Another conference day. Much of my time is spent talking with other attendees and doing "work", such as preparing my presentation, send emails, etc. However, I do take some time to actually sit in on other presentations, so here are two highlights:
PARSECs - This work explores rewriting some of the PARSEC benchmarks to use a task-based parallelism (OpenMP tasks), rather than pthreads. For many workloads, these changes provide improved scaling. For almost all workloads, the code size was reduced as the original thread pools, job queues, etc could be removed. In the near future, these revised versions should be released.
HRF-Relaxed - The original OpenCL had no memory model; however, many vendors implemented one. Now, C++ and other languages use SC for DRF (sequential consistency for data-race-free programs). Unfortunately, if you use this consistency model in OpenCL, you will lose performance. Instead, this work proposes a hierarchical race-free model, where the races are only checked at a certain scope of the program.
No comments:
Post a Comment