View Single Post
  #44 (permalink)  
Old October 4, 2007, 03:14 PM
pscout pscout is offline
Fallen Folder
F@H
 
Join Date: Mar 2007
Location: folding in GTA
Posts: 375
Default

Quote:
Originally Posted by sswilson View Post
I'm going to have to do a bit of playing with this.... am I right in assuming that more mem dedicated to the job is going to equate to better PPD, or are the work packets small enough that I'm actually wasting valuable mem space by setting it to the max?

Since I'm not doing much other than the folding within the VM, all I need for resources dedicated to it are those directly related to the folding.

I don't suppose anybody has already done the legwork on this have they??? :)
with the current wu's, having a vm larger than ~440 MB is wasteful once the client is running. You might be able to run with less but it was good enuf for me since they are dedicated rigs. I could increase it's refresh frequency to more that 10 minutes which i use now but like 10 minutes ... the default is 5 minutes.

Having more than that makes the ui to the guest linux more responsive, and gets the vm and smp client started faster. Since i seldom touch the linux ui or client once it is folding, on my 1 gb quads i make each vm 440 MB and set the host memory to 880 (default is 896 on a 1 gb rig).

The largest wu's when i last looked (ie in native linux via system monitor) took around 400 MB total, so 440 is reasonably safe i think.

This is one reason for not running fahmon in the linux guest .... run 1 copy on wxp to monitor all your vms. Fahmon can be a bit of a pig on my wxp monitoring rig but probably because it is monitoring 27 instances and reports network share errors regularly.

Last edited by pscout; October 4, 2007 at 03:17 PM.
Reply With Quote