Topic: "CUDA, ROCm, oneAPI – All for One or One for All?"

Speaker: Armin Sobhani, SHARCNET

Video link

Recording

 

--- 

For a long time, CUDA was the platform of choice for developing applications running on NVIDIA’s GPUs. That is starting to change in recent years with the introduction of AMD’s ROCm and Intel’s oneAPI which both support GPUs by other vendors. While ROCm targets both AMD and NVIDIA GPUs, using the recently released drivers by CodePlay, oneAPI applications can run on NVIDIA and AMD in addition to Intel’s GPUs. The question this seminar is trying to answer is if in 2023 you want to start a project targeting GPUs, what would be your platform of choice? Should you go with one or all of them? Later in the seminar, a boilerplate framework named one4all will be introduced that streamlines the process of developing applications targeting all the above platforms. Unit testing with Catch2 as well as benchmarking with Google benchmark are already supported by the framework.

The Compute Ontario Colloquia are weekly Zoom presentations on Advanced Research Computing, High Performance Computing, Research Data Management, and Research Software topics, delivered by staff from three Compute Ontario consortia (CAC, SciNet, SHARCNET) and guest speakers. The series began January 2023 and superseded similar series previously delivered by individual consortia (e.g. General Interest Seminars by SHARCNET or User Group Meeting TechTalks by SciNet). The colloquia are one hour long and include time for questions. No registration is required. Presentations are usually recorded and uploaded to the hosting consortium video channel (colloquia hosted by SHARCNET go to our youtube channel).