• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

stress testing these days

M

Momo

Gawd
Joined
Mar 3, 2005
Messages
869
What do we use to stress test a CPU these days?
I've got a fresh Win10pro install, and when i run prime95, using the blend test, i get an immediate win10 BSOD.


System is otherwise stable.

No overclocking at all, default bios values. Running i7 6700k, intel PCI-E M.2 SSD, and QVL RAM for this motherboard.

Anyone?
 
anyone? system is stable otherwise. Hours of gaming is fine.

Prime95 bug/issue?
 
I like AIDA64's stress tester for synthetic testing.

Outside of that, I'll queue up a couple blu-ray conversions on Handbrake...if it can take converting a couple movies in a row, I'll call it stable.
 
I used Asus RealBench and AIDA64 Stress test. CineBench seems like a good bet too. Real and Cine use actual video encoding for the bulk of their stressing, so if you're stable in those tools, you'll be good to rip and encode to your heart's content. AIDA is synthetic, but also more flexible. You can choose what system to stress (including whether or not to try out the power hungry AVX floating point instructions) or do whole-system.

I don't use Prime95, because I don't expect to be using my computer to find prime numbers. It's not a relevant workload for me. Some others use it as the high standard for stability. They're not wrong. I just don't need it. I encode vids, play games, mess around in Linux VMs in Hyper-V and occasionally compile stuff. I've never crashed while doing any of that and my stressing consisted of 1 hour of RB and 15 minutes of AIDA (to check temps mostly).
 
Prime95 is a stress test really for AVX loads. It will test your cooling/voltage/thermals to the max. If you can find a prime95 stability, odds are the others will be stable too. I used p95 20+hrs, realbench 16+, occt 24 hrs to stress my system. 7 years ago, I did the same to a 920@4.0ghz and I was using it till end of Feb this year.
 
I use intel burn test V2 by Agent GOD. Works great and if my cpu overclock and voltages have any discrepancies, running this program for less then a few seconds will yield me a fail. I like using it more then Prime95

Edit-Also, with you describing your issues with blend test, I would test your ram to make sure non of your modules are faulty. Download memtest86 onto a usb stick and run it overnight to rule that out
 
gaming/ realbench some prime have to watch using newer prime because of avx can pull really dangerous wattage and push temps up, funny enough i found mass effect andromeda really demanding have seen it use 85% of my 6800k and cause freezes where the system looked stable in prime and realbench.
 
Slade said:
Prime95 is a stress test really for AVX loads. It will test your cooling/voltage/thermals to the max. If you can find a prime95 stability, odds are the others will be stable too. I used p95 20+hrs, realbench 16+, occt 24 hrs to stress my system. 7 years ago, I did the same to a 920@4.0ghz and I was using it till end of Feb this year.
Click to expand...
Depends which version you have whether it has AVX support.
AVX testing doesnt start straight away when P95 is run on Blend, his initial problem has another cause.
 
Momo said:
What do we use to stress test a CPU these days?
I've got a fresh Win10pro install, and when i run prime95, using the blend test, i get an immediate win10 BSOD.


System is otherwise stable.

No overclocking at all, default bios values. Running i7 6700k, intel PCI-E M.2 SSD, and QVL RAM for this motherboard.

Anyone?
Click to expand...
Set your memory to 2133MHz and see if it improves.
If it is already at 2133MHz, reset your CMOS and see if it improves.
Also try 1 memory stick.

I am looking to see if it can be made stable.
 
The most recent prime95 uses AVX, this is assuming you are working on a new build, you would be running the latest versions of stress test programs. The reason we run a whole host, not just 1 is test is to stress the system in a variety of ways. Prime95 with AVX will push the limits of your temp solution. Since the extensions are 256bit and 512bit, the chances for errors run high and will push the temp aspects of your CPU cooling. Memtest can be used to test stability of your ram. Realbench and OCCT are a system wide testing which will push the combination of the CPU, memory and GPU, Furmark for GPU and Intel Burn Test for another cooker.

Find your stable setting that applies to all these stress test, and you should have a decent running system you can trust.

You can't fault a stress test for finding fault in your mobo/cpu/cooling/memory/gpu/power etc because that's WHY you are running these tests, to find fault. Dismissing the tests as invalid and saying, oh I can't pass this test but everything is fine for everything else, what's the point in this? Overclocks are not guaranteed, you run this stuff to ensure that you aren't left hanging 2 weeks down the line doing something important on your pc.If a test exposes a weakness, run another test which tests similar aspects and then either crank up voltage and deal with heat or back off settings. These days PC's are a breeze to tweak, I recall having to modify traces to select voltages on CPU or redoing rewiring my geforce3 to redirect a buildup in charge from a capacitor and dump it to ground or else it would lock the system under load....

I would not consider a crash in a video game to be an indicator of instability alone as a video game has too many variables involved which can lead to instability where the "stress" programs listed above are closed systems and fairly bug free as of late.
 
Last edited:
My response was tailored for the ops problem to reduce confusion.
I'm not sure what you are doing.
 
adia64 and handbrake does not come near the stress level or fault detection of prime95.

Its more likely you system is just unstable under the load of prime95. but can take weaker loads like handbrake and aida64
it still means you system is unstable you are just ignorering/accepting it.
 
Slade said:
The most recent prime95 uses AVX, this is assuming you are working on a new build, you would be running the latest versions of stress test programs. The reason we run a whole host, not just 1 is test is to stress the system in a variety of ways. Prime95 with AVX will push the limits of your temp solution. Since the extensions are 256bit and 512bit, the chances for errors run high and will push the temp aspects of your CPU cooling. Memtest can be used to test stability of your ram. Realbench and OCCT are a system wide testing which will push the combination of the CPU, memory and GPU, Furmark for GPU and Intel Burn Test for another cooker.

Find your stable setting that applies to all these stress test, and you should have a decent running system you can trust.

You can't fault a stress test for finding fault in your mobo/cpu/cooling/memory/gpu/power etc because that's WHY you are running these tests, to find fault. Dismissing the tests as invalid and saying, oh I can't pass this test but everything is fine for everything else, what's the point in this? Overclocks are not guaranteed, you run this stuff to ensure that you aren't left hanging 2 weeks down the line doing something important on your pc.If a test exposes a weakness, run another test which tests similar aspects and then either crank up voltage and deal with heat or back off settings. These days PC's are a breeze to tweak, I recall having to modify traces to select voltages on CPU or redoing rewiring my geforce3 to redirect a buildup in charge from a capacitor and dump it to ground or else it would lock the system under load....

I would not consider a crash in a video game to be an indicator of instability alone as a video game has too many variables involved which can lead to instability where the "stress" programs listed above are closed systems and fairly bug free as of late.
Click to expand...

Here's the thing, Prime95 will crash some systems at stock settings. Systems which are certified by the OEM to run perfectly fine, and do outside of that one app. That's not realistic. That's not even useful. It's probably not good for the hardware (then again, most components will become obsolete long before they actually fail, so whatever). I will admit that GPU tests are essential for GPU OC's. That's an oversight on my part.


SvenBent said:
adia64 and handbrake does not come near the stress level or fault detection of prime95.

Its more likely you system is just unstable under the load of prime95. but can take weaker loads like handbrake and aida64
it still means you system is unstable you are just ignorering/accepting it.
Click to expand...

That's just absurd. Stability is never absolute and it is always conditional. Think your rig is rock stable sitting on your desk? Okay, what about sitting on a table, outside in the sun, at noon during high summer? No? That isn't realistic? Exactly.
 
STR said:
Here's the thing, Prime95 will crash some systems at stock settings. Systems which are certified by the OEM to run perfectly fine, and do outside of that one app. That's not realistic. That's not even useful. It's probably not good for the hardware (then again, most components will become obsolete long before they actually fail, so whatever). I will admit that GPU tests are essential for GPU OC's. That's an oversight on my part.

That's just absurd. Stability is never absolute and it is always conditional. Think your rig is rock stable sitting on your desk? Okay, what about sitting on a table, outside in the sun, at noon during high summer? No? That isn't realistic? Exactly.
Click to expand...

Never had a systme crash on stock settings and if it did you OEM made you a bad Computer. that is not something new for OEM crappy PC's. dell release a entire line with the new Pentium4 that was constant throttling under load
Prim95 is not some magic that does something your CPU is not supposed to do.. Its simple uses you CPU that is.

and correct. then you system is not built well enough for being in the sun. still does not make you unstable computer suddenly acceptable at stable.
Again beeing a failure at figure out to stress test in the enviroment you are using you system is not a fault for the software. but the person doing a petty horribly attempt at stress testing

again.. being horrible at stuff does not make it the stuff's fault... just learn how to do it well
 
I have the same "won't prime out-of-the-box" with my rig from sig, regardless if oldest or any Bios in between, default 2133 or XMP-3600, it failed prime-95 AVX any time.

The reason I think is in the IMC as raising VCCIO cured it alone for 2133 and for 3600 it also needed more vDIMM than 1.35 default XMP value.

If that is a reason to give it back I dont know. Maybe it would do it with 2 modules but not 4 as I have. My RAM is also not on QVL, so I may not blame anybody but me.


I could fix it with said Voltages and Prime95-AVX runs stable till 5.0GHz, just temps get VERY HOT and the CPU cries for a delid, which I most likely wont do as I dont prime for money and RealWorld AVX in Handbrake is way lower in temps and Volatges than prime is.

I have seen a few posts with this issue and I guess most could be fixed with more Volts.
 
You must log in or register to reply here.
Back
Top