LLM-Inference-Bench
收藏资源简介:
LLM-Inference-Bench是由阿贡国家实验室创建的一个综合基准测试套件,旨在评估大型语言模型在多种AI加速器上的推理性能。该数据集涵盖了从7亿到70亿参数的多种LLM模型,包括LLaMA、Mistral和Qwen系列,并分析了不同硬件平台(如Nvidia和AMD的GPU以及Intel Habana和SambaNova的AI加速器)上的性能。数据集的创建过程包括对多种推理框架(如vLLM、TensorRT-LLM、llama.cpp和Deepspeed-MII)的全面评估,旨在为研究人员提供优化LLM性能和硬件选择的参考。该数据集主要应用于自然语言处理、内容生成和决策支持系统等领域,旨在解决LLM在不同硬件平台上的性能瓶颈问题。
LLM-Inference-Bench is a comprehensive benchmark suite developed by Argonne National Laboratory, designed to evaluate the inference performance of large language models across various AI accelerators. This suite covers a diverse range of LLM models with parameter sizes from 700 million to 7 billion, including the LLaMA, Mistral and Qwen model families, and analyzes performance across different hardware platforms such as NVIDIA and AMD GPUs, as well as AI accelerators from Intel Habana and SambaNova. The development of this benchmark suite involves comprehensive evaluations of multiple inference frameworks including vLLM, TensorRT-LLM, llama.cpp and DeepSpeed-MII, aiming to provide researchers with valuable references for optimizing LLM performance and selecting appropriate hardware. This benchmark suite is primarily applied in fields such as natural language processing, content generation and decision support systems, with the core goal of addressing performance bottlenecks of LLMs across different hardware platforms.
LLM-Inference-Bench
数据集概述
LLM-Inference-Bench 是一个用于在AI加速器上对大型语言模型(LLM)进行推理基准测试的数据集。
评估的框架和硬件
| 框架/硬件 | NVIDIA A100 | NVIDIA H100 | NVIDIA GH200 | AMD MI250 | AMD MI300X | Intel Max1550 | Habana Gaudi2 | Sambanova SN40L |
|---|---|---|---|---|---|---|---|---|
| vLLM | Yes | Yes | Yes | Yes | Yes | Yes | No | N/A |
| llama.cpp | Yes | Yes | Yes | Yes | Yes | Yes | N/A | N/A |
| TensorRT-LLM | Yes | Yes | Yes | N/A | N/A | N/A | N/A | N/A |
| DeepSpeed-MII | Yes | No | No | No | No | No | Yes | N/A |
| Sambaflow | N/A | N/A | N/A | N/A | N/A | N/A | N/A | Yes |
引用信息
@INPROCEEDINGS{####, author={Krishna Teja Chitty-Venkata and Siddhisanket Raskar and Bharat Kale and Farah Ferdaus and Aditya Tanikanti and Ken Raffenetti and Valerie Taylor and Murali Emani and Venkatram Vishwanath}, booktitle={2024 IEEE/ACM International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS)}, title={LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators}, year={2024}, volume={}, number={}, pages={}, keywords={Large Language Models, AI Accelerators, Performance Evaluation, Benchmarking }, doi={}}

- 1LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators阿贡国家实验室 · 2024年



