D120.de/forum

In todays exercise session Jens provided the logins for the SVN accounts to those, who were present. If you don't have your SVN login yet, please stop by Jens' office and get yours. You can reach your repository under: https://pcgris-svn.igd.fraunhofer.de/teaching/pmpp-09-projects/teamXX where "XX" ...

in fact you do want to compare the benefit of cuda over a cpu implementation. this includes the necessary transfer and setup operations. it is true that you cannot synchronize your host code with the syncthreads statement. the appropriate function is cudaThreadSynchronize(), which can be found in th...

the assignment of the final projects is available at http://www.gris.informatik.tu-darmstadt.de/~mgoesele/PMPP-SS2009/download/projects-assignment.html if you are not satisfied with your current assignment, you can still change to one of the unassigned projects. in that case please contact me via em...

did you reduce your matrix size? if this helps you definitely know where this is coming from. i can recall a similar situation. it is possible that you get the same interval of device memory allocated, every time you run your program. now if you don't set a different seed for your pseudo-random numb...

Hello Sylvester, your code looks right. You stumbled over a certain problem in CUDA. Your program exited prematurely without any error message. This can easily be verified by looking at the result of your computation. I came across the same problem after adjusting my matrix multiplication dimensions...

Exercise 1 is now available. The submission deadline is May 12th, 2pm.

There will be one bonus point if and only if you solve all items of the exercise. Make sure to include a textfile that presents your results from 3)