5KK73 GPU assignment website 2014/2015

GPU assignment website

Welcome to the GPU assignment website, here we will post announcements, schedules , examples, etc.

Final Deadline!!

Please see the announcements. The final deadline to hand in your reports is Sunday 21-12-2014. Note that even if the server crashes, this deadline will not be moved!

Winner Competition

With great pleasure I can announce to you we have a winner in the mining competition! Sander Vocke has battled his way to the top and even outdid our own GPU expert in disguise J.Argent! Congratulations Sander! The prize for the competition will be awarded in the next year. You can however already see the full stats of the players STATS. We hope you all enjoyed the contest as much as we did folowing you guys from the side line. Have a merry christmas and a happy new year! Mark and Luc.

Submitting the assignment

The assignment has to be submitted, BEFORE the deadline via the submission page.

Assignment deadline

The deadline of the assignment will be moved to the 14th of December 23:59 (for the creative people, that is UTC+1 (with current daylight saving time)). This due to server issues during the last weekend.

Forum only via VPN

For security reasons the forum will now only be accessible via TU/e VPN.

Competition has started!

From today (30-10-2014, 18:00) we have started the competition server. You can participate by putting your code in the 'production' folder, see the competition page for more detailed instructions. Please also make sure you are running the latest software, since the first release we have allowed more nonces (now also capitals and numbers can be used).
Small note: if you mined before the starting date and time, these results have been reset.

Assignment kick-off

Thursday the 23rd of October this assignment will be officially handed out, you can then start with familiarizing yourself with OpenCL and (if you are interrested) CUDA. There are two versions of an example for matrix multiplication: one for OpenCL and another for CUDA. The CUDA version will treat code optimizations for GPU while the OpenCL version focusses more on the code structure of the OpenCV host and kernel files. Most optimizations described for CUDA can also be applied to OpenCL and it might be a good start into OpenCL to try these.
Also, the source code for the application is already available for people who want to start quickly. The application for this year's assignment will be a variant on the bitcoin mining applications that were very popular in recent years.


Competition

The week after the kick-off we will release more information about a competition we want to launch. This competition will involve a benchmark server which will load a 'production' version of your mining application. Every student gets in turn the same amount of processing time to mine as many coins as they can in that time, based on the amount of coins found we will keep a score-board of the best performing implementations. In this contest every student will mine on the same input data, this means found coins are not unique but it will give everyone a fair chance.