mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2010-01-25, 04:48   #1
hj47
 
hj47's Avatar
 
Oct 2008

26 Posts
Default Intel Burn Test & LL integrity

Sup dawgs

So I've been crunching at 3.2GHz on my i5 750 for a week or so now but have decided to crank up the clocks. I have the system on 3.8GHz with 1.312V, and it has passed the intel burn test (linx) successfully set on maximum memory saturation and ran through 50 runs (which took about 80 minutes).

Am I right to say I shouldn't have jeopardised the integrity of my (future) LL results?

Last fiddled with by hj47 on 2010-01-25 at 04:49
hj47 is offline   Reply With Quote
Old 2010-01-25, 05:23   #2
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

6,287 Posts
Default

I think if you want to ensure that your system is stable for Prime95 then you should be running Prime95. Running some other test program will only prove it to be stable for that other test program.
retina is online now   Reply With Quote
Old 2010-01-25, 06:35   #3
hj47
 
hj47's Avatar
 
Oct 2008

26 Posts
Default

Quote:
Originally Posted by retina View Post
I think if you want to ensure that your system is stable for Prime95 then you should be running Prime95. Running some other test program will only prove it to be stable for that other test program.
Yes but I have had prime95 run for an hour without error, then as soon as I use linx it often fails within the first or second test...
hj47 is offline   Reply With Quote
Old 2010-01-25, 08:11   #4
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

6,287 Posts
Default

Quote:
Originally Posted by hj47 View Post
Yes but I have had prime95 run for an hour without error, then as soon as I use linx it often fails within the first or second test...
Well same thing applies I guess. If you want to ensure that your system is stable for Linx then you should run Linx. Running Prime95 will only prove it to be stable for Prime95. Other apps/OSes will use different resources/drivers/hardware etc. So even if Prime95 passes all tests perfectly for a week that still doesn't mean your X graphics driver is going to work perfectly when you start editing a PNG file.

Last fiddled with by retina on 2010-01-25 at 08:11
retina is online now   Reply With Quote
Old 2010-01-25, 10:52   #5
hj47
 
hj47's Avatar
 
Oct 2008

26 Posts
Default

In that case I guess doing 100 runs of linx + 12 hours of blend + 12 hours small FFT + 12 hours of large FFT's test should suffice?

Last fiddled with by hj47 on 2010-01-25 at 11:08
hj47 is offline   Reply With Quote
Old 2010-01-25, 13:59   #6
garo
 
garo's Avatar
 
Aug 2002
Termonfeckin, IE

22×691 Posts
Default

Not really. It might find a problem or it might not find a problem. Your best bet would be to dedicate your hottest core to doublechecks and check if you get any errors/mismatches. BTW, 1.312 is a bit low. I got to 3.8 but had to up it to 1.375 to avoid once a week errors.
garo is offline   Reply With Quote
Old 2010-01-25, 14:50   #7
petrw1
1976 Toyota Corona years forever!
 
petrw1's Avatar
 
"Wayne"
Nov 2006
Saskatchewan, Canada

10010101000102 Posts
Default

Quote:
Originally Posted by garo View Post
Not really. It might find a problem or it might not find a problem. Your best bet would be to dedicate your hottest core to doublechecks and check if you get any errors/mismatches. BTW, 1.312 is a bit low. I got to 3.8 but had to up it to 1.375 to avoid once a week errors.
So you've been running 3.8 since November without any errors?
I need to try to bump mine up more.

I used the stardard EasyTune which maxes at 3.2. I'm getting DC iteration times of 20 to 21 ms.

Thanks
petrw1 is online now   Reply With Quote
Old 2010-01-25, 18:27   #8
Jeff Gilchrist
 
Jeff Gilchrist's Avatar
 
Jun 2003
Ottawa, Canada

3·17·23 Posts
Default

Quote:
Originally Posted by hj47 View Post
In that case I guess doing 100 runs of linx + 12 hours of blend + 12 hours small FFT + 12 hours of large FFT's test should suffice?
As Garo said, the best thing to do is let Prime95 run in the OS you will be using to do double checks. If you are running say 4 double checks at the same time to stress all 4 of your physical cores, then when that set is done, you can check the Prime95 server to see if the results match the previous one. If they all match you should be good since it will take days to complete the test.

I have had cases where running a day of Intel Burn Test and a day of Prime95 torture tests did not find any problems but running something for days or weeks would error out. Turned out in my case the voltage was a little to low but not enough to notice any problems for every day stuff.

Jeff.
Jeff Gilchrist is offline   Reply With Quote
Old 2010-01-25, 21:53   #9
garo
 
garo's Avatar
 
Aug 2002
Termonfeckin, IE

22·691 Posts
Default

I was running 3.8GHz at 1.35V and I got errors. I have not had any errors since I upped the voltage to 1.375V. I still have a couple of unconfirmed results so I'll wait until they are confirmed/rejected before deciding to go down to 3.6V.
garo is offline   Reply With Quote
Old 2010-01-26, 02:27   #10
hj47
 
hj47's Avatar
 
Oct 2008

10000002 Posts
Default

This is interesting, argh now I've gotta find the sweetspot for performance/voltage/temperature/reliability.

Doublechecking is the only way to go I guess.

Cheers
hj47 is offline   Reply With Quote
Old 2010-01-26, 04:07   #11
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

This thread may be of some relevance here--there's an example of a setup that passed Intel Burn Test but failed Prime95's Large and Blend tests. Perhaps the Intel Burn Test only does the equivalent of Prime95's small-FFT test? (Disclaimer: I've never used the IBT myself.)

Last fiddled with by mdettweiler on 2010-01-26 at 04:07
mdettweiler is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Modifying the Lucas Lehmer Primality Test into a fast test of nothing Trilo Miscellaneous Math 25 2018-03-11 23:20
Integrity check of Prime95 archive? WaiCeeh Software 18 2017-06-23 14:26
Intel software emulator lets you test drive AVX-512 tServo Hardware 0 2016-04-01 15:31
Anomaly after ECM report; possible ECM data base integrity problem cheesehead PrimeNet 8 2013-09-01 04:27
A primality test for Fermat numbers faster than P├ępin's test ? T.Rex Math 0 2004-10-26 21:37

All times are UTC. The time now is 22:02.


Sat Oct 23 22:02:49 UTC 2021 up 92 days, 16:31, 0 users, load averages: 1.33, 1.25, 1.16

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.