Separate names with a comma.
Discussion in 'nVidia Flavor' started by Master_shake_, Oct 29, 2018.
Upper 80's low 90's.
Woah. If the temp sensors are accurate that's some hot RAM.
Maybe it's more like cooling failure? Non contact or heat sinks are too thermally loaded?
A few guys removed the heatsink and checked the thermal pads and said the thermal pads were forsure squished and making contact.. so no idea.
Hot ram would make sense...
Has anyone with a card underwater had issues? I don’t think so?
Wonder if it is just a case of bad batch GDDR6 since the failure seems to be focus on cards with serial number starting 0323.
The board design has not changed from 1080 ti FE at all in regards to where the power phase sits relative to the ram/ gpu. I guess they got away with it on the 10 series because of lower TDP.
It looks like 20 series FE cards are all going to fail. Does any one know if the EVGA cards are FE nvidia boards or custom like MSI ?
I also wonder how long will it take for the bios updated 366 TDP gigabyte boards to fail. Bravo Nvidia. What a shit show...
I think they’ll be fine if you don’t raise the TDP limit. At standard TDP limit the card doesn’t get that hot. It’s when you start raising it - it easily uses everything you’ll give it.
I've been running fire strike and time spy for the past few days in an attempt to break my card. Also been loading up FFXV and just fighting large mob groups. So far, I haven't seen any problems, but I'll keep pushing.
So far, none here. My cards max out under load around 50C, haven't checked memory temps specifically though.
The card I returned was the EVGA 2080ti variant. Thing ran super hot and I began seeing early signs of death....IE stuttering, flickering, etc so I just did a standard no questions asked return. I'm waiting still for a week now for the actual refund to appear....
I'm on week 1 of my new FE variant direct from Nvidia and so far so good...I have an odd serial on it however that I've only seen a few other times being reported in forums so perhaps a new bin or something?
My exchange with nVidia customer support:
"Santosh: Hi, my name is Santosh. How may I help you?
NukeDukem: Nuke 12:47 PM (4 minutes ago) to info Hello, I pre-ordered a ZOTAC RTX 2080 Ti. After one month of use, the card died on me and I have to RMA with ZOTAC, one of your AIB partners. They have refused to compensate me for shipping to the RMA center. I have heard nVidia has acknowledged there are problems with the initial batches of RTX 2080 Ti cards. As a $1199.99 product, I am upset that ZOTAC is not offering me an Advanced RMA as nVidia.com direct customers are reporting being offered. My question is, would you be willing to compensate me for the price of shipping my 2080 Ti back to Zotac? I am a member of several tech forums, and would love to report to them that nVidia is taking responsibility for these very expensive faulty cards, even to those customers who purchased through your board partners. I do not feel I should be obligated to pay RMA shipping for what could be MULTIPLE RMA situations with ZOTAC. I have been a supporter and customer of nVidia for years, and would appreciate very much you consideration on this matter. Thank You.
Santosh: I am sorry for the inconvenience you have been through please note NVIDIA has many add in partners like PNY,EVGA,ZOTAC,MSI,DELL,HP we sell chip set to our partners and the card is manufactured by them.
Santosh: Unfortunately there is nothing much that we can assist you in this regard.
NukeDukem: Thanks for your time, I'll report your response on all tech forums I belong to."
It was worth a shot I suppose. I could have pressed "well the chip set itself seems to be the problem, you jerks", but I can't prove it. I just hope they get lots more negative flak for selling defective products and then blaming their AIB partners. The integrity of this company is on par with that of Enron as far as I'm now concerned.
Always buy evga if one can. Best customer support and Rma process is king.
NukeDukem That is expected. The AIB companies handle support and RMA for their products. Sad that ZOTAC is not coming correct, but that is on them and not Nvidia.
Oh I didn't expect them to help me. But don't delude yourself, this whole thing IS on nVidia. They released the defective product, they should make it right. But they won't (although I'm sure the AIB's will be compensated for their losses by nVidia). Screw them AND Zotac. We who pre-ordered are fools. Never again.
Is the serial # on the card itself? Or on the box? I have a
2080Ti card but not using it yet till I build my system. I see the SKU and parts # on the box and it does have a SN:N184100007100 so not sure if that's it. Not seeing the 323/4 numbers. (Or am I wrong?)
The serial #'s are completely different on Zotac cards. Can't confirm what batch they came from. Mine is a brick and S/N is N18370000****.
Ah okay...Got it later than most and even if I have to return it eventually I still have the Titan card if needed. Thanks!
Seems Nvidia is denying there is a broad issue.
Makes me ambivalent about potentially upgrading, was seriously considering it until these issues became known.
I have not heard of any failures of EVGA's FTW3 cards, but a lot of people are noting higher temps on the upper mem bank. Some people are seeing 80+ degrees...some up to the high 80s. This is what I saw after 30 min's of 4k Haven. Do you think that 80c temp is a reason to be concerned?
I've had my two 2080 Ti FEs for about a month now, no hardware problems thus far. Had issues with a build of Shadow of the Tomb Raider that would BSOD my sig rig running Windows 10 1809. Other than that nothing else thus far. NVLink/SLI provides incredible performance at 4k in games that support it.
Haven't done any overclocking yet.
Probably better asked on the EVGA forums. 80C isn’t catastrophic for ICs but it’s really weird since it’s way higher than the die itself... the max is 95C.
Temps that high on memory and they wont last long as 90 degrees is critical so you will be losing lifespan quickly on those chips. On a personal level I am unhappy with anything in the 80's and you achieved that in 30 minutes, might want to see how hot it gets playing a game for a few hours where all equipment should be heat saturated at that point. Perhaps Nvidia pushed this GDDR6 a little too hard but the memory shouldn't be that hot.
Worked in the auto industry for years and people pay far more for a vehicle then a video card and yeah you had to get the car to a repair facility on your dime not the manufacturers unless you paid extra for that service.
I just thought " Hey, nVidia is paying Advanced RMA for their direct customers, maybe they would throw me a bone." I understand how it works with RMA, but this seems to be an unusual situation. The cards are defective at abnormally high rates. Why not ask the source of the problem to help me? The money isn't the issue, but I didn't gain a seven-figure net worth by throwing small amounts of money away. This is really the only hobby where I splurge on the latest and greatest. I just love PC gaming. What I hate is this industry and the crap this particular company pulls in the utter absence of competition.
How the hell did nVidia not even test these things for defects? This is what they call a premium product and charge $1200 for? "It just works" - my ass. Might as well take up some of their time, because they're taking a bunch of mine. And time, unlike money, is something one can never make back.
I think some of it is people running inadequate PSUs or PSUs in the wrong mode (multirail instead of single) not giving enough amps.
A lot of reports is that the cards die days or weeks after which is unrealistic to ask them to test that long.
Its not the PSU's. Those things simply cook themselves. Even the FTW3 are running way too hot. Mid to high 70's surpassing 80's. This release is a hot mess.
Honestly it's why I'm waiting to try and snag one from Amazon. 30 days zero question return with no fees. Plus have until Jan 31st to return.
I'm so glad I held off on this release. It will be interesting to see how this all plays out. Tech Jesus Steve @ GN said in his video last night that board partners have not seen anything out of the ordinary as far as RMAs but the cards have not been out for very long.
Welcome to Hell.
Is that your latest card?
I understand the frustration and there is no harm in asking, you just go the expected answer. Your right tho they should have tested the cards better.
Well, I have 2 2080 Tis I bought for NVLink. One is from the original batch (0323) and the other is the replacement that I already RMA'd (0324).
I guess it's possible the other original card is bad too, I don't know. Or maybe there is some other problem like a bad driver.
man that's a USB Driver issue or it seems to be
check the minidump and see if this appear DRIVER_IRQL_NOT_LESS_OR_EQUAL it seems to be happening with some USB 3.0 driver.
if it's something like KMODE_EXCEPTION_NOT_HANDLED it's a GPU driver related probably..
and PAGE_FAULT_IN_NONPAGED_AREA generally related to the commit usage when windows doesn't manage correctly the pagefile.. set a fixed pagefile, I was having issues with all the drivers newer than 411 with this error while gaming, it turned to be the pagefile I set a minimum of 8192 and max of 16384 no more issues so far.
Lol they put qr codes on bsods in windows 9
Araxie It says SYSTEM_SERVICE_EXCEPTION. It's just strange that it seem to happen with Far Cry 5 specifically, and I never saw this before the 2080 Ti card was installed in the system.
Maybe nVidia surreptitiously supporting competition after all? Anything runs faster than a non-operational card...
Yeah, that's a USB BSOD. It could be some issues folks are having are due to chipset/driver issues with the USB-C controller on the 2080's.
Good thinking, you might be on to something here ....
I should note, this was with power limit to 120%, which I was doing to test the system. Put it back at 80% and was able to play around 1 hour without a crash.
I specifically have not been installing the USB C drivers since I got my cards and haven't had an issue yet. Obviously this is just anecdotal but when I saw those drivers with the install package and I knew I wouldn't use them, I simply omitted them.