Retailer says Radeon 5000 series has higher failure rate than Turing cards

midian182

Posts: 9,633   +120
Staff member
In a nutshell: In the battle between AMD’s and Nvidia’s most recent graphics cards, it’s fair to say that team green’s series has been the more successful by far—but are they more reliable? Yes, according to statistics from one German retailer, though the RTX 2080 Ti has the highest RMA rate.

Germany’s largest PC retailer, Mindfactory, has published RMA (Return Merchandise Authorization) data for AMD’s Radeon 5000 series (5500 XT and above) and Nvidia’s Turing cards (GTX 1660 Ti and above).

The first stat that jumps out is just how much Nvidia outperforms its competitor in terms of sales. The data shows that 76,280 cards from Team Green were sold, while AMD managed 44,100 GPU sales.

When it comes to failures, Nvidia had 1,607 units RMA’d and AMD saw 1,452 units. Taking into account the total sales numbers, that means Nvidia had a RMA rate of 2.1 percent and AMD had a 3.3 percent rate—about 50 percent more than its rival.

Graphics card series RMA Rate
AMD 5700 3.6%
AMD 5700 XT 3.58%
AMD 5600 XT 1.99%
AMD 5500 XT 0.95%
Nvidia RTX 2080 Ti 5.35%
Nvidia RTX 2080 3.63%
Nvidia RTX 2070 3.04%
Nvidia RTX 2060 Super 1.85%
Nvidia RTX 2060 1.82%
Nvidia RTX 2070 Super 1.24%
Nvidia RTX 2080 Super 1.24%
Nvidia GTX 1060 Ti 1.20%

Looking at individual series’ RMA rates, the RTX 2080 Ti has the unwanted distinction of the highest failure rate: 5.3 percent. The GPU SKU that failed most, however, was the Red Devil PowerColor AMD RX 5700 with a 13.18 percent failure rate. PowerColor, whose video cards are limited to AMD lines, had the worst RMA rate of all vendors at 5.59 percent.

On AMD’s side, the company’s 5700 and 5700XT cards had a failure rate of 3.6 percent, but the 5500 XT’s RMA was just 0.95 percent.

Why the cards were RMA’d isn’t revealed, and it’s best not to read too much into data from just one retailer. Moreover, a particular batch could have shipped with problems, and improvements to drivers often mean later cards are more reliable than their predecessors. Still, Nvidia will be happy.

Permalink to story.

 
Looks like the Super refresh are much more reliable than the original lineups.
For 2080 Ti I think the RMA rate was inflated slightly due to people just keep exchanging them in search of "golden chip". Still I must say 2080 Ti sales are pretty good for 1200usd+ GPU.
 
I think everybody should study the sheet carefully.

MSI really shines on the said doc. 1% failure rate. I never owned an MSI card but I will seriously consider getting one next upgrade cycle.

Palit, another China co. returned abt 3% failure rate. I guess I am lucky my Palit Game Rock 1080 has lasted so long.
 
These numbers are interesting, but obviously they don't have that much to do with AMD or nVidia themselves, and more about the quality of the partner cards.

Asrock is surprisingly low for AMD cards, which is interesting. They are the newest manufacturer and are doing well hardware wise, which is good to see. Even the card that was considered a garbage one, the Asrock 5700(X) Challenger has a low RMA rate. The second best is Sapphire, which is no surprise. They are my go-to brand for AMD cards and I've always considered them reliable. PowerColor is the worst, which to me is also no surprise... XFX is worse than expected though.

It's also interesting that the MSI 2080Ti Lightning Z has an 11% RMA rate. And despite this, MSI seems to be more reliable than Gigabyte... Zotac is the most reliable, which again to me is no surprise. It's the same parent company as Sapphire. Palit is the worst, which is again expected.
 
No surprises here, everyone knows AMD has poorer quality standards for their partners than Nvidia do.
 
This is bait
No it isn’t. It’s fact, the power standards, clocking standards are much tighter on Nvidia boards. Most of the time people criticise Nvidia for locking things out on boards! But here we have clear data of Radeon cards being more likely to fail than GeForce cards. If it’s not quality control then what is it? Are you suggesting that it’s not the board partners fault? Is it the GPUs themselves?
 
No it isn’t. It’s fact, the power standards, clocking standards are much tighter on Nvidia boards. Most of the time people criticise Nvidia for locking things out on boards! But here we have clear data of Radeon cards being more likely to fail than GeForce cards. If it’s not quality control then what is it? Are you suggesting that it’s not the board partners fault? Is it the GPUs themselves?

If it's fact you must have links for said standards. Provide them.
 
Last edited:
Like other posters already pointed out, the higher return rates for AMD based cards are mostly due to Power Color's very high RMA rate. For a reliable nVidia vs AMD comparison these neee to be discarded as they are skewing the results.

Other than that, RMA rates are very close.

A positive mention is deserved by the 1660Ti. It has an RMA rate of 0.01% which is great.

 
Like other posters already pointed out, the higher return rates for AMD based cards are mostly due to Power Color's very high RMA rate. For a reliable nVidia vs AMD comparison these neee to be discarded as they are skewing the results.

Other than that, RMA rates are very close.

A positive mention is deserved by the 1660Ti. It has an RMA rate of 0.01% which is great.
Nope. PowerColor sell lots of AMD cards. Their results should count. I don’t understand how it’s fair to discount them. I understand you are here on damage control unofficially on behalf of AMD. But there is no logic in removing PowerColor at all. If you are a true AMD fan you should be sending strongly worded emails to PowerColor for letting your side down.

You can do all sorts of tweaks to the data, if you remove the 2080ti then GeForce is by far better than Radeon. But that wouldn’t be fair, although fairer than removing an entire board partner.

Nvidia cards are less likely to fail than AMD cards. The data shows it clear as day.
 
He has no proof just an intel / Nv fanboy. Any story to show something negative about amd and you can better a paycheck he will be posting some nonsense.
Bwahahaha I do actually. Nvidia launched their greenlight programme some years ago. The tech press criticised Nvidia for it at the time. As far as I am aware AMD don’t do this, there are no checks on AMD partner boards by AMD.




So, no I am not talking nonsense. I’m trying to teach you why this might have happened...
 
I haven't had any issues with my PowerColor RX 5700 XT Red Devil, but I don't doubt the data because of all the issues reported with AMD drivers. Issues I haven't personally had with this card, but did have with my ASUS R9 390 Strix. Issues that go back for years without a proper response from AMD until this year.
 
Last edited:
If you are a true AMD fan you should be sending strongly worded emails to PowerColor for letting your side down.

There is no point in sending them a strongly worded email. I knew 15 years ago not to touch that brand and still don't to this day. There are far better brands on the AMD side and the numbers clearly show why PowerColor is to be avoided.

That is the only thing I get from this story is to avoid that brand the rest are fine. RMA's happen to any brand I've been fortunate that I've never had to RMA in 20 years of building with mostly ATI/AMD videocards.
 
There is no point in sending them a strongly worded email. I knew 15 years ago not to touch that brand and still don't to this day. There are far better brands on the AMD side and the numbers clearly show why PowerColor is to be avoided.

That is the only thing I get from this story is to avoid that brand the rest are fine. RMA's happen to any brand I've been fortunate that I've never had to RMA in 20 years of building with mostly ATI/AMD videocards.
Yeah I learnt to avoid PowerColor too. But I do think AMD need to be approving partner boards like Nvidia do. It would definitely help people who end up with a PowerColor card!

I believe the reason AMD don’t is not because they don’t want to. But because the partners don’t want to go through this extra approval. Nvidia are far bigger and partners can’t really say no. But with AMD they probably have more leverage.
 
I believe the reason AMD don’t is not because they don’t want to. But because the partners don’t want to go through this extra approval. Nvidia are far bigger and partners can’t really say no. But with AMD they probably have more leverage.

Extra validation cost money.

Is it that NV is big and they can't so no or NV pays for the extra cost?
 
Extra validation cost money.

Is it that NV is big and they can't so no or NV pays for the extra cost?
I definitely think it is something to do with Nvidias size. It would cost the partners more to go through approval. I think it’s because they can’t say no to Nvidia. But they could turn down AMD who are a lot smaller so AMD don’t ask to validate parts.

Personally I do think something could be done, returning a faulty card is a bad user experience and if it’s got aRadeon GPU on there it’s going to reflect on AMD.
 
Personally I do think something could be done, returning a faulty card is a bad user experience and if it’s got aRadeon GPU on there it’s going to reflect on AMD.

Ι agree. This is bad press for AMD and the ppl who buy Powerfail China cards and return them are disappointed customers who likely won't buy AMD GPU's again.

AMD has to do smt.
 
Nope. PowerColor sell lots of AMD cards. Their results should count. I don’t understand how it’s fair to discount them.

It is true that Powercolor represents a quarter of all AMD based cards sold by Mindfactory. However, their RMA rate is three times that of e.g. Sapphire.
As the same AMD GPU sold in a Sapphire card has a 2 % RMA while it‘s 6% for Powercolor, this is a strong indication that there is a problem with Powercolor cards in particular. Their RMA rate is also not in line with any other AMD GPU RMA, so if you want to compare AMD to nVidia, you actually need to not exclude them.
The same of course applies to any OEM with an abnormally high nVidia card RMA rate.

I understand you are here on damage control unofficially on behalf of AMD.
Nah, in that case I‘d argue that the RMA were due to people looking for golden samples.

You can do all sorts of tweaks to the data, if you remove the 2080ti then GeForce is by far better than Radeon. But that wouldn’t be fair, although fairer than removing an entire board partner.

Nvidia cards are less likely to fail than AMD cards. The data shows it clear as day.

For a valid statistical analysis, you need to tweak the data to look for anomalies. Since we want to compare nVidia to AMD GPU, we are looking for a causal relationship between GPU manufacturer and RMA rate. If one OEM shows noticeably worse results than others, we can infer a causal relationship between the OEM and RMA rate. Since that is not what we are looking for, that data needs to be excluded.

As for the higher 2080Ti RMA rate, it could be due to the defective cards after launch (if that time frame is covered here) but an alternative explanation could be that buyers who chose a 2080Ti do so because they want the best and are not willing to put up with minor issues that buyers of $300 cards would not be bothered by.

Edit: I am not arguing that AMD based cards are returned less than nVidia based ones, however the difference in nVidia‘s favor is not as large as it may at first appear due to skewed data.
 
It is true that Powercolor represents a quarter of all AMD based cards sold by Mindfactory. However, their RMA rate is three times that of e.g. Sapphire.
As the same AMD GPU sold in a Sapphire card has a 2 % RMA while it‘s 6% for Powercolor, this is a strong indication that there is a problem with Powercolor cards in particular. Their RMA rate is also not in line with any other AMD GPU RMA, so if you want to compare AMD to nVidia, you actually need to not exclude them.
The same of course applies to any OEM with an abnormally high nVidia card RMA rate.


Nah, in that case I‘d argue that the RMA were due to people looking for golden samples.



For a valid statistical analysis, you need to tweak the data to look for anomalies. Since we want to compare nVidia to AMD GPU, we are looking for a causal relationship between GPU manufacturer and RMA rate. If one OEM shows noticeably worse results than others, we can infer a causal relationship between the OEM and RMA rate. Since that is not what we are looking for, that data needs to be excluded.

As for the higher 2080Ti RMA rate, it could be due to the defective cards after launch (if that time frame is covered here) but an alternative explanation could be that buyers who chose a 2080Ti do so because they want the best and are not willing to put up with minor issues that buyers of $300 cards would not be bothered by.
You can’t just remove the results of a board partner because you don’t like them. They may be worse than other partners but that doesn’t mean we can’t count that data. What are AMD doing to fix this? What has gone wrong with AMDs quality control to lead to such a high return rate from one board partner?

I understand that you don’t like hearing criticism of AMD and you are attempting to deflect this from AMD onto AMDs partner PowerColor. But they are “partners”. AMD definitely should take some responsibility. Otherwise whats to stop any old cheap manufacturer from buying some AMD GPUs, chucking them on some cheap custom board, slapping the AMD Radeon branding to it and ruining AMDs reputation?
 
Back