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

TL;DR

LLM-Inference-Bench is introduced, a comprehensive benchmarking suite to evaluate the hardware inference performance of LLMs, and the strengths and limitations of various models, hardware platforms, and inference frameworks are revealed.

Abstract

Large Language Models (LLMs) have propelled groundbreaking advancements across several domains and are commonly used for text generation applications. However, the computational demands of these complex models pose significant challenges, requiring efficient hardware acceleration. Benchmarking the performance of LLMs across diverse hardware platforms is crucial to understanding their scalability and throughput characteristics. We introduce LLM-Inference-Bench, a comprehensive benchmarking suite to evaluate the hardware inference performance of LLMs. We thoroughly analyze diverse hardware platforms, including GPUs from Nvidia and AMD and specialized AI accelerators, Intel Habana and SambaNova. Our evaluation includes several LLM inference frameworks and models from LLaMA, Mistral, and Qwen families with 7B and 70B parameters. Our benchmarking results reveal the strengths and limitations of various models, hardware platforms, and inference frameworks. We provide an interactive dashboard to help identify configurations for optimal performance for a given hardware platform.

Figures (38)

• Figure 1: LLaMA-3-8B on single A100
• Figure 2: KV Cache Performance Benchmarking
• Figure 3: LLaMA-3-8B Quantization Benchmarking
• Figure 4: NAS and SD on A100 GPU
• Figure 5: Parallelism Comparison within a node