Optimizing BOINC to retrieve more (cuda_femi) workunits?

JBertok's picture

Hello,

I've got an i790 Ultra motherboard and Core 2 Extreme QX9650 with three GTX 570 that I use for running BOINC.  The only project I am subscribed to is SETI@Home, so all resources are committed to SETI 100%.

This system has no trouble staying supplied with "SETI@home Enhanced 6.03" workunits but it constantly is running out of "SETI@home Enhanced 6.10 (cuda_femi) workunits.  I can't find any settings or parameters to make it prefer to get the Femi wu's.  The work buffer is already set to the maximum of 10 days.

I'm curious what others' experiences are with multi-GPU systems producing Fermi Workunit results every 1-2 minutes.

Thank You

Comments

Phil Horney's picture

I don't think there is really

I don't think there is really a way to retrieve more GPU work units.  Your system makes requests for WUs and the server sends you what it has, so if there aren't any GPU WUs available it won't send you any.  The number of various work units available is always fluctuating especially if the splitters aren't running.

One thing you can check is to make sure your computing preferences are set correctly.  In the BOINC 7.x versions, they changed the way the buffers work.  Whereas in the 6.x clients you could just set the maximum buffer, in the 7.x version there are now two buffer controls - the "minimum work buffer" and the "max additional work buffer."  You want to set the minimum to something like 7 days and the max additional to something very small, like 0.1 days so that it keeps requesting new units to fill the additional buffer.

-=Phil (Deltoid) - webmaster@gpuug.org

lylahonks's picture

It depends on the connection

It depends on the connection if it load fast. Most of computer users were buy motherboard for the CPU that they concerned.

 

LadyL's picture

Which BOINC version?

Which BOINC version?