Inno3d RTX 3090 Problem

[Andy112]

Yesterday I got the aforementioned Inno3d RTX 3090 from OCUK. Happy day.
Worked just fine under low load. Videos. Desktop. Browsing. Non intensive graphic games.

Cue cyberpunk 2077. Went up to 83 Degrees once at the start . Then it went back down. Fans working to cool stuff down.
Pink dots on the screen 1 sec and then No signal. Computer was still working. Had to manually turn it off and back on.
When trying to play. Same kind of crash except this time the computer rebooted by itself.
3rd time it happened I was unable to turn it back on. The computer turned on but the graphics card would show no signal (even though it looked as if it had power).
Turned it off. Left it for 5 minutes worked just fine under low load afterwards.
Edit : I could play for a while , the crashes just happen randomly. Also all the gaming ,browsing and testing was done on 2k resolution with max other settings.

Decided to get a benchmark test (Furmark) to see if that triggers it --- it did it within 5 seconds - pink dots into no signal.
At this time I was monitoring the temperature and it didn't even have time to reach 70 degrees.

Any ideas ?

My specs are i7-10700k , 2x WD Blue SN550 NVMe SSDs (500gb and 1tb) , 1x wd 1tb HDD , 32gb ram and a 850W Rog Thor PSU , 3 Case fans , the 120mm water cooler fan for the CPU. All this in a Fractal design Vector RS blackout case. MB is ASUS Z490-F


Mentioned so far: No damage. Has been reseated. Bios (not updated but changelog on asus website does not suggest anything to do with this problem). Airflow is at least reasonable. Different ports not checked (no DP cable). On different power lines (but same rail) on PSU.

 

[Andy112]

I have already reseated it. Did not make a difference.
The 2x 6+2 connectors are from different cables from the PSU however the PSU is single rail. Randos from discord seem to suggest the OCP is triggering this but I tend to disagree since as Grim5 mentioned the visual artefacts tend to suggest a memory issue.

Diagnosing past problems that I can fix is pointless however. This is more of me just making sure I covered everything I could before I contact OCUK.

Thanks for all the suggestions and I am still open to new ones until OCUK opens and I can call them

 

[Andy112]

Two front 140mm fans intake . 1 in the back for exhaust also CPU water cooling exhausting on topside. Mesh filters have all been thoroughly cleaned in preparation of installing the card. To be fair it barely fits even though this is a spacious case.

Edit . Front and back fans are all stock though.
2nd edit : checking bios now . Hold for updates
3rd edit : Could i have the link to what you found regarding bios ? I am quite proud of my google fu and i found nothing of the sort.

 

[Andy112]

No curve. I did not need one as I had a 1060 3gb that did not require such finesse . However I manually set all the fans to 100% In preparation of the test. (the one where it crashed in 5 sec)

 

[Andy112]

And that was in the main menu of all things. I wouldn't say seconds but less than a minute anyway and with fans running flat out.

Also the fact that it doesn't work immediately but after a few minutes of computer being turned off it does is telling me it's a heating issue as well.

Also also. Why would the stress test crash it in 5 seconds?

My money is on a faulty ram module.

 

[Andy112]

No drops from me. This thing cost an arm and a leg.
Inspected it when I first opened it. Looked mint.
Inspected it again when I reseated it. Still looked mint.
The only thing is I haven't peeled off the plastic film from the middle of the fans as it is REALLY stuck on there and I didn't want to risk damaging the card.

 

[Andy112]

I do not have any display port cables . And it only has 1x HDMI. Tried a different cable already.
Also the fact that some of the time it rebooted the whole computer other times it didn't (discord still worked for a few seconds before freezing all whilst having no video feed) points to not being the connector's fault since that wouldn't freeze the pc (?).

 

[Andy112]

Does that imply calling them and asking for it or going on to inno3d's website and getting one ?

 

[Andy112]

Great. Updating top post to reflect conversation and waiting for 9AM to give them a ring.

Thanks for all the help Guys! - If nothing else at least I didn't forget to try something basic .

 

[Andy112]

Final update , decided to give it one more go .

Reinstalled the 3090. Double and triple checked everything. Ran Furmark on 2k. It managed to run the test 2 times (while slowly heating up). On the 3rd attempt upon reaching 71 Degrees Celsius it did the same - Pink Dots into no signal on monitor.
I Will give OCUK a call in 20 mins and see what they can do to help.

 

[Andy112]

I just got off the phone with them and that's exactly what they said. They'll test it and if the problem i reported checks out i'll get it replaced.
Made the "mistake" of being truthful and saying 3 tests . If what you're saying is true you're not giving me much hope .... goodbye 2000 quid ?

I hope they don't run their tests in the freezer then.

Edit: This card pops really fast under load tho.... I sent a webnote asking them to ammend the recorded fault saying it has to warm up first. Hopefully things turn out okay.

 

[Andy112]

Final result : RMA was confirmed . Faulty card.