Eric J Korpela wrote: > On Thu, Jul 23, 2009 at 7:07 AM, Martin<[email protected]> wrote: >> [email protected] wrote: >>> A few thoughts: >>> >>> There is no guarantee that any two computation devices (general sense here) >>> will be equally efficient at a particular computation. The CPU and GPU in >>> a single box may very well have different efficiencies, and should be >>> treated as separate computation devices for credit comparisons. > >>> Goal: The credit granted for all tasks for a single WU needs to be >>> identical, no mater what computation devices it runs on. >> Sorry, disagree... What happens for the case where the CPU application >> is highly optimised whereas the GPU application is brazenly wasteful of >> the GPU resource? The GPU does more physical work and gains less credit >> per unit of physical processing work done. > > As it should be. Some applications don't work well on a GPU and may > even be slower on a GPU than a CPU. Such an application should not be > ported to a GPU in the first place.
s...@h VLARs?... >> I consider a fairer scheme is that credit is granted for the resource >> actually used. > > A resource wasted is different than a resource used. A resource squandered by a project should still be rewarded for the participant. It's up to the project to get more science out of their participant's participation. >> If for example, the GPU application (wastefully) for a WU performs twice >> as many calculations as the equivalent CPU application, then the GPU >> processed WU should be awarded twice the credit of the CPU processed WU >> (to reflect the additional resource used/squandered). > > Definitely not. The user should be encouraged (through lower credit) > to use their GPU for a task that is more suited to it. Is there not a better way by having the Boinc client allocate resource share by the resource used? The projects will then be encouraged to optimise to get more science from their applications. (The awarded credits share will then also match the user's selected resource share.) Perhaps there needs to be an independent measure of "efficiency"? Regards, Martin -- -------------------- Martin Lomas m_boincdev ml1 co uk.ddSPAM.dd -------------------- _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
