
Mentor: Shehzeen Malik
Company: University of Engineering & Technology, Lahore, Pakistan
This project aims to develop and verify a custom RISC-V Vector (RVV) v1.0 compliant coprocessor to accelerate an Electrocardiogram (ECG) detection algorithm. Building on our previous work with a CGRA accelerator, this project addresses its key limitations by leveraging the standardized RISC-V Vector extension, which offers superior toolchain support and predictable data-level parallelism.
The mentee will focus on two core pillars:
- Verification and Integration of RVV Modules: Rigorously test and integrate pre-designed hardware modules to build a functional vector coprocessor.
- Algorithm Development: Implement and optimize vectorized ECG signal processing kernels using RVV intrinsics.
Why This Project?
Our previous LFX project using the OpenEdgeCGRA revealed critical issues: poor documentation, a custom compiler, and inefficient spatial-parallelism leading to excessive NOPs. By adopting the RISC-V Vector standard, we solve these problems. The mentee will gain hands-on experience with industry-standard verification methodologies and hardware-accelerated algorithm design, contributing directly to an open-source biomedical computing platform.
Project Goals:
- Verify and integrate the components of a lightweight RVV coprocessor (VLEN=128, 4 lanes).
- Develop comprehensive testbenches and a verification plan for the vector processor.
- Design and optimize vectorized ECG detection algorithms using RVV intrinsics.
- Benchmark the performance of the vectorized algorithms against scalar implementations.
- Create detailed documentation, including a verification report.
Repository URL: https://github.com/meds-ee-uet/UET-RVMCU
Learning Objectives:
Technical Skills:
- Gain a deep, practical understanding of the RISC-V Vector Extension v1.0 through hands-on verification and programming.
- Master modern digital design verification methodologies, including testbench creation, and debugging of complex hardware modules.
- Understand the full stack of hardware acceleration by developing performance-critical software that targets a custom co-processor.
- Learn biomedical signal processing fundamentals through the implementation of a real-world ECG detection pipeline.
Professional Development:
- Learn industry-standard open-source collaboration tools and workflows (Git, GitHub).
- Practice creating professional-grade technical documentation, including verification reports and algorithm specifications.
- Build a strong portfolio with tangible contributions to a significant open-source hardware/software project.
- Receive structured mentorship with weekly syncs and code reviews, preparing for professional engineering roles.
Mentorship Structure:
- Weekly 1-hour technical meetings for progress review and strategic guidance.
- Daily asynchronous support via Slack/email for immediate problem-solving.
- Bi-weekly deep-dive code and verification plan reviews.
Coding Challenge: https://drive.google.com/file/d/1KkfX5T3XT47w9ORSXTra0-uge1sIm3ce/view?usp=sharing
To apply for this job please visit mentorship.lfx.linuxfoundation.org.