CUDA Subdivide Edges

[Evan Lezar]

Hi all

I have been having some hassles with the serial version of the cuda edge_index_calculator.  I was getting "unspecified launch errors"  for the kernels as the benchmark meshes got larger after I had tried to do more of the resizing etc on the GPU.  This may have been somethings else causing it, but nonetheless - I was concerned with the 5s timeout for the CUDA kernels and thus restructured the code somewhat.  This did not, however solve the problem, so I rolled back the changes that I had made - and started again (with the resizing of the edge lists etc on the device) and now have a working implementation once more.  In terms of performance, it has taken quite a knock, but I this may be due to some (a lot) debugging code still present in cuda_subdivide_edges.

Next step is to bring in the adjacency calculator ( that I have already mentioned ) and work on creating a copy of the device mesh on the device instead of assigning the Input to the Ouput (shallow copy).  I should have both of these done by the end of the weekend.

Then after that, I will take Bart's suggestion and implement a cuda version of a mesh source, suce as a 3d mesh array - which is also used extensively in the mesh benchmarks.

Evan

PS.  Bart, if you're listening, I hope you are enjoying your vacation.

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development

[bART Janssens-2]

On Thursday 07 August 2008 23:10:40 Evan Lezar wrote:
> Next step is to bring in the adjacency calculator ( that I have already
> mentioned ) and work on creating a copy of the device mesh on the device
> instead of assigning the Input to the Ouput (shallow copy).  I should have
> both of these done by the end of the weekend.
>
> Then after that, I will take Bart's suggestion and implement a cuda version
> of a mesh source, suce as a 3d mesh array - which is also used extensively
> in the mesh benchmarks.

OK, sounds good!

> PS.  Bart, if you're listening, I hope you are enjoying your vacation.

I'm writing this from the side of the pool, so it's quite enjoyable, thanks :)

Cheers,

Bart

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
K3d-development mailing list
K3d-development@...
https://lists.sourceforge.net/lists/listinfo/k3d-development