Jump to content

why two primes are uneven? sign of instability


Recommended Posts

I was running for dual prime with the -a0 and -a1 switch added but i noticed one instance is using significantly more memory than the other. I've set the affinity for each core before starting blend test. One instance is going through the second cycle while the other one is still on the first test. I ask cause two instances of folding will use the same amount of resource for each core.

 

At 260x10 1.430V and 3-4-4-8 2.7V 1T 1:1, prime won't fail but the icons become black sqaures instead of showing the pictures for running apps. Does this mean the settings will eventually fail?

Share this post


Link to post
Share on other sites

I usually do a custom test, with each core allocated 800 MB. This still allows a lot of memory to be tested and much stress put on the cores, but enough for Windows to run responsively and not cause it to have to page every little thing to/from the HD. IMO that ruins the stress test run. On my machine, it tries to use 1790 MB with one instance, and since you can't start two copies of P95 at once, one hogs the majority of it before the other copy can get to it. I think that's why the usage is way different.

Share this post


Link to post
Share on other sites

As I said, default blend will do somewhere around 1790 MB in just one copy. Now when you run the -A0 copy on default blend, it'll suck up that amount. Then you go and fire up the -A1 copy on default blend. Where's it going to get that RAM from? That's why I said to do a custom run with 800 MB per core. There's no contention for RAM.

 

Edit: I see what you were saying, and I mistyped what I stated above. When I said you can't run two P95 copies at once, I meant that you couldn't start the stress test on both copies at the exact same time, unless you can pause time somehow :P.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...