• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Prime Ninety Five-2 instances?

C

Chomp

Gawd
Joined
Sep 16, 2003
Messages
547
Well I'm overclocked now (sig) and it's time to run some benchmarks and stability tests.

Since with a P4 w/HT you can set affinities for programs, I can run 2 instances of prime95, one on each 'processor' or affinity. Is this the correct way to test for stability, or should I just run 1 instance of prime95?

How long should it run to be considered stable? I ran 2 instances, one on each affinity, for 7 hours with no errors. However, last night I ran it (on both affinities again), and one instance got an error after 2 hours, while the otherone was still running for 9 hours.
 
Don't even mess with those....Affinity is more for actual 2 physical processors and I have tested it and in real world testing you are better off for performance not to touch any priorities or affinity.....They just don't seem to work properly with HT....test it out and you will see what I mean...


1 instance of prime will load only 50% on the HT cpu.....

Simply then go to start - run - cmd.... Then use CD to change directory to the folder prime95 is and then type prime95 -a ...This will launch a second instance....

Notice 2 instances will run 100% load....


I actually like the running one instance of prime95 and do a 4 hour high VBR avi to mpeg2 encoding with TMPGenc.....

If I am going to be by the computer's side for the duration when the tmpgenc is done I just do it again or run SETI or FH, or maybe run a 1800frame high resolution CAD rendering....

If I am going to bed 2x pirme95 is just pretty hard to beat and 8-12 hours later when I get up I get a good sense of stability. If you will be doing distributed computing let this run for up to 24 hours....

The main goal is to run the hT thread at 100% for as long as possible....I have never had that pass and not have 2 instances of prime95 pass as well....
 
So basically, I should run 2 instances of prime95, but don't mess with the affinity stuff, right?

How long should I run it before it's considered stable?


Lastly, why does it only use 50% cpu, whereas prime95 on an AMD would use 100%?
 
Yes don't mess with priorities or affinity....

Run it 8-12 hours for pretty much most things...24 hours if you do distributed computing programs like SETI or FH....


It show 50% because windows sees 2 cpus (though virtual and not physical), and the single thread of prime95 is being handled by 1 of those (virtual) cpus.....IN actuality it is more then 50% but that is the way the OS sees it....You will notice running in HT with 2 instances of prime95, you will get higher temps as a more complete area of the cpu is active and running at all times thus generating more stress and more heat.....


I would say in most the apps I have ran HT as deliverd 15-22% in alot of them ( non gaming) versus HT off. IN SETI I saw a 40% increase (2 instances) and in FH about the same.....So the benefits are real...
 
Alright, thanks for the help.


BTW, what kinda cooling setup do you have


2.4 @ 3.5 = :eek:
 
Air cooling!!! I have a swiftech MCX400 with a 92mm Panaflo fan (57cfm).....

I have a chemming case (antec knockoff) with 2 intakes and 2 exhaust....

Stock NB chipset cooler from my abit IC7 but used AS5 on it....


Things I want to do.....

1) PLace heatsinks on Mosfets...
2) Better cooler for NB chipset
3) SP94 cause it allows for better airflow and less exhausting of cpu heat onto the ram, mosfets, NB chipset, and top of VGA card...


Edit: Als that is with only 1.6v actual so not much vcore boosting....
 
That's awesome. I almost got that same Heatsink, but I couldn't make up my mind so I just picked the SP-94.


BTW, on prime95, is it better to run Small FFTs(less ram) or Large FFTs(more ram tested).
 
I do the one with less ram and then use memtest to thorughly test it...I don't think prime95 is as good a memory tester and the lots of ram test runs slow and the heat doesn't build up like the large FFT....


I also like good old version 22.12 but have version 23.8...version 23.7 had some bugs.....

Version 22.12 was back in the day when there was one test and you know what you got and could somewhat interpret it...I think they over complicated it and now it seems to be getting buggy....
 
You must log in or register to reply here.
Back
Top