Intel burn test ?

the_electrician

the_electrician

the_electrician

Associate
Joined
5 Jan 2009
Posts
229
Why does ibt fail after 2 runs running 8 threads even at stock speeds?

8 threads stresses the cpu to 100% while 16 threads puts approx 70% load on the cpu.
So why do people assume there cpu is stable after 20 runs as the cpu has not been fully loaded?
 
When I used to use IBT, it was the opposite way round to what you have described, 8 threads would only put 70% load on the chip, whereas 16 threads would max it out..

I prefer to use LinX these days.
 
I prefer to use LinX these days.[/QUOTE]

how long would you advise running linx?
I cant run my mc overnight thats why i thought id give ibt a go as the consensus is that 20 runs is sufficient.
 
the_electrician said:
how long would you advise running linx?
I cant run my mc overnight thats why i thought id give ibt a go as the consensus is that 20 runs is sufficient.
Click to expand...

LinX is basically the same thing as IBT, as they both use Linpack to stress the CPU, so 20 passes using all available RAM should be good enough.

I generally use a mix of LinX, Prime95 (blend test) and day to day use to test for stability.
 
Some versions had issues with 8 threads failing under any circumstances, I'd have thought this was fixed by now. I'm using 16 threads, maximum available ram, loading at 100% quite happily. Takes about 5 hours to run through 15 loops.

Respectfully suggest that your systems are failing is a sign of instability, consider redownloading ibt and reinstalling windows to confirm.
 
I can confirm IBT has an issue running 8 threads. After sorting my duff ra issue I still had problems getting stability until I found out about using 8 threads.

CAn't recall what version I have but 8 threads stresses cpu 100% but always fails on third pass. Same settings using 16 threads and it passes 30 (eventually - takes a while... ;) )
 
I have been doing some testing using this program this morning. At 191x21 the system crashed twice on standard, so I decided to turn off that option in the BIOS (forgot what it's called) and tested again. So far it's passed Standard, High, Very High and Maximum using 10 loops. I left the threads on auto, as I didn't know the correct settings for manually choosing it.

Anyway, here are the results I got:
2jbqo3b.jpg


Should I test again using a specific thread setting?
 
looks like thats only using 4 threads, change setting listed as auto to 16 and run task manager at same time to check cpu is running at 100% load. Or download real temp which shows core temps as well as loading in real time.
 
15 loops takes five hours for me. Passed this on maximum memory quite happily, crashed under normal use. I conclude that either ibt is inadequate or that five hours of it is inadequate.
 
20 runs of IBT takes a couple of hours for me, i always run maximum ram on 16 threads (set manually) ive tried LIN X, but it doesnt seem to generate as high coretemps as IBT, i also use p95 small fft and blend for 8 hours each as theyre my old favourite test utilitys. i particularly find blend to be very useful as i do a lot of video encoding.
 
Thi shappened to me occasionally. I presume it's IBT's way of crashing, I put it down to it trrying to access more memory than was available mid run - You click on max memory available but as IBT spools up the ram another program in the background pinches some.
 
I found the only way I could pass 20 runs was to lower my ram usage to under 100Mb, which is ridiculous. Have been running orthos for the last two days so when I get home I'll see if it made it through that cleanly and try reinstalling IBT
 
You must log in or register to reply here.
Back
Top Bottom