Exercise 0

Moderator: Programming Massively Parallel Processors

baecher
Neuling
Neuling
Beiträge: 8
Registriert: 6. Apr 2009 14:42

Exercise 0

Beitrag von baecher » 21. Apr 2009 20:44

... is now available on the course page. The submission deadline is April 28th, 2pm - please submit the entire project directory including a Makefile.

Sylvester
Neuling
Neuling
Beiträge: 6
Registriert: 22. Apr 2009 20:51

Re: Exercise 0

Beitrag von Sylvester » 22. Apr 2009 20:54

Are we allowed to assume the matrix dimensions as mulitples of BLOCK_SIZE (thus not really arbitrary)?

baecher
Neuling
Neuling
Beiträge: 8
Registriert: 6. Apr 2009 14:42

Re: Exercise 0

Beitrag von baecher » 22. Apr 2009 22:24

Hi,
Sylvester hat geschrieben:Are we allowed to assume the matrix dimensions as mulitples of BLOCK_SIZE (thus not really arbitrary)?
your program should be able to handle that case. It doesn't have to be anything fancy/fast/efficient though, e.g. a thread which is "out of bounds" could just perform a no-op. Another approach would be to expand an align the matrix dimensions to BLOCK_SIZE and discard the excess result elements.

Antworten

Zurück zu „Programming Massively Parallel Processors“