Parallel STL Applications

Apr 17, 2014 at 10:33 AM
Very nice implementation. I really think parallel algorithms is the way to go. Having lots of benchmarks would be a good way to demonstrate this.
In addition to your matrix multiply, sort and image processing, you can find a few more applications here, which are also following the API proposed in n3960.
I will be implementing more in the coming days, mainly data parallel applications (ported from CUDA and OpenCL).
This repository also contains an implementation of n3960 I was working on in my spare time, which is not nearly as complete as yours, but aims to be platform independent and use the threading and concurrency utilities from C++11 for parallel::par.
Coordinator
Apr 17, 2014 at 5:55 PM
Hi Thibaut,

Thanks for the links, we'll definitely check it out!

Our implementation is on Windows because we want to take advantage of the Windows Thread Pool, and not roll our own scheduler, but I would be very interested to know how that compares with your portable solution based on std::thread etc.