Project Difficulty
Intermediate Audience: Computer Science (CS) / Electrical Engineering (EE)
Description
This project aims to benchmark inference on Arm-based servers using the MLPerf Inference benchmark suite. The project spans performance analysis across different configurations of Arm-based servers. The main deliverable is a comprehensive benchmarking setup that can evaluate the performance of large language models (LLMs) on various Arm server configurations in addition to a report highlighting the performance difference and how to recreate the results. This project will provide practical experience in benchmarking, performance analysis, and working with Arm-based server architectures. The final output will be a detailed report and a functional benchmarking infrastructure that can be used for further research and development.
Estimated Project Duration
The project is estimated to take 8-12 weeks to complete, involving a team of 3-5 participants. There is no hard deadline, but timely completion is encouraged to maximize learning outcomes.
Hardware / Software Requirements
- Languages: Python, C++
- Tooling: MLPerf, TensorFlow, PyTorch
- Hardware: Arm-based server, access to cloud service providers
- IP access: Arm Academic Access member (link to get if they don’t have it)
Resources
- MLPerf Inference
- MLPerf Inference Benchmark Suite
- Blog on Arm Server inference performance
- Previous project submissions: GitHub link to past projects
Benefits
-
Standout projects could be internally referred for relevant positions at Arm!
-
If your submission is approved, you will receive a recognised badge that you can list on your CV and shared on LinkedIn. A great way to stand out from the crowd!
-
It’s a great way to demonstrate your initiative and commitment to your field.
-
It offers the opportunity to learn valuable skills that are highly relevant to a successful career at Arm!
