So I recently upgraded from a GTX 260 to a GTX 660 Ti, and I was wondering if it was possible to set up so 2 WU run on the 660, and 1 WU runs on the 260? I've messed around with cc_config and the new app_config files during the outage, but couldn't really seem to get it to work. I'm not even sure if it's possible, but figured someone might know :)

Cheers!

So I recently upgraded from a GTX 260 to a GTX 660 Ti, and I was wondering if it was possible to set up so 2 WU run on the 660, and 1 WU runs on the 260? I've messed around with cc_config and the new app_config files during the outage, but couldn't really seem to get it to work. I'm not even sure if it's possible, but figured someone might know :)

Cheers!

At this point in time it is not possible with one instance of BOINC. Unless the different cards used different apps. Then each app could be configured separately via the app_config.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Ohhh thanks :D I might just set the 260 to use the x38g_cuda32 app and the 660 to use the x41g version :) I'll see if it works!

So I recently upgraded from a GTX 260 to a GTX 660 Ti, and I was wondering if it was possible to set up so 2 WU run on the 660, and 1 WU runs on the 260? I've messed around with cc_config and the new app_config files during the outage, but couldn't really seem to get it to work. I'm not even sure if it's possible, but figured someone might know :)

Cheers!

With x41zc you can set different priorities for the different GPUs, and different parameters.

Claggy

Ohhh thanks :D I might just set the 260 to use the x38g_cuda32 app and the 660 to use the x41g version :) I'll see if it works!

I'm not sure that will work. You can add multiple apps to your app_into.xml, but I don't recall a way to bind an app to a card. IIRC BOINC will just use each app as it more or less feels like it on each card.

Claggy mentioned the x41zc has options to configure per card, but the docs only list processpriority, pfblockspersm, & pfperiodsperlaunch. I don't see an per card instance control, but I might have missed something.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

It should be done on BOINC level (BOINC should know how many instances to launch on each device) not on science app level.
AFAIK current BOINC version can't do that.

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

Hmm thanks for all the replies! It seems it may just be easier to run 2 WU on each card, and just take the performance hit on the 260. Surely the gains on the 660 would more than make up for it :)

Hmm thanks for all the replies! It seems it may just be easier to run 2 WU on each card, and just take the performance hit on the 260. Surely the gains on the 660 would more than make up for it :)

You could run 2 instances of BOINC with 1 GPU in each instance, but that can get messy. Each would be viewed by the stats and stats as a separate host. however each instance would have its own queue to fill.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Go to the Vyper´s site, he teaches how to do that (2WU on 1 GPU, 1 WU on the other), it´s easy to do if you follow his instructions.

http://vyper.kafit.se/wp/index.php/2011/02/04/running-different-nvidia-architectures-most-optimal-at-setihome/

---