Monday, March 17, 2025

Re: Questions regarding GPU Accelerated Signal Processing Blocks

Hi Lunar,

Thank you for reaching out with these questions - will do my best to answer them.  Happy to see you are interested in this topic.

1 - we can look into finding a machine with a cuda supported GPU, but this will probably be easier if you have your own.  
2 - personally I find all of the subproject areas interesting and would be useful to the project, so it is hard to assign priority.  It would be beneficial as you say to propose a development sequence based on feasibility and impact.
3 - unit tests for individual GPU-assisted computation blocks will be sufficient.

Hope this helps, and look forward to your proposal.

Josh

On Wed, Mar 12, 2025 at 3:14 AM Mohammad Mosiur Lunar <mmlunar@hotmail.com> wrote:
Hello Everyone,
I hope this email finds you well. My name is Lunar, and I am currently a Ph.D. student at the University of Nebraska-Lincoln, specializing in wireless communication experimentation and data analysis using deep learning. Through my research, I have gained extensive experience with GNU Radio and USRP devices, and I am particularly interested in leveraging GPU acceleration for real-time signal processing.
The GPU-Accelerated Signal Processing Blocks project strongly aligns with my research interests, and I am eager to contribute. Before submitting my detailed proposal, I would appreciate some clarifications regarding the project scope:
  1. Hardware/Computing Resources: Since this project requires GPU acceleration, will GNU Radio provide contributors with access to a CUDA-supported GPU, either via a development machine or through a cloud computing service? Or is it expected that contributors arrange this themselves?
  2. Task Prioritization: The project outlines several subprojects. Is there a predefined priority among them, or should the contributor propose a development sequence based on feasibility and impact?
  3. Testing Scope: Will unit tests for individual GPU-assisted computation blocks be sufficient, or is there an expectation to include integration tests with USRP devices to validate end-to-end performance?
Gaining clarity on these points will help me structure my proposal effectively. I appreciate your time and look forward to your guidance.
Best regards,
Lunar

No comments:

Post a Comment