Precision 7960 Tower, Artifact, glitches, flickering issue.

Hello,

among the attempts, changing bios settings was also tried (and I assume you are also in contact with nvidia and adobe) ?
Since in one case the sleep mode also acted bad, maybe disabling S3 ( and / or S2 ) state could be a valid test.

A few questions : 

- are the computers shut down after work , or hibernated / put to sleep ?
- are they put to sleep during the day, normally ?
- are the issues showing up before , after , or both before and after sleeping ?

- are the computers ever rebooted ? if yes are they displaying issues from the reboot to the shutdown (assume NO sleep or hibernation between the reboot and the shutdown )

- Nvidia control panel > Workstation > manage gpu utilization : Enable error correction code

Hi @mazzinia_ 

Please see response.

- are the computers shut down after work , or hibernated / put to sleep ? SHUTDOWN 
- are they put to sleep during the day, normally ? NO, it just happened randomly
- are the issues showing up before , after , or both before and after sleeping ? AFTER RANDOM SLEEP, THE UNIT NEVER WAKE UP not after 2-3hours RANDOMLY

- are the computers ever rebooted ? if yes are they displaying issues from the reboot to the shutdown (assume NO sleep or hibernation between the reboot and the shutdown ) NO

among the attempts, changing bios settings was also tried (and I assume you are also in contact with nvidia and adobe) ? NO WE HAVEN'T CONTACTED ADOBE OR NVIDIA SINCE THE UNIT IS UNDER WARRANTY BY DELL

Given the specification of the unit, the client assumes that everything runs smoothly, now it can't even handle the web browser due to the issue encountered.

I'll ask the client to reformat the unit if it is possible.

@Eajc​  If the computers are never rebooted : reboot one, and see if things stay the same ( reboot = clean slate . Shutdown = any lingering os issue in windows will be preserved )

Point 2, check the ecc setting of the gpu

Point 3, seriously test on one unit  disabling S3 and S2 states in the bios ( disable sleep and hibernation states at bios level )

(and check if Resizable bar is turned on in bios)

Add on : you should also speak with nvidia because I'm reading that some of those rtx 6000 required a newer vbios and that was obtained by contacting the nvidia support ( and rma'ing the cards )

Would also recommend to contact Dell Pro Support. You mentioned being under warranty ... 

I'm wondering if it's electrical glitches.  Are they plugged into a UPS?  If so, plug one straight into the outlet and check results.

Are they all on their own 20-amp circuit?

I'm also with mazzinia on contacting Dell Pro Support.

@bradthetechnut​ 

Hi, they were all connected directly to power outlet, not using any UPS or any protection appliance.

Also, I'd like to clarify the customer's current connection from GPU to monitor if they are using any adapter like DP port to HDMI adapter.

I'll get back to you once clarified.

Thanks.

Hi @all

We have an update:

We have replaced the motherboard of the unit, yet the same problem exist.

Before replacing the motherboard, we have replaced a GPU with the same problem.

We also re-install the Operating System and update everything, yet the same problem encountered.

The artifact, glitching and lines on the monitor still the same despite the replacement of the parts.

We have supplied a display port to display port cable end to end, GPU to monitor still the issue manifesting.

S3 ( and / or S2 ) (sleep) is also disabled, yet this features happens randomly and frequently resulting the unit to considered as decommissioned and unreliable to use.

The unit will be pulled out to customer.

Hoping Dell would pinpoint the root cause of the issue.

As of the moment, no clear solution is provided, the customer refuse to have another replacement parts. 

Sad to hear. I guess that trying a different monitor didn't help too.
I know this is probably weird as a question, but has it been tried to operate the unit in a totally different place ? ( any room where people are operating computers without suffering issues )... just saying to eliminate another factor.

And 2ndly, maybe somehow there's something that causes a small short inside the chassis... you changed the mb, if the old one is still in your possession or you want to use a little time to do a test, maybe keep the mb out, add the gpu like if being on a bench, extend the power cable outside and see what happens ?

Hi @mazzinia_ 

Since this unit is a problematic one, yeah we did bring it to other room and sadly it exhibit the same issue.

There are about atleast 20 I guess workstation in that room that is used  to edit videos since this customer is a big time local tv station, (5 of them are this model, and other 15 are other brand) with the same monitor and just works fine except this Dell brand and model.

We are being pressured every time we support this customer so no, we didn't try take out everything out of the chasis to test whether the issue exists or not out of the chasis.

I said there where 5 of this Dell model in that room that exhibits same exact issue, however one of them is very problematic. There are times that after logging in the Windows, everything you see is like pixelated, unreadable texts and unrecognisable icons. This one is already pulled out today and to be handed over to Dell to further isolate the issue where we can't perform under customer premises due to pressure. And the other 4 of them though exhibiting some issues still usable.

Take note that there are 2 monitors connected to the unit, when you dragged the window to other monitor, what you see in the main monitor still visible to the secondary, so i don't think replacing monitor will help so as the display cable.

I'll keep you updated once we have new updates.

Thanks.

Well, it's definitely a migraine level of issue. Before giving it to dell, isn't that you have maybe an old quadro nvs or a p600 or anyway any other simple card to swap in place of the current gpu ?

Just to see if something changes

Sad to say that we don't have any GPU laying around for testing purposes. To be honest I am very much challenged to settle this issue, however due to a tension rising up we have no choice. This unit is not a single dollar value workstation, imagine how much it costs just a single stick of memory and with a total of 1TB of it added with a very high end GPU. We put ourselves in customer's shoe and we get this kind of performance and imagine there were 5 of it, it is very frustrating on customer's side, resulting their production on delay. We just nod out heads whenever the customer bad-mouthing on us hahaha.

If only we have spare gpu we'll be gladly test it to see if anything changes, like I said we were unlucky to have one.

Well let's see what happens this coming following days or weeks.

@Eajc​ oh yes, i can fully understand. A base configuration is already having a cost, add the extras and the fact that now ram is also affected by shortage... it's a major investment, and indeed multiplied by 5...

I would have another test idea, but it would be more for fun since it would add more annoyance "if" working ( idea being lower the ram to 32 gb or the minimum allowed by the dimms installed ).