Mentor: Kurt Keville
Company: MIT
We plan to refactor and recode Community AI / ML and HPC (double precision) applications to port them to RISC-V this summer. This is a large, potentially daunting effort but we anticipate building scripts and methods to be able to substantially automate the process. Priority will be given to codes that make the most impact. Here is the initial list of 400 codes we are optimizing and compiling for RISC-V this summer… https://tinyurl.com/4v22ujvt
Repository URL: https://github.com/clusterchallenge/Hardware-Abstraction-Layer-Transitional-Libraries
Learning Objectives:
We will start with training and setting a few project design priorities. The mentees will engage with mentors to understand the project scope, objectives, and existing literature on RISC-V, and the necessary concepts for the project including IDEs, Linux, and git. Additionally, initial research will be conducted on current related solutions and techniques. This foundational knowledge will set the stage for the mentee to familiarize themselves with the relevant tools, open-source infrastructure, and community practices that will be essential as the project progresses. Subsequently, the mentees will begin porting applications, potentially inefficiently at first until their skills improve, collaborating closely with the mentors to iterate through requirements and revisions. This will involve creating design documentation and identifying potential hardware acceleration features.
By the end of the summer the students will have learned tips and tricks for software optimizations with RISC-V as a target as well as learning techniques that will further serve to inform the students on RISC-V hardware design. We will lean heavily on the textbook “Numerical Recipes : The Art of Scientific Computing”
Coding Challenge:
https://docs.google.com/document/d/17b1WxQdUgGKqB6ixp-FyVxumYI2-o-xWhDn9Fhc4b7k/edit?usp=sharing
To apply for this job please visit mentorship.lfx.linuxfoundation.org.