![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Not with gpu3.
Using -advmethods with gpu3 forces the client to DL core11 WU instead of core15 WU.
Fermi will only run core15. If you run gpu3 with a Fermi card and put in the -advmethods flag, your client will sit there doing nothing.
-advmethods should only be used with gpu3 if you are running older cards (pre-Fermi) and want to run core11 WU which yield higher PPD on older cards than the core15 WU do. (essentially giving gpu3 users with old cards the same PPD as gpu2 would)
and the bastards aren't even carrying any lube!
EVGA is where it is today largely because of -bigadv. Their top folders are all producing huge due to extensive server farms. Before -bigadv they had production spikes because of the bucks program but were nowhere near overtaking us by millions of PPD advantage. We don't have detailed statistics but they may have even ventured into -bigadv through a broad deployment earlier than we did. IIRC, we were a little late in that respect with the exception of a few individuals on the team. I think lelliot may have been first on this team.
EVGA is where it is today largely because of -bigadv. Their top folders are all producing huge due to extensive server farms and the most commonly used VM image for -bigadv was EVGA's. Before -bigadv they had production spikes because of the bucks program but were nowhere near overtaking us by millions of PPD advantage. We don't have detailed statistics but they may have even ventured into -bigadv through a broad deployment earlier than we did. IIRC, we were a little late in that respect with the exception of a few individuals on the team. I think lelliot may have been first on this team.
I also wonder what will happen when we finally see the openCL ATI client - hopefully they will be stuffed then
Who is MIA? PM me if you know.
We may have more than them and that would be evident from the plethora of threads about the SR-2 here. Heck, we have more threads and info on the SR-2 than the multiprocessor section or even 2CPU.com FTM, LOL.
Not with gpu3.
Using -advmethods with gpu3 forces the client to DL core11 WU instead of core15 WU.
Fermi will only run core15. If you run gpu3 with a Fermi card and put in the -advmethods flag, your client will sit there doing nothing.
-advmethods should only be used with gpu3 if you are running older cards (pre-Fermi) and want to run core11 WU which yield higher PPD on older cards than the core15 WU do. (essentially giving gpu3 users with old cards the same PPD as gpu2 would)
What projects are you getting now?I love these 57xx units, I'm getting higher PPD values on them. My 260s are estimating near 8k and my GT is running nearly 5600k....I hope these keep coming.
Yeah they changed it....I've been running -advmethods, I figured I'd remove it tonight (therein doing core15 only) and see if it got rid of my EUE problems. Anyway I noticed after removing -advmethods it will do both, actually. I've had gpu3 DL and crunch both core11 and core 15 WU now, so they've definitely changed it and -advmethods does not force core11 anymore.
I love these 57xx units, I'm getting higher PPD values on them. My 260s are estimating near 8k and my GT is running nearly 5600k....I hope these keep coming.
Ideally you should, otherwise you get warning message spam in your logfile saying the client can't identify the "GPU Species". Proper project delivery also works best when you are using more recent drivers.
OK, I'll update the drivers this week when I perform some scheduled maintenance. I was going to update the drivers in the near future and might as well do it now. I noticed the message about not identifying the GPU family. I received it on one system but not on another with very recent drivers. Maybe new drivers combined with the GPU3 client will help with the lags of some apps that plagued the GPU2 client.
The 353s are horrible when also folding -bigadv. On a system with several cards these WUs can be highly predatory on the SMP client. It doesn't affect my systems that run regular SMP because I always have one core dedicated to the GPUs, but the 353s wreak untold havoc on my -bigadv clients that have access to all cores.