PC restarts after GPU swap

[Guest]

10 minutes ago, IIIIIIIIII said:

Yes.

http://jongerow.com/no-sense/

OMG!  I forgot I even wrote that!  LOL!

 

[IIIIIIIIII]

11 minutes ago, jonnyGURU said:

OMG!  I forgot I even wrote that!  LOL!

 

Can't you add that link to jongerow.com?

It can only be found by Google search

 

[Guest]

1 hour ago, IIIIIIIIII said:

Can't you add that link to jongerow.com?

It can only be found by Google search

 

Maybe I wasn't done with it?  LOL!  I have no idea why that page exists.  Looks good to me, though! 

 

I think I was just riffing after someone banned me from a sub-Reddit for telling someone to leave off the sense wires. 

 

 

[Spotty]

15 minutes ago, jonnyGURU said:

Maybe I wasn't done with it?  LOL!  I have no idea why that page exists.  Looks good to me, though! 

 

I think I was just riffing after someone banned me from a sub-Reddit for telling someone to leave off the sense wires. 

 

 

[Guest]

6 minutes ago, Spotty said:

 

 

That might not have been it.  I'm friends with the CableMod guys.  But I can also see how end users can see "missing pins" as "wrong cables" and freak out. Sense wires are good, no doubt.  But they're usually not necessary.  But you need a freaking novel to explain that to people and the average intelligence of the average PC builder has lowered considerably over the last 10 years, so there's that problem as well....

 

[Sleepycat3]

On 9/9/2021 at 7:38 PM, taavi said:

I removed sense pin and for now I have not experienced any random restarts.

 

Can there be any negative side effects when removing the pin, or is it just a control for "smooth" voltage delivery?

It's for detecting a voltage drop under load (or other reasons), and the PSU will increase voltage to compensate. Think of it like Vdroop & LLC for Ryzen CPUs.

[Sleepycat3]

5 hours ago, Spotty said:

 

 

 

Nice write up Jonnyguru. Basically without the sense wires, the PSU operates like a PSU of the 2010's. The voltage delivery from your testing read very similar to my Corsair HX850 from 2008. And back then, the HX850 was highly rated by reputable sites, I recall HardOCP rated it highly. 

[taavi]

On 9/10/2021 at 7:28 AM, IIIIIIIIII said:

Yes.

http://jongerow.com/no-sense/

I actually had read that before  Was the only thorough writeup in the internet of what sense wires actually are and do.

[Guest]

I guess I should add it to my main page navigation.  I really didn't think anyone ever read my shit.  I usually just put it up so I can point people to it when they ask the same questions over and over again.

 

[SAVRAYO]

On 9/4/2021 at 4:02 PM, jonnyGURU said:

Yeah. Though really it's more the card to blame than anything.  The power management on Ampere is horrible.

 

And it's not transient spikes tripping OPP.  People keep saying that, but I don't know where that rumor started.  I mean, it's not a bad guess, 

 

The problem that was observed is that under high loads, the ampere cards actually feeds noise back to the PSU.  The PSU sees this noise as a fault and shuts down.  Depending on the design of the PSU, this noise can be filtered out.  I've even seen some OEMs do something as simple as putting a ferrite bead on the +12V sense lead.  For most Seasonic users, I just tell them to remove (cut, de-populate, whatever) the +12V sense wire from the 24-pin cable and everything rights itself.

Hi .
I also have a shutdown problem.
For 9 months I did not have any problem, but a month ago the pc started to turn off in gaming

 

My PC is:
i9 10900kf
asus maximus xii apex
Evga 3090FTW3
Seasonic Prime Ultra Titanium 1000w. Purchased in 2019

 

I want to solve the problem using a ferrite bead. I want to put the ferrite bead in the 10 pin connector that connects to the psu. I'm a bit lost and I can't find the  + 12v sense wire to put the ferrite bead.
Is the wire number 4? (see photo in yellow).

 

 

[IIIIIIIIII]

2 hours ago, SAVRAYO said:

Hi .
I also have a shutdown problem.
For 9 months I did not have any problem, but a month ago the pc started to turn off in gaming

 

My PC is:
i9 10900kf
asus maximus xii apex
Evga 3090FTW3
Seasonic Prime Ultra Titanium 1000w. Purchased in 2019

 

I want to solve the problem using a ferrite bead. I want to put the ferrite bead in the 10 pin connector that connects to the psu. I'm a bit lost and I can't find the  + 12v sense wire to put the ferrite bead.
Is the wire number 4? (see photo in yellow).

 

 

 

[Sleepycat3]

On 9/14/2021 at 8:40 AM, SAVRAYO said:

Hi .
I also have a shutdown problem.
For 9 months I did not have any problem, but a month ago the pc started to turn off in gaming

 

My PC is:
i9 10900kf
asus maximus xii apex
Evga 3090FTW3
Seasonic Prime Ultra Titanium 1000w. Purchased in 2019

 

I want to solve the problem using a ferrite bead. I want to put the ferrite bead in the 10 pin connector that connects to the psu. I'm a bit lost and I can't find the  + 12v sense wire to put the ferrite bead.
Is the wire number 4? (see photo in yellow).

 

Yes, wire #4  in your photo of the 10 pin connector.

[SAVRAYO]

On 9/13/2021 at 9:52 PM, IIIIIIIIII said:

 

Thank you !

15 hours ago, Sleepycat3 said:

Yes, wire #4  in your photo of the 10 pin connector.

Thanks for the information!.

Did you only put wire number 4 inside the ferrite choke? (+ 12v sense wire).

I also have a 5mm ferrite choke. What material did you use to fill the gap between the wire and the ferrite choke?

 

On the other end of the cable (24-pin cable that connects to the motherboard), is a ferrite choke also placed?

[Sleepycat3]

12 hours ago, SAVRAYO said:

Thank you !

Thanks for the information!.

Did you only put wire number 4 inside the ferrite choke? (+ 12v sense wire).

I also have a 5mm ferrite choke. What material did you use to fill the gap between the wire and the ferrite choke?

 

On the other end of the cable (24-pin cable that connects to the motherboard), is a ferrite choke also placed?

Yup, only wire #4. I used some heat-resistant foam to push the wire against the inside of the choke.

 

Only use 1 choke, on the PSU side connector on wire #4. I did not put anything else on.

[taavi]

Seems like the sense wire didn't help after all in my case. Random shutdowns are back. I noticed that it happens more often when computer has been running for at least a couple of days and when starting Overwatch (even crashes in the menu).

 

Will try Windows reinstall also, but it seems to be more of a hardware than a software issue.

[Guest]

2 hours ago, taavi said:

Seems like the sense wire didn't help after all in my case. Random shutdowns are back. I noticed that it happens more often when computer has been running for at least a couple of days and when starting Overwatch (even crashes in the menu).

 

Will try Windows reinstall also, but it seems to be more of a hardware than a software issue.

Thermals, then?

 

[taavi]

11 hours ago, jonnyGURU said:

Thermals, then?

 

Just ran Furmark for 10 minutes. GPU temps stabled out on 68C. I'm fairly confident it is not related to thermals, because running benchmarks like Furmark, Timespy Extreme, Superposition are all stable. Even when increasing GPU power limits to test higher power draw, all benchmarks run fine and without any reboots.

 

The only way to kind of (with varying chances of success) reproduce the issue is when my PC has been on (after restart) for more than a day and when I open Overwatch it instantly (sometimes even not reaching the menus) reboots without BSOD (and thus any dump files). Strange thing is, when I try to play or test it after the restart - everything runs fine and the game is not even maxing out my GPU or CPU. It is so weird. I have few times noticed the same behavior with MS Flight Simulator 2020 but with less "success".

 

Quote

 

[IIIIIIIIII]

10 minutes ago, taavi said:

Just ran Furmark for 10 minutes. GPU temps stabled out on 68C.

Try adding Prime95 to it. Furmark(1080p, 2x MSAA)&Prime95(All cores - small FFT)

[taavi]

3 minutes ago, IIIIIIIIII said:

Try adding Prime95 to it. Furmark(1080p, 2x MSAA)&Prime95(All cores - small FFT)

CPU thermal throttled heavily (expected because of the unrealistic load from P95) but the system runs fine. No restarts.

 

Quote

 

[Sleepycat3]

On 9/18/2021 at 2:45 AM, taavi said:

Seems like the sense wire didn't help after all in my case. Random shutdowns are back. I noticed that it happens more often when computer has been running for at least a couple of days and when starting Overwatch (even crashes in the menu).

 

Will try Windows reinstall also, but it seems to be more of a hardware than a software issue.

If it crashes in the Overwatch menu, I'd do the following steps to diagnose:

1) Turn on Vsync, confirm that fps in menu matches vsync. Then note any crashes in menu.

2) If it still crashes, turn off vsync, then run a CPU single core stability test. I like Corecycler, which cycles each core through Prime95 Small FFT AVX.

3) If it finds an unstable core, then you need to increase the CPU voltage by a tiny amount and retest just that core.

4) Test with Overwatch again, with vsync off.

Edited September 18, 2021 by Sleepycat3

[taavi]

On 9/19/2021 at 2:55 AM, Sleepycat3 said:

If it crashes in the Overwatch menu, I'd do the following steps to diagnose:

1) Turn on Vsync, confirm that fps in menu matches vsync. Then note any crashes in menu.

2) If it still crashes, turn off vsync, then run a CPU single core stability test. I like Corecycler, which cycles each core through Prime95 Small FFT AVX.

3) If it finds an unstable core, then you need to increase the CPU voltage by a tiny amount and retest just that core.

4) Test with Overwatch again, with vsync off.

I will try that. But remember, that these random reboots started after I had switched my GPU to 3090 so naturally the CPU most probably will not be the culprit here.

 

EDIT: 
A little update if anyone is interested. I have now ran computer for 48 hours straight without it crashing (even with OC'd GPU) when running Furmark or launching Overwatch. The only thing I changed is I reseated all PSU side cables to different PCIe output ports and reseated GPU side cables. And placed external PC power cable to different wall socket. Unfortunately I did it all at the same time (to save time ) so I cannot say which of the reseating's fixed it. I still have the sense wire "fix".

 

One thing I noticed after this is when I thought the crash should happen I saw GPU board power draw sudden increase only for a second. Default max TDP was 380W and it took about ~420W with this spike, so almost 10% over the max set power draw.

Edited September 23, 2021 by taavi
Update

[Sleepycat3]

An update from me with the ferrite choke "fix". I've been trying more difficult tensor core loads, mostly around AI based processing software, and I can again trigger a PSU shutdown, eventhough my GPU would be drawing only 200W. The total draw from the wall was about 400W. 

 

Interestingly, while I am running a heavy load, but low power draw on the tensor core, all I need to do is to change the power limit in MSI Afterburner down by 10%, and it will shut down as soon as I click the apply button. So it is not because of too much current through the supply, but something is still generating noise that is not being reduced by the choke, or is too large a spike, due to it also happening when I reduce GPU power draw through software.

 

Once I did this enough times, the PSU would not restart for me and I could hear a clicking sound coming from it when I try to turn the PC back on.

 

To force the PSU to start again, I had to hold the power button down and the PSU would keep clicking over and over, before starting after about 3 seconds of clicking (just like a car engine!). 

 

Looks like I will RMA this Seasonic PSU afterall. 

[BuzzingBee]

I also got random freeze then restarts only on GPU loaded games (HZD, Crysis remastered). On the other hand I never got random restarts from max CPU load. Took a lot of digging to realise it's because of seasonic PSU.

 

My card is Sapphire Nitro+ RX 6800XT and PSU is Seasonic Prime Platinum 1000W (SSR-1000PD). Admittedly my gfx card is OCed nearly at its maximum, and I know reducing OC or power limit may solve the issue, I wouldn't like to compromise my OC as it shouldn't be happening in the first place.

 

Because I bought my PSU second hand, returning isn't an option so I decided to follow jonnyguru's solution by removing the 12V sense wire out from PSU side and attach onto seperate connector so that the wire is safe. Now I'm benchmark-looping games and so far no restarts. If it doesn't restart after a week I would consider problem solved. 

[Sleepycat3]

My RMA'd Seasonic PSU is on its way back to me after being at the service location for 3 weeks. It should arrive early next week.

 

During the past 3 weeks, I have been running my system on my 13 year old Corsair HX850W, running AI workloads on my 3080, combined with Handbrake encoding on the CPU at the same time. And.... it has not skipped a beat at all! It's happy chugging away at 650W loads when I am pushing both CPU (145W) and GPU (271W) at the same time. It doesn't seem fussed about the RTX's noisy feedback.

 

Here's to hoping that I get a newer version of the Seasonic PSU without this issue. Otherwise, I'll probably sell it off and keep using my old Corsair.

[Sleepycat3]

Just received my replacement PSU. A new PX-850 to replace my old SSR-850PD. Manufacturing date is 2021 March, so hoping that it resolves the 12V sense issue.