Our paper with the name “Formal methods for GPGPU programming: is the demand met?” has been accepted on the Integrated Formal Methods conference, which takes place 16–20 November 2020.
In this paper, we looked at Stack Overflow, to see what issues developers struggled with whilst programming for GPUs, using OpenCL or CUDA. We made an overview of these problems, and we looked at research and industry for their the solutions for these problems. We think many problems are already properly addressed, but there is still a need for easy to use sound tools. Alternatively, languages or programming styles can be created, that allows for easier checking for soundness.