jmke

Linpack by Intel(R) is an extremely stressful program that will
put even the most powerful X86/X64 CPU in the world at its knees. Load temp
under Linpack will be up to 22*C higher than the competing software Prime95.
This program will make usage of Linpack easier and more practical.

Benefits of using Linpack:
1. More accurate than Prime95 Small FFTs/Blend.
2. Takes less time to tell if your CPU/RAM is unstable than Prime95 (usually
something like 8 minutes Linpack vs 40 hours under Prime95).
3. Use the same stress-testing engine that Intel uses to test their products
before they are packed and put on shelves for sale.

http://www.xtremesystems.org/forums/...d.php?t=197835

__________________

leeghoofd

It's a nice program to determine quickly if ya CPU can hold up BUT :

1) It's best tested under 64bit OS conditions
2) It's not that conclusive on NB and ram stability
3) Stresses CPU way to far, so users with inadequate cooling can damage their hardware
4) Claim for 8 minute testing is totally ridicilous as with 4-6Gb ram and max ram test you run what ? 5 intenerations...prgram creators tell to le the users run at least 20 itenerations ( which o my rig still corresponds to at least 30 minutes of hardware torture)

npp

My T7200 hits the sweet 100°C mark after even 3 iterations Kinda funny to see such CPU temp.

jmke

E1200 @ 22°C room temp hits 60°C under load with stock cooler's fan undervolted to 5v

__________________

Kougar

Oh, I was thinking it was a new update!


This is no different than Prime95... Prime95 only tests 32bit regardless, as does all other programs using a 32bit OS.

Unfortunately, this is true. That is what you can use Prime95 Blend for and other memory testing programs. Intelburn rules out the CPU.

CPU's will underclock and undervolt if they get too hot, and shut down the machine if they reach their max temp threshold.

If the CPU cooler can't handle IntelBurn then it is already likely getting too hot during full program loads. Stock Intel coolers can handle IntelBurn just fine.

As opposed to 24 hours of Prime95?? 30 minutes is nothing by comparison

dazzawul

The idea is, errors are more statistically likely to happen when you've overclocked, the higher the overclock, the more likely they are to happen.

Now, running a benchmark for 30 minutes is going to pick up anything glaringly obvious, but what he's trying to say is that an error is more likely to show up under less stress and over a longer period of time, than more stress on a shorter period of time.

It also explains why you can have an overclock thats prime stable but not linpack stable, or linpack stable and not prime stable.

leeghoofd

Prime also has a 64bit version


If you think running a CPU at eg 90°C is a good test, I beg to differ. There's absolutely no program that heats up your CPU like IBT (besides LinX which is another version). But is more heatoutput then a better test ? I can run IBT for eg 30 runs , yet cap out under Prime or a game in minutes. And like Dazzawul said vice versa. There's no bulletproof test program. OCCT is getting there as it combines, prime, IBT and stuff... IBT is hyped because it heats up ya CPU like nothing else and therefore for many it's seems as the ultimate test program.

For ya statement on the stock coolers, I run all my CPU's overclocked, and never ever use a stock cooler. If you test a stock clocked and boxed cooled CPU itshould pass any test program as the CPU usually runs at a voltage higher than required.

Another issue I have is that many many 45nm CPu's have borked, stuck temp sensors so many users have absolutely no idea how hot the CPU really is. You can say it starts to throttle, is that a good thing ? if you push hardware beyond it's limit and it has to throtlle and is possible lower Vcore. Still isn't a stable then to me as the cooler is inadequate, yet IBT might have completed the tests succesfully.

Well this where I have problems with all these test programs. If you read the thread ( and I followed it from the start ) you eg claim to have CPU stability with a 10 inteneration run, but your PC sometimes BSOD's or reboots, someone will tell you that a 10 run is not conclusive.Then they tell you to do a 50 run and so on. There are users that run this program with 100-300 itenerations set. Which is absolute madness. I also never have run Prime longer than 8 hours. Best stability test is daily usage. And a mix of test programs is far better then to focus on just solely on eg IBT. I use LinX, Prime Blend, Hyperpi and then the 3dmarks to get a sort of good feeling that my rig has a stable OC.