What stress testing program should I use with a Haswell CPU?

[sltech]

I just got a laptop with a 4910MQ CPU (a Haswell CPU), and I want to do some benchmarking/stress testing on it. I found some old posts saying that Prime95 and IntelBurnTest might damage Haswell CPUs. So what is a suitable program to use to stress test my CPU?

 

[techne]

I just got a laptop with a 4910MQ CPU (a Haswell CPU), and I want to do some benchmarking/stress testing on it. I found some old posts saying that Prime95 and IntelBurnTest might damage Haswell CPUs. So what is a suitable program to use to stress test my CPU?

Hyper PI worked here decently.

 

[LTC8K6]

Intel XTU has nice stress tests.

 

[BonzaiDuck]

Intel XTU has nice stress tests.

This is always a fascinating topic. If I use XTU on my Sandy Bridge systems (ancient!), it is for "casual" stress-testing. What passes in XTU for the SB-K CPUs may easily fail under IBT, LinX, OCCT or Prime95. Aida64 also offers some decent stress-testing like XTU, but with the same failure prospects for the other tests.

Of course -- relatively speaking -- Haswell is a "whole new ball-game." As is Skylake, for that matter . . .

 

[LTC8K6]

My thinking is that you probably aren't going to see anything worse than the XTU stress test and memory test in the real world use of your computer.

Although I have never built a computer that failed with IBT or LinX, etc.

They seem to get the CPU hotter than it will ever get in real world use.

I also tend to use XTU as a heatsink test. If it stays cool with XTU, then the cooler is fine for reality.

 

[BTRY B 529th FA BN]

Asus Real Bench is a good stress test

 

[Nhirlathothep]

i can t find a reason to stress your cpu, but if you wanna do it, you can use linx.

but why? don t do it

 

[sm625]

I been kind of liking silverbench lately

http://silver.urih.com/

Mainly because I can just google it and start running it without having to download anything.

Then I end up downloading hwmonitor anyway lol. Because you need some sort of decent temp monitor.

 

[cytg111]

If you want to test throtttling then prime95 avx2 edition.
(We dont need to have another discussion as to if that is a relevant workload or not, you know, calculating primes... its all good.)

 

[BonzaiDuck]

My thinking is that you probably aren't going to see anything worse than the XTU stress test and memory test in the real world use of your computer.

Although I have never built a computer that failed with IBT or LinX, etc.

They seem to get the CPU hotter than it will ever get in real world use.

I also tend to use XTU as a heatsink test. If it stays cool with XTU, then the cooler is fine for reality.

Remember Dan Brown's "Digital Fortress?" Someone else -- an Anandtech colleague in Poland -- called these programs "power viruses."

And it occurs to me that Intel itself would know what reasonable real-world stress would be for their processor, then adjust their specs to insure it by enough to greatly diminish any residual chance of failure.

On the other hand, Intel could not anticipate the subsequent release of game or other programs which might exceed their expectations, but the software developers would probably do that. Why market a game if only a segment in the already-exclusive realm of Enthusiasts can play them on their 12-core Xeons or Broadwell-E processors?

But there is thermal stress, and simply the stable and reliable operation of CPU components, following spec or OC'd situations. OCCT's own "CPU" test runs cooler than Linpack stressors. But it will discover instability faster than those latter, and most profoundly faster than Prime95 sFFT or Blend.

 

[Ranulf]

OCCT

http://www.ocbase.com/