Thanks for the suggestion. :) I will try atomic operations and see if they
can fit my need. :)


Best regards,

./francis



If I understand problem correctly, you can try to use one thread
> to compute length of each list. Then you can use atomic functions
> to find the longest list. This way everything will be done on the GPU,
> and CPU will receive the final result.
>
> Atomic functions (e.g. atomicInc) are available in devices
> with computing capability 1.1 and higher. In practice only
> GeForce 8x00 do not have them, and all other (better) cards
> support those functions.
>
> Hope that helps.
>
_______________________________________________
PyCUDA mailing list
[email protected]
http://lists.tiker.net/listinfo/pycuda

Reply via email to