Stress Test Powers Machine Down

SwartSkaap

SwartSkaap

SwartSkaap

Associate
Joined
22 Nov 2013
Posts
12
Location
UK
Hey Guys,

I am currently having an issue where my PC turns off during a stress test using RealBench. After 4-5mins, its pretty consistent. The GPU is new. I ran the stress test as the machine was randomly powering down while streaming and playing games. Now I can reproduce the power down with the stress test.

I am not trying to OC the machine, just to get it running at stock to begin with. It was running an MSI GeForce GTX 960 Gaming 2048MB GDDR5 PCI-Express Graphics Card before the 1080ti below with no issues. The powering off only started when I bought the 1080ti.

Here are my PC Specs:

CPU: Intel i7 6700K @ 4.0GHz
MotherBoard: Gigabyte Z170X-Gaming 5 BIOS Version F21.
Memory: 2 x 8GB DDR4 (Corsair Vengeance LPX 16GB (2x8GB) DDR4 PC4-24000C15 3000MHz Dual Channel)
GPU: Nvidia GeForce Aorus GTX 1080ti (GPU Boost Clock 1683MHz | Memory Clock 11010MHz)
PSU: Corsair GS600W

- XMP Is enabled, was disabled, but I enabled this last night. It hasnt affected the power off issue but it has improved performance of my memory, as it was running at 1066MHz now idles at 1500MHz.
- CPU Vcore set to Normal (1.230v).
- Dynamic Vcore (DVID) set to +0.070v.
- All other CPU Voltage options set to auto.

Power Loading Feature
Now this board has something called Power Loading which acts like some power load balancer type thing, not entirely sure. It has 3 settings Auto/Enabled/Disabled. When the PC powers off in the stress test, it is set to Auto. If i set it to Enabled the stress test completes and there is no power off, BUT parts of the stress test seem to crash, the rendering app LuxMark i think its called, stops responding and the monitors flicker, both of which doesnt happen on the Auto setting.

Now the test completing doesnt fill me with confidence as parts of the test crash out themselves.

Worth noting that changing the setting this feature has an effect on the stress test:

Auto - No screen flicker, all apps run ok, but powers off after 4-5mins into the stress test.
Enabled - Screen flicker, Luxmark App within RealBench test crashes, but no powers off after 4-5mins into the stress test, and stress test completes "successfully" after 15mins.

Here is a youtube video of the Power Off (Power Off actually happens at 4m30secs) as it normally happens during a stress test. Ignore the sound its all background noise from the family.

Motherboard Manual

Currently the Power Loading is set to enabled and I am going to run some streaming/games and see how it behaves.

My thoughts are to replace the PSU I have with this one and see if that resolves the issue, but everything I have read states that my current PSU should be ok, unless of course it has a fault somewhere.

Ive never had to adjust voltages etc before to get things working/stable. So I am new/novice to all this and would appreciate any assistance from someone who has more experience with this.

Kind regards,
Swart Skaap
 
Last edited:
@SwartSkaap

Unigine Valley is a good GPU stress bench mark, doesn't place to much stress on CPU .
could load that with intel Xtreme Benchmark and will load system quite high ..

have a feeling it may be power related .

you able to save your Bios profile. then hit restore default values and then save exit - and try benchmarks again to see if OC or any settings are making it unstable . If it still happens at default/stock bios mode, again think it might be power related

My basket at Overclockers UK:
Total: £334.66 (includes shipping: £11.70)
​
 
Last edited:
@orbitalwalsh

Thanks for your reply.

Both UV and IEB have been run, no issues. Trouble is I dont believe they stress the PC as much as Realbench does.

I did the default BIOS thing and it was the same. Its not the OC settings making it unstable. I think i need to tune the OC settings TO make it stable.

I will be getting a Corsair RM650x at the end of the month, see if that makes any difference.

Would appreciate any other suggestions/insights.

Regards,

Swart Skaap
 
SwartSkaap said:
@orbitalwalsh

Thanks for your reply.

Both UV and IEB have been run, no issues. Trouble is I dont believe they stress the PC as much as Realbench does.

I did the default BIOS thing and it was the same. Its not the OC settings making it unstable. I think i need to tune the OC settings TO make it stable.

I will be getting a Corsair RM650x at the end of the month, see if that makes any difference.

Would appreciate any other suggestions/insights.

Regards,

Swart Skaap
Click to expand...

was trying to see if it would power down as you mentioned it did it while gaming . Real Bench stresses system more then any game or 90% of application can - not the best for fault finding and what isn't stable in real bench is normally stable with everything else. Most of us set an AVX offset in overclocks for this reason.

Did the system pass Real Bench with everything set to default ?
 
@orbitalwalsh Fair enough, that's actually what I was thinking.

Would be great to get it to realbench ok though, technically the hardware should be able to handle it with the right settings?

Was gaming last night and no power down, but it doesn't do it every time, going to carry on though as see where i get to.

I like the IEB interface, seems pretty simple to configure the things!

HWinfo doesnt report any lack of power issues tho.
 
SwartSkaap said:
I'll let you know.

Trouble is, if it doesnt solve it, I dont know what settings the BIOS will need.
Click to expand...

great thing about forums, members will be able to help you out, ive got z270 gigabyte mobo- though not powered up so can try and recall setting i've used. others are solid overclockers on here as well as using Gigabyte boards and then if it need some serious help, sure rep can pitch in if needs be
 
orbitalwalsh said:
great thing about forums, members will be able to help you out, ive got z270 gigabyte mobo- though not powered up so can try and recall setting i've used. others are solid overclockers on here as well as using Gigabyte boards and then if it need some serious help, sure rep can pitch in if needs be
Click to expand...

Thanks, I do feel I came to the right place :P
 
SwartSkaap said:
PSU: Corsair GS600W
Click to expand...
How old is that PSU?
That line seems to have been released half dozen years ago and despite of Corsair pretending to be automatic quality brand it has cheap capacitors in secondary.
So might be now clearly less than 600W capable PSU if you're enthusiast user with lots of power on hours for PC.
And some factory overclocked 1080 Ti can certainly put lots of load on PSU.

You could try running simultaneously CPU heavy program (Intel Burn Test/LinX or Prime95) with some GPU stress test to see if behaviour/symptom changes.
Also have you checked Event Viewer for any kind errors for time of those crashes?
And is that case light (showing in video) always on when PC is connected to power/wall?
Because if that light is illuminated only when PC is powered then it's not actual hard power down but some crash.

SwartSkaap said:
HWinfo doesnt report any lack of power issues tho.
Click to expand...
There's no reliable way for any software to tell that some instability is caused by struggling PSU.



orbitalwalsh said:
not the best for fault finding and what isn't stable in real bench is normally stable with everything else.
Click to expand...
If something fails in full load stress test then it certainly isn't 100% stable.
I mean would you consider it functional car if you could drive it only at 60kmh while 100kmh would break its engine fast?
 
EsaT said:
How old is that PSU?
That line seems to have been released half dozen years ago and despite of Corsair pretending to be automatic quality brand it has cheap capacitors in secondary.
So might be now clearly less than 600W capable PSU if you're enthusiast user with lots of power on hours for PC.
And some factory overclocked 1080 Ti can certainly put lots of load on PSU.

You could try running simultaneously CPU heavy program (Intel Burn Test/LinX or Prime95) with some GPU stress test to see if behaviour/symptom changes.
Also have you checked Event Viewer for any kind errors for time of those crashes?
And is that case light (showing in video) always on when PC is connected to power/wall?
Because if that light is illuminated only when PC is powered then it's not actual hard power down but some crash.

There's no reliable way for any software to tell that some instability is caused by struggling PSU.




If something fails in full load stress test then it certainly isn't 100% stable.
I mean would you consider it functional car if you could drive it only at 60kmh while 100kmh would break its engine fast?
Click to expand...

if you've been to dreamhack/EGX , you'll see quite a lot of vendor only rigs on display, overclocked - they certainly aint run asus real bench for 24hr let alone 2 ;)
also games don't run AVX2 like asus realbench - half of the time users dont use offsets and it crashes due to this, but would be 100% stable for Gaming/ web browsing and VR porn :D

do agree though its worth passing 30 mins off it - think i've given up the days running it stable for 24 hours haha
 
EsaT said:
How old is that PSU?
Click to expand...

Bought it from OC in Feb 2013, and yes I do leave it over night most days.

EsaT said:
You could try running simultaneously CPU heavy program (Intel Burn Test/LinX or Prime95) with some GPU stress test to see if behaviour/symptom changes.
Click to expand...

What does this prove over the RealBench test?

EsaT said:
Also have you checked Event Viewer for any kind errors for time of those crashes?
Click to expand...

Yes, nothing that sheds any light on the issue.

EsaT said:
And is that case light (showing in video) always on when PC is connected to power/wall? Because if that light is illuminated only when PC is powered then it's not actual hard power down but some crash.
Click to expand...

When the PC is off, blue light is off too.

When it crashed in the video, the light goes out, then comes back on and stays on as you see in the video until I reset by holding in the power button.

If you like I can get another video of the light while the test is being done?

EsaT said:
There's no reliable way for any software to tell that some instability is caused by struggling PSU.

If something fails in full load stress test then it certainly isn't 100% stable.
I mean would you consider it functional car if you could drive it only at 60kmh while 100kmh would break its engine fast?
Click to expand...

This is my thought, sure it can run the games etc, but I couldnt live with it crashing out like that, knowing that something can be done to fix that.
 
SwartSkaap said:
Bought it from OC in Feb 2013, and yes I do leave it over night most days.


Yes, nothing that sheds any light on the issue.

When the PC is off, blue light is off too.
Click to expand...
Then there's good reason to consider that fake-"high quality" Corsair having reached its design goal of failing stage.
(you bought it thinking it as high quality PSU, right?)

Wearing of capacitors causes increase of ripple which can't be measured by multimeter, or even less by some software.
That makes it very insidious problem because too high ripple puts greater stress on PC parts and VRMs of CPU and GPU.
So before visible symptoms appear PSU's capacitors going bad could have decreased life of PC's parts.
High load spikes can trigger visible symptoms long before they would appear in normal use...
But it still hard to know if ripple has stayed in reasonable levels.

And that's why I'm critical towards PSUs using capacitors chosen for cheap price...
Cougars are such PSUs.
Using quality capacitors in 5V standby to avoid PSU developing cold boot problems as "early warning" is just hypocrisy.


Lack of errors in Windows certainly hints it's not driver/software kicking the bucket, but more like hardware "hard" crashing.

And that light staying powered hints that it isn't PSU's overcurrent/voltage protections triggering.
Because clearing triggered PSU's protections should need power cycling whole PSU.
(ATX spec requires them to be latching type)
And if PSU didn't follow standard and reseted itself that should cause restart of whole PC instead of leaving it to some crashed state.


So with age of that PSU would certainly recommend ordering new in days instead of weeks.
And Seasonic Focus Plus is where top performance/quality meets the price.
https://www.overclockers.co.uk/search?sSearch=seasonic+focus+plus
Unless you prefer to pay more for brand who also fools people into buying standard stuff as quality product...

Antec EarthWatts Gold Pro has same insides as Focus Plus Gold with semi-modular cables.
And for lowest price with fixed cables BitFenix Formula has quality capacitors unlike that Cougar GX-F.

Though considering price of your graphics card Seasonic Focus Plus certainly can't be expensive.
More like Seasonic Prime serie Platinum/Titanium efficiency PSU being in same category...
 
If your worried about piece of Focus unit, just search for Riotoro G2 . Though this rolled into the UK and was cleared off the shelf with Distribution pre-orders .
Should be about £20 off the Focus Price , just finding it :(
Handy now they increased to match OEM 10 year warranty, was going to be 7 to highlight lower price tag by changed their minds
 
To give you an idea, 850w Riotoro G2 is cheaper then the focus 650w :)
Though hard to find, again same units just rebranded . I have a feeling they arent reaching resellers anymore :(
That or Seasonic isn't to pleased with agressaggr pricing
 
Quick update,

I have bought the SF 650W Platinum, I have a new case and am in the process of moving all the internals into the new case. Should be a couple days before it's up again and I can test it with the new PSU.
 
You must log in or register to reply here.
Back
Top Bottom