Prime95 v IntelBurnTest

Toggle sidebar Toggle sidebar
T

TheSunSmellsLoud

Honorable
Jul 10, 2013
7
0
10,510
Hi,

I was wondering on the difference between intelburntest and prime95 for stress testing. I'm currently OCed at 4.2GHz (1.18v Core Voltage) on my 4670k, and I've tried running both programs.

IntelBurnTest pushes temperatures around 83 degrees doing 20 runs of a high stress test, and after an hour, Prime95 gets temps around 67 degrees on the blend test.

Will any application ever exceed the stress put under it by prime95? If so, will it ever reach the levels of intelburntest? As I'm not willing to overclock more if it's likely that it will reach similar temperatures to the IBT ones.
 
Solution
Roland Of Gilead


Well, different horses for different courses. Each poster will have a preference but they both do the same thing - put your CPU under full load to...
Sort by date Sort by votes
Roland Of Gilead

Roland Of Gilead

Titan
Aug 8, 2008
10,299
2,249
77,690


Well, different horses for different courses. Each poster will have a preference but they both do the same thing - put your CPU under full load to test for stability. Personally I use Prime 95, and small ffts only (this will stress only CPU and cache) the blend test won't get it to full load. I run it for 8+ hours overnight, and then I know it's stable :)


As long as it's not throwing up errors whilst your testing, you should be happy. :)
 
Upvote 0 Downvote
Solution
P

Pinhedd

Champion
Aug 7, 2007
10,342
20
56,965
Both IBT and Prime95 are similar in that they stress floating point arithmetic and memory subsystems. They are different in that IBT uses Linpack (solving linear equations) while Prime95 calculates Mersene Primes.

IBT is generally regarded as being far more aggressive in the short term, which makes it great for testing ultimate stability. IBT will easily drive load temps up to 20 degrees higher than Prime95, this is well known and is a defining feature of the program.

Unfortunately, the Linpack benchmark was designed for supercomputers (hence the floating point part, for modeling continuous phenomenon) so it really pushes desktops to the limit, far beyond what any application will do. This means that IBT may fail on commercial CPUs that are running at stock settings simply because Intel doesn't test them to that extent.

Some applications such as games with substantial amounts of software physics may reach Prime95 load levels. However, almost nothing at all will ever reach IBT levels. The only exceptions that I can think of would be Bitcoin mining and various physics simulations.
 
Upvote 0 Downvote
boju

boju

Titan
Ambassador
Jul 7, 2008
20,681
2,212
85,890
There are lots of different opinions on which and how long for.

Pretty much intelburntest for a minimum of 10 runs maxing out the most memory available. An unstable overclock will fail within the first couple of tests. If you make it to 10 then its quite stable and if you've made it to 20 its a good sign and is probably set in stone solid.

Running prime95 overnight or even 48hrs will help find any tiny instabilities intelburntest wont find, I have read. You can set the priority higher, default is 1, goes upto 10 using more memory. Differing opinions here also, some say its quicker raising the value and some say and even the program's information on this says if you leave it over a period of 24hrs or more, leaving it on 1 will do the job.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom