Heterogeneous programming is a topic. The topic of numerous articles on resources for developers and reports at thematic conferences. The topic for the Intel blog, after all, the Intel oneAPI Toolkit and Data Parallel C ++ were recently born. Finally, a great topic for our rubric, “Ask a question to an Intel expert,” don’t think that we forgot about our tradition of communicating with knowledgeable people - we were just waiting for a suitable occasion.
So, this time we are talking about programming for various architectures - CPU, GPU, FPGA, and tools created for this: OpenCL, SYCL, DPC ++ and, in fact, oneAPI. Our expert is Konstantin Vladimirov .
About himself, Konstantin, our today's expert, said just a few words:
- I have been working as a programmer for the past 15 years, 10 of them in the field of compilers.
- I am leading a team in the Moscow office of Intel developing vector optimizations for a graphical compiler.
- I teach programming at the Moscow Institute of Physics and Technology at the Department of Microprocessor Technologies in Intelligent Systems.
Brevity is the sister of leading Intel experts. Although, this does not apply to answers to questions. Konstantin expects you to have any questions about heterogeneous programming within the open standards supported by Intel. That is, let's talk about OpenCL, SYCL, DPC ++, One API and SPIRV, but leave aside third-party technologies such as CUDA or HIP. We will give the opportunity to tell experts of other companies about them.
Traditionally, the author of the best question will receive a prize from Intel (the prize is delivered from Moscow within the Russian Federation). Ask your questions in the comments to this post, saul's private messages, or to intelhabra@gmail.com by Wednesday, December 4th , inclusive. Next, the expert will prepare his answers, and we will publish them on the blog. Together with the name of the winner.