Matthew Knepley <[email protected]> writes:

>> These users compute redundantly and set MAT_NO_OFF_PROC_ENTRIES.
>
>
> Fine, we should have a flag like that.

We do, it's called MAT_NO_OFF_PROC_ENTRIES.

>> What if you do the sort/reduce thing within thread blocks, and only
>> write the reduced version to global storage?
>>
>
> I think it should be easy, but we will have to see what is out there for
> thread blocks.

Cool.  Maybe we'll be able to work up a proof of concept this week, will
see how it goes tomorrow.

Attachment: pgpUVqfbHiWFD.pgp
Description: PGP signature

Reply via email to