Prime95 or intel burntest?

awsomeness

awsomeness

awsomeness

Associate
Joined
18 May 2009
Posts
93
Hi im just wondering which is the best to do? im currently doing a 8 hour prime test and wondering if i should do a intel burn test after? my current 100% load temps are - 66 - 74*c (i think my cooler needs lapping because my temps are kinda like 66, 69, 70, 74)
 
Prime 95 for northbridge vs memory.
Intel Burn test for cpu temperature world records, and melting your plastic watercooling system tubes.

The thing about Intel burn test temps, is that you will never see them in real life. Where as prime 95 + leaving a single core running something like furmark, is a fair approximation of running games.
 
Last edited:
Agree with the above, do both.

Prime95 for testing errors and stability if you really have the time, and Intel Burn Test for checking how well your CPU cooler is doing it's job.

Personally for me if it lasts 15 minutes or so for something like IBT, or even OCCT it is more than enough.
 
so far I've found that if it does 5 passes of IBT the system is rock solid stable.

I have never liked prime95 as I've had 12 hour sessions that have crashed out on me in a game, and I've had totally stable systems that only crashed on prime95.

Some people just love their prime sessions though......
 
just wondering how do i use memtest? what do i download and can i run it in windows? or do i need to burn it onto a disk and load it on startup? also where can i download the latest version of memtest :p?
 
awsomeness said:
Prime95 or intel burntest?
Click to expand...
Both . . . at the same time! :p

36ghzulv.gif


"Everyone wants to go to the party . . . but no one wants to stay behind and help clean up the mess!"

I like that saying and it can be ported across to the wonderful world of overclocking too. It seems there are a lot of overclockers who don't care much for stability testing, and lets be honest who can blame them. The process of torture-testing and subjecting your machine to ludicrous synthetic workloads is pretty boring and can verge on being tedious, I take no pleasure in it myself but personally I approach overclocking from a scientists viewpoint, my computer is the lab where I run my experiements and I do my absolute best to make sure any results I publish are good and tested to an extreme level.

Prime95 and IBT are both great tools for testing an overclocked or tweaked rig but they only target certain parts of the system so as already mentioned its possible that a machine can pass over 24 hours of Prime and 100's of IBT loops yet still *CRASH* or *REBOOT* after 5 minutes of gaming :o

The screenshot above just happens to be what I am doing atm, running some undervolting testing on a new(used) E8400. After fifty hours of Prime testing with a 50 loop run of IBT thrown in gives me good confidence that the machine is stable for number-crunching tasks only. The next level of testing will be to throw as much 3D-Rendering at the machine as I can, looping 3D-Mark overnight and as many hours of gaming as I can squeeze in! :p

I normally am happy with 24 hours of prime so don't think 50 hours is normal however on the above clock the system was *REBOOTING* after about 15 hours of Prime which I finally tracked down to a slight lack of vTT voltage. It had done this a number of times before I made the adjustement so I just put Prime on and left it for a few days while I played on another machine!

I wish more overclockers would take their time to really give their overclocked systems a thorough testing before publishing half-baked results (normally followed some time later by a retraction "Oops it just crashed!).

The net result of a Methodical testing proceedure before sharing results would be a steep decline in frustrated overclockers who have built up uBer *expectations* from having seen tons of screenshots from machines that were in fact not 100% stable.

Prime95 and IBT testing alone do not guarantee a 100% stable system but it sure narrrows it down and gives a fair bit of credibility that rock solid stability is just a few tweaks away! :cool:
 
I admire your systematic approach, which is great when you have the time, which is something that's essential when it comes to getting a good OC. If only there was a 'set it and forget it' solution to overclocking, though if it were that easy everybody would do it.
 
Personally i prefer p95, i first test an oc on small fft, to ensure stability (8 hours minimum)then i test ram settings with blend, same test time. IBT is a quick test that doesnt represent my pc usage in a given day.
 
As some have already stated, its all about time management.

I absolutely hate having to put my PC through endless hours of primes and burn ins just to check that its stable. So when I discovered IBT, it was a breath of fresh air.

I dont know what settings others are using that mean that prime is stressing the whole PC more than IBT, but if you use the right test, nothing imo stresses your system more than IBT. I had a PC that was prime stable 10 hours and it failed in IBT (sure it completed it, but failed it). Other overclocks would boot into windows play all my games etc but start IBT and after a few minutes the PC would show symptoms of instability. This is a marked improvement than having to wait 3-4+ hours to show any signs with prime etc.

Although the heat produced is excessive and no real software will cause that sort of heat, I find the test better because its quicker to find a sign of instability.

I mean sure, if the overclock passes IBT then maybe you can run prime for 10s of hours to check stability with more chance of success. In the past, prime was mostly it really, so it was a tedious cycle of making a bios change, booting up, everything seems okay, run prime then it fails after 3 hours, going back to bios, making another change, booting into windows, everything seems okay, prime fails now after 5 hours, etc, the cycle continues.
 
When i had my q6600 at 3.8ghz, it passed 8 hours on small fft and blend, tried ibt and all it done was up the temps a bit, the system was as stable as i needed it to be, all IBT done was generate a bit more heat. Now on a q9550 @3.4ghz on stock volts (1.200) so i may give IBT another run.
 
I ran 8 hours prime when I got my system, as it seemed like a decent amount of time without taking me away from my new toy for too ridiculously long.

I figure that makes it stable enough for most normal use, and I can leave it running sometime from going to bed to coming back from work (best part of 20 hours) when I get a chance, without needing to urgently, just to make sure it's bulletproof.

Apparently the new IBT may be AMD-compatible, which should make life more fun - a good point is raised above that IBT tends to fail faster, making it an excellent indicator of unstable systems.

The faster you can make it fail, the faster you can get on with it. Once it passes IBT, it's time to give it a proper torture test.
 
Anyone else run furmark at the same time? I do so because my cpu and gpu are in the same loop, but it is probably good practice to stress the graphics subsystem at the same time as the cpu anyway
 
You must log in or register to reply here.
Back
Top Bottom