I am retired at the moment. Last time I got bored fairly quickly. I expect that to happen again.
I had been working part time in the Parallel Runtimes Engineering group at Intel, which is part of Developer Software Engineering. I was also working on a few research problems for a government customer via Intel Federal. Intel, in its current retrenchment, dangled a package that was too good to ignore.
What I was doing at PRE was mostly Intel SHMEM, which is an implementation of OpenSHMEM supporting extensions for Intel GPUs. Essentially, SYCL kernels running on your GPU can do PUTs, GETs, atomic operations, and collective operations to GPU memory throughout the entire parallel application.
You can read about, download, and try Intel SHMEM via https://github.com/oneapi-src/ishmem
We wrote a paper, which was presented at Supercomputing 2024, and is available at https://arxiv.org/abs/2409.20476
The research work was a general investigation into how to reduce messaging latency for short messages. So far the only public results are https://www.intel.com/content/www/us/en/developer/articles/technical/fast-core-to-core-communications.html
I am kinda sorta looking around for a part time researchy gig. I like working with smart people and I like to learn new things. Fairly often good things come out of my efforts. See