• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Intel Burn Test

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Beware of ur cooling milomak, IBT will stres ur procie with high temp. I use IBT v.2.5 while testing my i5 750 at 4.2ghz it went to 77c with venomousX RT.... i saw review venomousX performance is abit same with corsair H70, now u have corsair H50 i guess ur temp will be higher than H70.
 
Sorry u use linux OS hehe i though microsoft windows ^_^ mmm dont have info about linx OS, just googling IBT and read it, see if IBT can works in linux OS, or not.
 
I am not sure exactly how IBT works but Phoronix has a couple cpu tests in it. Will run CPU at 100% for as long as you would like.

Link
 
The thing is, 100% CPU usage does not mean full heat output. Compare Prime95 and IBT on Windows. Massive temperature differences and they both use "100%". The programs are not created equally and there is a reason we use LinX/LinkPack.

There is a Linux version, but I can't seem to locate it.
 
This is just a guess but if you write a loop to do 1+1 for some ridiculous amount of times you can utilize 100% of the CPU cycle but not actually use 100% of the cpu... am i close?
 
Basically, 100% of the CPU is based on how busy the processor is. You could write a program that uses 100% of the CPU time, but does absolutely nothing and temperatures would be near or at idle.

LinX/LinPack/IBT attempts to use as much of the processor at the same time as possible to increase heat output.
 
A page fault (something that isn't in cache, so the CPU has to send off to get it for ram, a process that takes hundreds to thousands of cpu cycles) leaves the cpu 100% utilized, but doing nothing.

To monitoring software there is no difference between constant 1,000 cycle page faults (so operating maybe 10 cycles per 1000) and doing short integer/floating point math that does an operation every cycle.
Both show up as 100% usage, but the page fault version is only actually doing something 1% of the time, while the short math version is doing something more like 90-95% of the time (getting stuff from L2/L3 (especially L3) takes time too).

An even nastier program would do both integer and floating point math at the same time, most modern CPUs do those two types of math in different parts and hence can do them at the same time if neither process requires any of the same data/registers.
 
Back