute Partitioning on NVIDIA GPUs for Composable Systems

KISim: Kubernetes Intelligent Scheduling Simulator

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

Efficient GPU Implementation of Multi-Precision Integer Division

exa-AMD: Exascale Accelerated Materials Discovery

Accelerated discovery and design of Fe-Co-Zr magnets with tunable magnetic anisotropy through machine learning and parallel computing

ParEval: A Parallel Code Evaluation Benchmark

ParEval-Repo: A Benchmark Suite for Evaluating LLMs with Repository-level HPC Translation Tasks

FlashSparse: Minimizing Computation Redundancy for Fast Sparse Matrix Multiplications on Tensor Cores

Libra: Synergizing CUDA and Tensor Cores for High-Performance Sparse Matrix Multiplication

WiLLM: An Open Wireless LLM Communication System

Vcc: the Vulkan Clang Compiler

No More Shading Languages: Compiling C++ to Vulkan Shaders

hpcbench: A set of benchmarking utilities for biomolecular simulation tools

Engineering Supercomputing Platforms for Biomolecular Applications

See all packages

* * *

high performance computing on graphics processing units: hgpu.org

InfiniteHiP: Extending Language Model Context Up to 3 Million Tokens on a Single GPU

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

Efficiently Training 7B LLM with 1 Million Sequence Length on 8 GPUs

A Survey on Hardware Accelerators for Large Language Models

Cramming: Training a Language Model on a Single GPU in One Day

EnergonAI: An Inference System for 10-100 Billion Parameter Transformer Models

Towards making the most of NLP-based device mapping optimization for OpenCL kernels

Long Code for Code Search

MapReduce for Counting Word Frequencies with MPI and GPUs

Lossless Acceleration for Seq2seq Generation with Aggressive Decoding

The Ecological Footprint of Neural Machine Translation Systems

Autotuning CUDA: Applying NLP Techniques to LS-CAT

Recent source codes

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

Hardware Compute Partitioning on NVIDIA GPUs for Composable Systems

KISim: Kubernetes Intelligent Scheduling Simulator

Efficient GPU Implementation of Multi-Precision Integer Division

exa-AMD: Exascale Accelerated Materials Discovery

ParEval: A Parallel Code Evaluation Benchmark

FlashSparse: Minimizing Computation Redundancy for Fast Sparse Matrix Multiplications on Tensor Cores

WiLLM: An Open Wireless LLM Communication System

Vcc: the Vulkan Clang Compiler

hpcbench: A set of benchmarking utilities for biomolecular simulation tools

Most viewed papers (last 30 days)