My terrible experience with the Gigabyte Support

[Myself5]

Hello everyone,

I've been a long time lurker but unfortunately came across an experience that I had to share.

 

Let's start with a short timeline:

• 10th of February 2021: In the middle of Crypto related GPU-pricehikes i am in need of a new GPU. I finally manage to grab an Gigabyte RTX 3080 Aorus Master (Rev 1.0) for 1335€ (~$1459 US)
• End 2022 - Early 2023: Card keeps on "crashing" in games. All screens turn black, the fans go to 100%. System keeps running and can be interacted with, theres just no output.
• I suspected the RAM/CPU Overclock. The issue appeared not too often so it was hard to test. Finally discovered its the GPU when testing with OCCT.
• 9th of February 2023 (yes, really): I contact my retailer (Alternate Germany) and request an RMA.
• 14.2: Card arrives at Alternate
• 18.2: Card is processed (We later learn Alternate was able to reproduce the issue)
• 27.2: Card is sent off to Gigabyte
• 22.3: Alternate is getting the card back from Gigabyte.
• 25.3: The card arrives at my place. This is where the story begins.

 

Foreshadowed by the Alternate support that the issue has not been resolved as "Gigabyte could not reproduce the issue", I unpacked my GPU and noticed that the card had been damaged in a way that it could not have happened in shipping from Alternate to me. It was later confirmed by Alternate that this must've happened between Gigabyte and Alternate that that they had sent it to me bent like that.

 

Spoiler

 

Upon contacting Alternate I was recommended to "just bend it back". Assuming this might have happend through some kind of drop or similiar while keeping in mind that the card was still not fixed I was worried about the warranty and refused to just "bend it back". After some horribly slow Email exchanges I eventually sent the card back in to Alternate.

 

• 4.4: The card arrived at Alternate
• 12.4: The card has been unpacked
• 17.4: The bent slot cover had been replaced, but all of a sudden Alternate is no longer able to reproduce my original issue.
• 24.4: Alternate agrees to send the card back in to Gigabyte at my own expenses because they claim the card they confirmed to be broken a month earlier is, without any interaction, no longer broken.

Being more than just mildly upset with the time Alternate takes to reply to my questions I had contacted the Gigabyte support in the meantime. There I was told that I should please "provide detailed steps on how to reproduce the issue so that it can be checked when I send the card back in". I did that by explaining how 5 Minutes of running OCCTs 3D Test test will easily cause the card to fail.

 

I also proceeded to explain how this only appears at the cards very performance peak and that a workaround would be to use the silent BIOS or use it in PCIe 3.0 Mode. Both of which are no solution to me as I want to use the card at its designed performance for obvious reasons. I also explained that, according to my online reasearch, this would mostly happen with AMD Systems and that they should test the card in an AMD system, if possible.

 

Gigabyte replied, telling me that they will not be using OCCT as they can no confirm that OCCT is a "3rd party utility"  that they will not be using "because [they] cannot guarantee it reads [their] design and [that it would provide] the correct test report." This left me with more questions than before. We're not talking about OCCT reporting any errors. The issue is 3 Black screens and a 100% fan spin. This also can be reproduced in games (like COD:MW2) just not as easily as OCCT. Either way, Gigabyte agreed they would try their best to reproduce the issue.

 

Which leads to today:
I just got this Answer from Gigabyte: 
 

Quote

Dear Myself5,

As checked with the service center, they have tested the card first with NVIDIA standard tool and run again with a burn-in test (for 84hr 38min) as the report attached. Neither test can find the issue on the product. There are many possibilities that could be the root cause of the issue, please test the graphics card on another PC / clean install OS with our driver from the product page to isolate the issue. If still you have concern, please contact your supplier help to arrange further service. The below image is attached for your reference.

They replied this to a ticket in which I mentioned that the issue was reproduced on a different system AND on my system with a clean OS install. But even worse: Below is the picture they mentioned.

 

Spoiler

The longer you look at this picture the more painful it gets.

 

Did they try OCCT? No. Fine, I had been given a somewhat reasonable explanation for that.

Why is the 3DMark Score that low? A Time Spy run on my system with that very card get's me a Score of 17631 (18328 GPU + 14506 CPU).

Did they use an AMD CPU? No. Ok, there were also reports that it happens on Intel.

 

But this is where they absolutely lost it. The Intel CPU used was an i5 8500T.

They tested my RTX 3080, which I carefully described to have issues only at it's peak performance, on an Power-Optimized midrange CPU from 2018. Running at 2111MHz. Using PCIe 3.0.

Even their Screen shows the card has not reached 100% load for once.

How is this supposed to be a stability test? How can you, with confidence, look at this screen and say that there are definitely no crashes for that CPU when under full load?

So to sum it up:

Gigabyte is telling me that there's "nothing wrong with the card".

I'm down 40€ because Alternate already refused to send it in the second time.

I have been without my GPU for over 3 months now and chances are very high that the issue still isn't fixed (after all, nobody has ever "repaired" any damage other than the slot cover).

The only way to send it back in a third time is through Alternate as Gigabyte does not offer direct customer support in Germany, and nobody is guaranteeing me that this time Gigabyte would finally reproduce the issue, or at least use a testbench that MIGHT be able to reproduce it. Not to mention the time it will take yet again.

 

I need this PC for university, currently run on a 6700 XT that was supposed to be somewhere else quite a while ago and it doesn't look like I will be getting back a fixed 3080 anytime soon.

 

Has anyone ever had similar issues with Gigabyte?

Does anyone have any other ideas I could try to resolve this?

[PDifolco]

7 minutes ago, Myself5 said:

Hello everyone,

I've been a long time lurker but unfortunately came across an experience that I had to share.

 

Let's start with a short timeline:

• 10th of February 2021: In the middle of Crypto related GPU-pricehikes i am in need of a new GPU. I finally manage to grab an Gigabyte RTX 3080 Aorus Master (Rev 1.0) for 1335€ (~$1459 US)
• End 2022 - Early 2023: Card keeps on "crashing" in games. All screens turn black, the fans go to 100%. System keeps running and can be interacted with, theres just no output.
• I suspected the RAM/CPU Overclock. The issue appeared not too often so it was hard to test. Finally discovered its the GPU when testing with OCCT.
• 9th of February 2023 (yes, really): I contact my retailer (Alternate Germany) and request an RMA.
• 14.2: Card arrives at Alternate
• 18.2: Card is processed (We later learn Alternate was able to reproduce the issue)
• 27.2: Card is sent off to Gigabyte
• 22.3: Alternate is getting the card back from Gigabyte.
• 25.3: The card arrives at my place. This is where the story begins.

 

Foreshadowed by the Alternate support that the issue has not been resolved as "Gigabyte could not reproduce the issue", I unpacked my GPU and noticed that the card had been damaged in a way that it could not have happened in shipping from Alternate to me. It was later confirmed by Alternate that this must've happened between Gigabyte and Alternate that that they had sent it to me bent like that.

 

  Reveal hidden contents

 

Upon contacting Alternate I was recommended to "just bend it back". Assuming this might have happend through some kind of drop or similiar while keeping in mind that the card was still not fixed I was worried about the warranty and refused to just "bend it back". After some horribly slow Email exchanges I eventually sent the card back in to Alternate.

 

• 4.4: The card arrived at Alternate
• 12.4: The card has been unpacked
• 17.4: The bent slot cover had been replaced, but all of a sudden Alternate is no longer able to reproduce my original issue.
• 24.4: Alternate agrees to send the card back in to Gigabyte at my own expenses because they claim the card they confirmed to be broken a month earlier is, without any interaction, no longer broken.

Being more than just mindly upset with the time Alternate takes to reply to my questions I had contacted the Gigabyte support in the meantime. There I was told that I should please "provide detailed steps on how to reproduce the issue so that it can be checked when I send the card back in". I did that by explaining how 5 Minutes of running OCCTs 3D Test test will easily cause the card to fail.

 

I also proceeded to explain how this only appears at the cards very performance peak and that a workaround would be to use the silent BIOS or use it in PCIe 3.0 Mode. Both of which are no solution to me as I want to use the card at its designed performance for obvious reasons. I also explained that, according to my online reasearch, this would mostly happen with AMD Systems and that they should test the card in an AMD system, if possible.

 

Gigabyte replied, telling me that they will not be using OCCT as they can no confirm that OCCT is a "3rd party utility"  that they will not be using "because [they] cannot guarantee it reads [their] design and [that it would provide] the correct test report." This left me with more questions than before. We're not talking about OCCT reporting any errors. The issue is 3 Black screens and a 100% fan spin. This also can be reproduced in games (like COD:MW2) just not as easily as OCCT. Either way, Gigabyte agreed they would try their best to reproduce the issue.

 

Which leads to today:
I just got this Answer from Gigabyte: 
 

They replied this to a ticket in which I mentioned that the issue was reproduced on a different system AND on my system with a clean OS install. But even worse: Below is the picture they mentioned.

 

  Reveal hidden contents

The longer you look at this picture the more painful it gets.

 

Did they try OCCT? No. Fine, I had been given a somewhat reasonable explanation for that.

Why is the 3DMark Score that low? A Time Spy run on my system with that very card get's me a Score of 17631 (18328 GPU + 14506 CPU).

Did they use an AMD CPU? No. Ok, there were also reports that it happens on Intel.

 

But this is where they absolutely lost it. The Intel CPU used was an i5 8500T.

They tested my RTX 3080, which I carefully described to have issues only at it's peak performance, on an Power-Optimized midrange CPU from 2018. Running at 2111MHz. Using PCIe 3.0.

Even their Screen shows the card has not reached 100% load for once.

How is this supposed to be a stability test? How can you, with confidence, look at this screen and say that there are definitely no crashes for that CPU when under full load?

So to sum it up:

Gigabyte is telling me that there's "nothing wrong with the card".

I'm down 40€ because Alternate already refused to send it in the second time.

I have been without my GPU for over 3 months now and chances are very high that the issue still isn't fixed (after all, nobody has ever "repaired" any damage other than the slot cover).

The only way to send it back in a third time is through Alternate as Gigabyte does not offer direct customer support in Germany, and nobody is guaranteeing me that this time Gigabyte would finally reproduce the issue, or at least use a testbench that MIGHT be able to reproduce it. Not to mention the time it will take yet again.

 

I need this PC for university, currently run on a 6700 XT that was supposed to be somewhere else quite a while ago and it doesn't look like I will be getting back a fixed 3080 anytime soon.

 

Has anyone ever had similar issues with Gigabyte?

Does anyone have any other ideas I could try to resolve this?

Gigabyte support may be sh#t but your card seems ok indeed

18K on 3DTimeSpy is expected score (average is 17K, my non OCable 3080 Eagle had 17.5K undervolted)

3080 isn't supposed to be stable at 2110MHz, normal boost is 1680 and 2000 is usually kinda max OC

 

[Myself5]

5 minutes ago, PDifolco said:

Gigabyte support may be sh#t but your card seems ok indeed

18K on 3DTimeSpy is expected score (average is 17K, my non OCable 3080 Eagle had 17.5K undervolted)

3080 isn't supposed to be stable at 2110MHz, normal boost is 1680 and 2000 is usually kinda max OC

 

The Time Spy score is from when the card was still fine, but even then it was't overclocked apart from running it in the stock "Overclocking" BIOS.

Havn't had anything recent and currently it's on it's way back to me. Unfixed, ofc. The thing is that even before I sent the card it was fine most of the time, just crashed during games (or OCCT) whenever it felt like it.

 

With the "low" score I was refering to the 7900 that Gigabyte got during their "stability test". Not sure if they ran all 3 tools at the same time but even then a CPU Score of 2900 seems laughable.

 

To further explain the issue: this is what happens. Both of those being Aorus 3080 Masters btw:

On the second one you can hear (through the controller rumble) that the system is still running fine, just the GPU output dies and the fans spin to 100%.

 

[Blqckqut]

11 minutes ago, Myself5 said:

The Time Spy score is from when the card was still fine, but even then it was't overclocked apart from running it in the stock "Overclocking" BIOS.

Havn't had anything recent and currently it's on it's way back to me. Unfixed, ofc. The thing is that even before I sent the card it was fine most of the time, just crashed during games (or OCCT) whenever it felt like it.

 

With the "low" score I was refering to the 7900 that Gigabyte got during their "stability test". Not sure if they ran all 3 tools at the same time but even then a CPU Score of 2900 seems laughable.

 

To further explain the issue: this is what happens. Both of those being Aorus 3080 Masters btw:

On the second one you can hear (through the controller rumble) that the system is still running fine, just the GPU output dies and the fans spin to 100%.

 

as much as i fidn the aorus series incredibly pretty cards, im never ever buying from gigabyte, read too many reports like this.

 

shame, i would have loved an lcd screen...

[Myself5]

On 5/11/2023 at 4:21 PM, Blqckqut said:

as much as i fidn the aorus series incredibly pretty cards, im never ever buying from gigabyte, read too many reports like this.

 

shame, i would have loved an lcd screen...

My hope at first that its only an issue with Alternate. But that PCIe 3.0 Stresstest absolutely killed it for me... Really wish I didn't buy an almost entire Gigabyte System...

[Blqckqut]

49 minutes ago, Myself5 said:

My hope at first that its only an issue with Alternate. But that PCIe 3.0 Stresstest absolutely killed it for me... Really wish I didn't buy an almost entire Gigabyte System...

i feel like this is illegal tbh

they are fully responsible, yet they are so lousy

i feel like you can deffinatly sue for false support claims or something like that

[DTReality]

I wish Linus would make a video about your post and put Gigabyte on blast. You should never treat your customer like they treated you. 

[Godlygamer23]

On 5/16/2023 at 5:31 PM, Myself5 said:

My hope at first that its only an issue with Alternate. But that PCIe 3.0 Stresstest absolutely killed it for me... Really wish I didn't buy an almost entire Gigabyte System...

Did you run 3DMark with the same settings they did? 

[Myself5]

On 5/17/2023 at 12:21 AM, Blqckqut said:

i feel like this is illegal tbh

they are fully responsible, yet they are so lousy

i feel like you can deffinatly sue for false support claims or something like that

Thought so too.
The problem I see here is that this is an issue with both, Alternate (the company that handles the warranty) and Gigabyte. They both fuck up on their own unique ways. Officially everything has to go through Alternate and there's no way way to directly speak to Gigabyte other than support tickets. Now if I was to sue Alternate they will claim they have nothing to do and it's Gigabyte. And I can't sue Gigabyte because the warranty has to go through Alternate.

 

Not to mention the cost which would by far exceed the price of a non-Gigaybte 4090 or 7900 XTX. It's just frustrating.

On 5/18/2023 at 6:35 PM, Godlygamer23 said:

Did you run 3DMark with the same settings they did? 

Other than that screenshot I have not been provided any Information about the test procedure.


Small Update: After calling them out for the PCIe 3.0 test and them claiming yet again that I should "test this on another system" which I repeatedly told them I did they forwarded me to their service center/partner RMA contacts.

They told me that they will replace the card with a refurbished 3080 1.0 or a new 3080 3.0 Yay. However, during that process they mentioned I need to send in the card again, which is when I learned that the card has been with Alternate since the 4th of May already. Just that Alternate once again didn't process it yet.

So for now I need to wait on Alternate (yet again) to process my card so that I can finally get it back, send it back in once more to Gigabyte and hope that they will finally honor the warranty and replace the card.

 

I was promised to get information on the current status with Alternate on Friday, which I once gain did not. Let's hope they have better news for me tomorrow. 

[Nick Brennan]

I had a gigabyte oc 4090. It was made junky as hell. Looked cheap. I returned it. I don't think I'll ever buy another gigabyte product again. Too many horror stories and their video cards are jank.

[DarkSmith2]

All excuses aside. they ran 84hours of Furmark, timespy and heaven even some of it at the same time.
At this point, if i would be alternate, i wouldnt send it in again aswell, BTW i highly doubt that alternate even tested your card, when they first got it. It is absolutely unusual to do this as reseller for first time send ins. There is probably nothing wrong with the card at all.

The symptoms pointing towards OCP, so it could be that your PSU is at fault for not being able to handle the GPUs transients.

Gigabyte has no benefit from denying RMA when the card has issues, you posted Videos of cases where the card was at fault yourself and they replaced them.

If you are 1000% convinced that the GPU has issues, you should give independant repair a shot, and if they find anything id communicate this with Alternate and Gigabyte and send them the bill. (diagnostic should be enough)
In this case you'd atleast have evidence of a defect and they cant deny fixing it/or replacing the card then, or you choose to pay the repair shop to fix it. (its probably a voltage regulator module not working correctly, if the card is damaged, shouldnt be to expansive)

[Myself5]

On 5/22/2023 at 2:14 AM, DarkSmith2 said:

All excuses aside. they ran 84hours of Furmark, timespy and heaven even some of it at the same time.

Would agree, if it wasn't on a PCIe System where their own screenshot is showing the card not hitting its max performance (or at least a consistent 99%) once as seen in GPU-Z Screenshot and Furmark in the background. That screenshot was from Gigabytes second test btw.

 

On 5/22/2023 at 2:14 AM, DarkSmith2 said:

At this point, if i would be alternate, i wouldnt send it in again aswell, BTW i highly doubt that alternate even tested your card, when they first got it. It is absolutely unusual to do this as reseller for first time send ins. There is probably nothing wrong with the card at all.

Fair point. But then again it would be them lying they tested it the first time cause they themselves told me that they tested AND replicated the error on their own. I wasn't "asking" if they did and got a "oh sure we did".

 

On 5/22/2023 at 2:14 AM, DarkSmith2 said:

Gigabyte has no benefit from denying RMA when the card has issues, you posted Videos of cases where the card was at fault yourself and they replaced them.

I'm not necessarily saying they don't "want" to fix the card, it's just that their testing measures seem so off that they straight up wouldn't be able to reproduce the issue even if they wanted (again point out the i5 8500T with PCIe 3.0 here). When I sent the card in I couldn't reproduce it on my PCIe 3.0 Intel system either. Which is why I specifically told them about my specs, mentioning AMD CPUs, the fact that it's a Performance Peak issue and probably only happens when connected through PCIe 4.0.

So basically the way they "confirmed" it to be working is not saying anything IMO.

 

On 5/22/2023 at 2:14 AM, DarkSmith2 said:

If you are 1000% convinced that the GPU has issues, you should give independant repair a shot, and if they find anything id communicate this with Alternate and Gigabyte and send them the bill. (diagnostic should be enough)
In this case you'd atleast have evidence of a defect and they cant deny fixing it/or replacing the card then, or you choose to pay the repair shop to fix it. (its probably a voltage regulator module not working correctly, if the card is damaged, shouldnt be to expansive)

Which would break my Warranty (of which I still have 1.5 years left) and probably cause another endless discussion with them. So unfortunately thats not really an option.

 

On 5/22/2023 at 2:14 AM, DarkSmith2 said:

The symptoms pointing towards OCP, so it could be that your PSU is at fault for not being able to handle the GPUs transients.

I had that suspicion as well, but the card is powered by a beQuiet Straight Power 11 1200W PSU with seperate cables on Multirail. Also, if the PSU was crapping out I would suspect the system to crash, which is not the case. Only the Video Output dies. Hell, the card even still seems to produce a picture (as in the Display Driver doesn't crash) given you can still interact with the system and get sound accordingly (e.g. in Games), you just have no output and a 100% fanspeed.

 

 

 

I got the card back yesterday, threw it in my system and now I can't reproduce the issue anymore either... But how would something that has been there prefectly reproducible across different installs and (if Alternate is to be trusted) different systems gone all of a sudden when nobody ever "fixed" anything? Doesn't really add up to me. Especially as Gigabyte finally gave in and told me they would replace the card, indirectly admitting that their i5 8500T test might not have been correct.

 

I will see if I can figure out what has been done to the card or how to reproduce the issue again in the coming days and, if necessary, send it back in. This time directly to Gigabyte who finally agreed to skip Alternate altogether.

[DarkSmith2]

I have a somewhat new speculation about this. There seems to be issues with PCI-E 4.0 and 4090 owners currently.
There is a thread with over 60pages on overclock.net about this, showing similar symptoms. (black screens, sometimes driver crashes) 

https://www.overclock.net/threads/massive-rtx-4090-problems-driver-or-hardware.1801381/

I cant tell for sure if its related to your problem, since you said your driver isnt crashing, but alot of it sounds familiar, maybe this could be the cause of your issue, maybe its not only related to RTX4090's, however most people having this issue upgraded from pci-e 3.0 cards. If you updated your BIOS on your Motherboard and things started happen i would consider it as a probable cause. There is also alot of people suggesting fixes that worked for them, so you might want to read a little bit into it. 

There seems to be a specific Asus Motherboard that is particulary often affected, aswell. Maybe a Bios rollback could fix it for you.
Some stated that the cooler and board combination doesnt allow for proper installation of the card, where the slot bracket would deny the card to be inserted all the way into the slot, which is causing issues with pci-e 4.0 but not 3.0. Really fancy things to read.

[Myself5]

4 hours ago, DarkSmith2 said:

I have a somewhat new speculation about this. There seems to be issues with PCI-E 4.0 and 4090 owners currently.
There is a thread with over 60pages on overclock.net about this, showing similar symptoms. (black screens, sometimes driver crashes) 

https://www.overclock.net/threads/massive-rtx-4090-problems-driver-or-hardware.1801381/

I cant tell for sure if its related to your problem, since you said your driver isnt crashing, but alot of it sounds familiar, maybe this could be the cause of your issue, maybe its not only related to RTX4090's, however most people having this issue upgraded from pci-e 3.0 cards. If you updated your BIOS on your Motherboard and things started happen i would consider it as a probable cause. There is also alot of people suggesting fixes that worked for them, so you might want to read a little bit into it. 

There seems to be a specific Asus Motherboard that is particulary often affected, aswell. Maybe a Bios rollback could fix it for you.
Some stated that the cooler and board combination doesnt allow for proper installation of the card, where the slot bracket would deny the card to be inserted all the way into the slot, which is causing issues with pci-e 4.0 but not 3.0. Really fancy things to read.

Thanks a ton, will read a little into this.

Actually noticed something very intresting in your message already.

 

4 hours ago, DarkSmith2 said:

If you updated your BIOS on your Motherboard and things started happen i would consider it as a probable cause.

My issues didn't seem to happen after a BIOS upgrade (at least not that I could specifically mark it down to that), but while the card was gone I upgraded my BIOS. Maybe that explains why it is working again all of a sudden. Still wouldn't explain why Alternate "reproduced" the issue, but given their entire support system this might just further support your assumption of them not testing it at all.