Hi all,

in my System I've got multiple GPU's
All of them are crunching just one Tasks.

Some of these GPU's can do more work and are underloaded :-D
Some can do just 1 Task.

How can I specify in the app_config.xml additional work for just one dedicated GPU ?

e.g.
GPU0: 2 Tasks
GPU1: 1 Task

How will this be possible ?

Thanks

Yours
Atratus

Here's the manual: http://boinc.berkeley.edu/wiki/Client_configuration

Per application/version you specify the fractions used. Probably if you visit a project forum that does GPU crunching you can find examples there, for instance https://www.gpugrid.net/forum_thread.php?id=3319

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

Thanks for Your Information.

Of course I know about configuring the cc_config and the app_config

Actually I've git two GPU's with the same Processor-Type, but with differenz CUDA-Cores, in my Sys and want to configure the App to use the Dev0 twice and the Dev1 in single Tasking.

So I can't configure different Apps (e.g. Cuda42 / Cuda50) because it's the same Type of Processor (same Instructions, Same nm Type. Dev0 with 384 Cuda-Cores, and Dev1 with about 1.000 Cuda-Cores)

So me detailed question is only about telling Boinc to Use one of this Dev's twice (2 Tasks) and one of these Devs in Single Mode (1 Task)

How will this be possible ?

There's also app_info.xml for detail configuration, but have no idea at all how that works, or how collaborative it is with app_config.xml or for that matter, the projects. IIRC, some are not too pleased with the use of app_info, from which reason app_config was born.

Edit: The manual http://boinc.berkeley.edu/wiki/Anonymous_platform

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt