Intel XTU benchmark causes coil whine

[silviu2008]

Hi everyone,

I just finished my build with a 9900k and a Gigabyte Z390 Aorus PRO motherboard. I did not OC the CPU yet, I did a lot of stress testing using various programs (Intel XTU stress test, Realbench, Prime95 etc) and also ran Memtest for couple of hours and everything seems to be working just fine, no BSODs/freezes whatsoever. I use a manual voltage of 1.23 volts along with a NH-D15S cooler and temps are OK (hit 90 degrees for split seconds while stress testing with AVX), and will not get hotter than 70c in games.

This is where it gets interesting: whenever I use Intel XTU Benchmarking app (and only this, other applications or games that I've tested so far will not cause anything) my mobo generates coil whine (I guess it's the VRM?). It's exactly like in this video:

. It also seems that the benchmarking generates random restarts or BSODs (like 1 in 10 runs). So what would be the issue? Since I successfully ran prime95 for like 20 hrs, I don't think that my BIOS settings are unstable. Could be that the benchmark app has issues/bugs? It's also the latest version. I am curios to see if someone else had this issue and if they managed to fix it somehow.

Thanks.

PS: i really didn't have time to mess that much with Bios settings, pretty much everything is on default settings except for XMP which I enabled and I manually put CPU vcore to 1.23Volts. I will try today to disable C-states or maybe mess a bit with CPU voltage to see if I still have the issue. And I also remember someone saying that using only one 8 pin (my mobo has 8 pin + 4pin eps) for CPU would cause coil whine. I will try to connect the 4 pin aswell.

 

[Ferrum Master]

Just accept it. It is a normal behavior, it is by design.

Get a better case, sound isolation and quit doing XTU and use the PC for proper things like productivity or gaming.

 

[silviu2008]

Yea I know that coil whine won't do any harm, however I don't understand why it only happens with Intel XTU Benchmark app. And to be honest I was more concerned with the random BSODs and restarts which happen only when using this app. Thanks for the reply

 

[Ferrum Master]

Yea I know that coil whine won't do any harm, however I don't understand why it only happens with Intel XTU Benchmark app. And to be honest I was more concerned with the random BSODs and restarts which happen only when using this app. Thanks for the reply

Because of the current flow. It does a task for certain amount of time. Does again. The loop period 1/T = frequency Hz. It is the specific data task.

BSOD? Your system is not stable. Then. Up the voltage or at least the vdroop. Just as always.

Download the latest linpack xtreme here at TPU. It should bsod also. For example for me if the system was not stable it passed first test with all 32GB RAM. It failed after an hour or so.

 

[Grog6]

Put a couple of drops of fingernail polish on the coil that's making noise; it should quiet it.

If the fingernail polish catches on fire, I'd back down my OC.

The sound is the wire moving.

Changing the phasing in the bios, if you have that option, may help.

 

[Bones]

Yea I know that coil whine won't do any harm, however I don't understand why it only happens with Intel XTU Benchmark app. And to be honest I was more concerned with the random BSODs and restarts which happen only when using this app. Thanks for the reply

It's due to the load placed on the circuitry when running XTU - It is a stress test you know.

As long as you're running XTU, get used to it.

 

[TheMadDutchDude]

XTU will degrade your CPU. Stop running it. It’s a one time stress test and then you leave it alone.

 

[Solaris17]

Yea I know that coil whine won't do any harm, however I don't understand why it only happens with Intel XTU Benchmark app. And to be honest I was more concerned with the random BSODs and restarts which happen only when using this app. Thanks for the reply

So you ran a stress testing tool, and your system restarts and blue screens. And your asking what’s wrong with the software.

 

[R-T-B]

XTU will degrade your CPU.

I don't see why if you are running within specifications and at a stable setup

So you ran a stress testing tool, and your system restarts and blue screens. And your asking what’s wrong with the software.

Basically yes but let's not get mad at him for reaching the wrong conclusion unless he denies it AFTER being kindly corrected.

 

[TheMadDutchDude]

Trust me... it degrades your CPU. It runs an extremely stressful AVX load on your CPU and will cause damage. I have seen it first hand when pushing for the two core record on a locked chip in the past. A chip that once needed 1.35v for 5G needed over 1.5v just to boot at 5G after 100+ runs of XTU at 1.15v (stock volts). It WILL damage your CPU.

Feel free to test it for yourself, though. Don't say I didn't warn you.

 

[R-T-B]

Feel free to test it for yourself, though. Don't say I didn't warn you.

I have but thanks. Also wasn't factoring in an OC though which probably makes your point valid.

 

[TheMadDutchDude]

That was stock, fella. You couldn't OC an i3 back then. It was at 102.9 BCLK with stock volts. I wouldn't deem that an overclock.

I'm talking 100+ runs, not just a one or two time thing. I just don't encourage using the tool at all.

 

[R-T-B]

That was stock, fella.

That's what I was saying yes. Even with OC: Lots of people test with Prime95 AVX and haven't reported issues, so, I'm doubtful, but ok.

 

[mstenholm]

And I also remember someone saying that using only one 8 pin (my mobo has 8 pin + 4pin eps) for CPU would cause coil whine. I will try to connect the 4 pin aswell.

I heard the same. Did connecting the 4 pin solve the problem?

 

[silviu2008]

So you ran a stress testing tool, and your system restarts and blue screens. And your asking what’s wrong with the software.

Yea actually, software (especially drivers) may cause instability and blue screens. Since I've ran other stress testing tools (including Prime95 AVX) for multiple hours and nothing bad happened, I don't see why this particular benchmarking software would crash my PC in like 5 seconds.

I heard the same. Did connecting the 4 pin solve the problem?

Unfortunately it did not fix the problem. I guess the power draw is huge in this case so that causes coil whine. I stopped worrying about it, I've had hours and hours of gaming sessions and CPU intensive tasks and I did not have any issues whatsoever. I just don't use the XTU benchmark anymore.

 

[phill]

XTU will degrade your CPU. Stop running it. It’s a one time stress test and then you leave it alone.

Trust me... it degrades your CPU. It runs an extremely stressful AVX load on your CPU and will cause damage. I have seen it first hand when pushing for the two core record on a locked chip in the past. A chip that once needed 1.35v for 5G needed over 1.5v just to boot at 5G after 100+ runs of XTU at 1.15v (stock volts). It WILL damage your CPU.

Feel free to test it for yourself, though. Don't say I didn't warn you.

That was stock, fella. You couldn't OC an i3 back then. It was at 102.9 BCLK with stock volts. I wouldn't deem that an overclock.

I'm talking 100+ runs, not just a one or two time thing. I just don't encourage using the tool at all.

I think we nicknamed it the CPU killer didn't we mate?? That i3 never stood a chance..... Still at least we had it RMA'd in time...

There's much better tools out there to test stability I can assure you and if you've run Prime it's fine

 

[TheMadDutchDude]

Yep! Everyone knows it. Hence why we run it in slow mode when on LN2...

Whatever, I apparently don’t know what I’m talking about.

I need someone with scripting experience to help me write a very basic script. I’ll put this to the test and document my findings.

 

[phill]

Yep! Everyone knows it. Hence why we run it in slow mode when on LN2...

Whatever, I apparently don’t know what I’m talking about.

I need someone with scripting experience to help me write a very basic script. I’ll put this to the test and document my findings.

Wasn't it something to do with the AVX/2 instructions it did (to show off those Intel CPUs being so much faster this gen than last gen??) or something or other?? Either way, crap bench in my opinion. Definitely prefer something like RealBench from Asus or PC Mark even to test a PC's stability..

 

[R-T-B]

Whatever, I apparently don’t know what I’m talking about.

Never said that. Just know I bench/stress test my 9900k using AVX2 XTU/Prime95 all the time and have yet to see an inch of degradation. I run a more mild overclock though (5Ghz All-core/uncore)

Maybe it's a generational thing. I do hear X99 had very touchy caches with voltage applied, no idea what you were using.

 

[DR4G00N]

There's much better tools out there to test stability I can assure you and if you've run Prime it's fine

FWIW, XTU's benchmark tool is Prime95.

What can kill the cpu's is the constant transition from idle to heavy AVX load and vice versa when you run it over and over again.
Though, the i3 6320 I had was used by the previous owner and then I to do 743pt XTU bench runs and it was still perfectly fine when I sold it.

 

[phill]

FWIW, XTU's benchmark tool is Prime95.

What can kill the cpu's is the constant transition from idle to heavy AVX load and vice versa when you run it over and over again.
Though, the i3 6320 I had was used by the previous owner and then I to do 743pt XTU bench runs and it was still perfectly fine when I sold it.

Well I shouldn't have been surprised at that but I am Prime uses AVX instructions doesn't it?? Or newer versions of it?

I gave up with the XTU bench when they released B Die at the time.. I was wasting hours doing it and then when it was struggling to pass the same speeds at higher volts it just was time to throw the towel in...
Maybe Brend and I had a bad batch of CPUs or even OCF board (I would always share my hardware ) but who knows... Just wouldn't use it myself for testing stability anymore. As I said previously, there's loads of other tests out there to try

 

[jaggerwild]

The question people should be asking is why r u using an AIO(cheap water cooler) on a top of the line CPU? Id suggest upping the V core but who care's as stated XTU is known to cause damage, preformed isn't really water cooling...............IT junk sold to make the company money off people scared to make a loop(cause it might leak!). Also while stress testing or not you should have active cooling on the VRM'S, At Least a good sized Fan.