AIB Partner quality discussion

[Niku-Sama]

considering now that EVGA is no longer going to be producing AIBs for nvidia i thought i would bring this up since i think they were regarded as a high quality AIB Partner.
so who makes a decent quality board and who doesnt?

i dont have much experience with many of them from the ones that i do i can say:

MSI is OK, they dont seem to break down or have cooling problems like some but theres room for improvement
Gigabyte is Crap, every video card of theirs i've had has had problems in one form or another, wither it be crap cooling solutions with defective parts or just outright dead cards for no real reason. Will not buy again.

XFX/Pine no longer does nvidia cards but their warranty sucked when they did, Sapphire is AMD/ATI, and PowerColor i cant provide input on because the one and only card i had of theirs was a GeForce 2 GTS

what are your experiences with the current AIB partners for nvidia

 

[TerranBrackiatt]

i'm curious as well. my first thoughts are that i'll probably go with ASUS cards (even though i kinda hate ASUS's software), but i dunno yet.
the most QA digging i really did was when the early RTX 20x0 were crashing hard
one of the youtube channels pointed out the 6 chips on the back of the board above the GPU cores, and at least 2 of them needed the high grade version. certain ASUS (i think, memory might be Bacon Deficient again) cards had 6/6 high grade versions while some others had 0/6, most had 2/6 high grade versions for stability. the ones with 0/6 were the ones that were crashing and burning.

 

[Celeron_Phreak]

When I do builds for other people and eVGA is not an option for them, I have never had any issues with ASUS Strix, and as of recently, TUF cards. I think ASUS will be my next AIB choice now that eVGA is done with the GPU game, and if ASUS is not available, Founders Edition, and lastly MSI.

If nVidia becomes a crapshoot, I had great success with XFX cards back in the 6800/7800 GT/GTX days.

 

[Woomack]

Every single brand had some problems with the quality of at least a single product, but some brands repeat the same mistakes for years. I remember when ASUS HD7970 Matrix was dying because of design flaws. I sent 2 of them to RMA and sold the last replacement. You would expect it's the top quality card.

I see it like this:
- Sapphire - makes probably the most reliable AMD cards, but their availability has been limited in recent years. In the last years (~3 years or something), fans can be sent separately to RMA without removing the cooler.
- XFX/HIS - both brands are one nowadays, and both had big problems with the quality of some series. I remember mixed memory IC, too relaxed timings, problems with coolers, overheating, and problems with PCB design. On top of that, I had one review sample which was supposed to be a retail product, and it was soldered/covered with paint. Later I got info that the whole batch had design flaws, and they manually corrected that. The last 2 XFX generations are much better, with pretty good and quiet coolers and better power design (I tested at least five different cards), so I assume they made big changes, and hopefully, it will improve. Just to add something, the current XFX is not the same brand as it was 5+ years ago.
- PowerColor - some cards are great, like the Red Dragon / Red Devil series. Some lower cards are clearly not the best and have some design flaws. There were repeatable cooler issues in some series. However, most of their sales are in Asia, and not all models are available in the EU/US.
- MSI - some cards are great, and some are mediocre. I mean, you can get a card with good PCB/VRM but below average cooler. Power limits are usually lower than on competitive cards, even lower than reference. Often, they design cheaper power designs or cheaper coolers for the most popular cards. They used to have better cards compared to the competition.
- Gigabyte - the same as MSI; some cards are great, and some cards are below the average. They release many models, some not much cheaper and have worse coolers like fewer heat pipes, no copper cold plate, and some more. Many cards have problems with coolers. Most problems are related to the most popular and cheaper models.
- ASRock - pretty good high-end models, just average everything else. It's not a popular graphics card brand, and hard to find them in many countries, so there is also not much feedback about quality. I have one, and it's great, but the design is a bit weird.
- ASUS - most cards are high quality, so I could say it's the best option if you pick Nvidia. Some models had single problems, but they didn't repeat the same issues later.
- Palit/Gainward - the same brand, almost everything pretty average quality. Low price, but when you compare it to other brands, you know why. Everything has warranty stickers - no sticker = no warranty.
- PNY - the same as Palit/Gainward
- Galax - I have mixed feelings about this brand. Top OC series seem great, but my HoF cards were whistling under load so loud that after tests, I sold them as I couldn't stand it.
- Colorful - some cards look interesting, but they sell them almost only in Asia and barely ever send any review samples, so I had no chance to check them.
Maybe I missed something.

In short, for me, ASUS is a choice for AMD and Nvidia. Sapphire is a choice for AMD. Everything else depends on a specific model.

 

[EarthDog]

I see it like this......

Every single brand had some problems with the quality of at least a single product,

So, people pointing at one or another brand as better or worse, is just down to their subjective experience. The reality is, RMA percents for almost every brand is extremely low. The chances of failure from a specific brand are, largely, not significantly different from brand to brand. A percent or two here or there doesn't make a difference.

Call me a sympathizer(?), but outside of something going 'really wrong' with a SKU, the chances of something going bad are pretty slim. RMA % are often low single digits. For example, Puget systems said overall they had an RMA rate of 1.63% for AIB graphics cards.



While this is clearly not The Gospel, it gives us a general idea of how few are actually RMA'd during the native warranty period (DOA and from the field). With information like this, is there really a significant difference between brands on the hardware side of things (outside of perhaps an outlier card that has a lot of failures)?

This is from 2018....

In addition to the supply constraints, there were also a number of software and driver issues we had to work through in 2018. However, from a physical hardware standpoint, the failure rates were pretty decent. Overall, we saw a 1.11% failure rate overall (1 in every 90), while our field failure rate after our production and QC process was just .5% (1 in every 200).

Interestingly enough, if we sort according to consumer (GeForce), workstation (Quadro), and the in-between Titan line, the Titan cards actually ended up being the most reliable. For Titan cards, we saw just a .45% failure rate overall (1 in every 222) with no failures at all for our customers in the field. For reference, GeForce was 1.24% and .62% for overall and field failure rates respectively, while Quadro was 1.16% and .39%.

Think about it!

EDIT: Now, that said, there's going to be some inevitable differences in cooler performance, noise levels, appearance, hardware specs (VRM count/type, clocks), etc... but that's, to me, a preference thing where an educated consumer with time, goes and looks at reviews and buys the card with the best specs appropriate for them. I'd buy any of the major partners cards without flinching.

 

[Woomack]

Answering to the topic title, I wasn't thinking about the RMA rate, but about the design quality of various brands. If we look at the RMA rate then you are right, no matter what brand you pick, it will probably work, at least stable. However, how it will work depends on some other factors (mainly power and cooling design).
Most users won't make RMA for coil whine, loud fans or overheating (as long as it only slightly throttles GPU, not causes it to freeze or crash in games), but they will complain around the web.
I just see that some brands make mistakes while designing PCB, power section or coolers ... or are just cheap. It still doesn't mean that these cards will go to RMA. We (read as overclockers/enthusiasts) see these things in a different way. A typical user won't make RMA for many things that we could call a design flaw. They will simply use it and assume it just works this way.

 

[EarthDog]

I think RMA rate has a lot to do with the build quality of various brands and answering the topic title. Low-quality parts should = higher RMA rates.....high-end less, right? Isn't that how it's supposed to work? But does it, really? Tough to prove since we don't have data which is why I like to look at RMA rates. What does it matter if it's using quality components and fails more (for w/e reason)? I'd look at the end result before I would look at most other things knowing how little a difference it (seemingly) is.

But yes, you're correct, some have minor issues but two of the reasons (loud fans and overheating) is something one should catch in a review. If your fan gets loud, RMA it...it isn't our fault some choose to keep it, right? Now, that's an enthusiast thing for the most part, but as you said, few RMA for coil whine and loud fans....but it's not a failure if nobody is turning it in as one.

Sadly, we can't disseminate between models on the Puget data to see which tier SKUs crap the bed, but worth noting is one of these from Puget listed the FE cards as the least problematic.... and we all know those are built with mediocre components and cooling. So it's tough to judge. What is the difference between say, the TUF models and Ventus....or Suprim X and ROG w/e? Is there really any outside of paper specs? We can look at say Ohhhhh 110A SPS Fets and ample, quiet cooling, but in the end, is it quality if it fails and needs to be RMA'd? That's why I look at RMA as part of build quality, knowing that most other specs are largely irrelevant unless you're trying to take these locked-down monsters cold.

 

[BugFreak]

I've only used a couple nVidia brands myself so can't speak to most. I thought eVGA handled the supply issues the best over the past couple years so I switched to them with this round of cards but prior was mainly ASUS. Not particularly fond of the ASUS up charge but that's life I guess.

With all the apps manufacturers want for driver updates, RGB control and all that I try to keep my brands to a minimum and I usually use ASUS for my motherboards so the switch back won't be a big deal for me.

Doing an RMA is a giant PITA regardless so I really haven't considered that.

 

[dejo]

My experience is that- I had a great experience with Power Color during an RMA for an ATI 5830HD. I have never needed to RMA an EVGA card. I had a horrid experience with MSI. I built a gaming pc for a close friend that included an MSI GTX980Ti. This card failed after about 1 yr. When I setup the RMA and got confirmation, they told me they didnt have any 980Ti cards to replace. They then stated they were going to send a check for $300- WHAT!! I can buy a GTX1060 for that money, and it also wont come close the the performance of the GTX980Ti. But... That is all they would offer. I have sworn off MSI products since that time. Sadly- I always liked their hardware and had good luck with most of them.

 

[EarthDog]

Doing an RMA is a giant PITA regardless so I really haven't considered that.

Right, so wouldn't users want one with the lowest RMA rates regardless of tier or hardware specs?

I mean you need to compare apples to apples (same tier vs tier), but, seeing as how the performance and overclocking headroom (ambient) between cards is very little, does it matter (serious question)? Outside of cooling capability and noise, I can't say I would care too much about how many power phases and amps the VRMs have on GPUs.

Am I missing something in that thinking?


EDIT: I'm not going to share my experiences because good or bad, they are but one card of 10s of thousands sold and a drop in the bucket compared to successful RMAs and not. I've had cards/boards/stoage go bad across almost every brand at some point in my PC life.

 

[Woomack]

EVGA was the top choice if you wanted custom water cooling. They never made problems if you removed warranty stickers, and most water cooling brands were designing full-cover blocks for EVGA. Try to find full covers for anything less popular. Also, EVGA always had top support.

About the power design. AMD forced partners to use their design or better in all higher Radeon cards, so there are no issues with power delivery for all RX6xxx cards. Problems are only with some fans, as I mentioned earlier. Some brands save on things they shouldn't, especially when graphics cards are so overpriced and their margin is always high.

 

[EarthDog]

About the power design. AMD forced partners to use their design or better in all higher Radeon cards,

NV does this also, right? Reference (FE) spec or better...? They don't let partners make lesser base designs?

 

[Woomack]

NV does this also, right? Reference (FE) spec or better...? They don't let partners make lesser base designs?

If I'm right, then Nvidia is not forcing on their partners' power design or anything else but warranty stickers. They only provide reference designs and suggest how the product should look/work (firmware, max voltage/current/clocks etc.). Warranty stickers are a must if you want to have GPUs under warranty at Nvidia. What you (as a graphics card manufacturer) do with RMA from clients, it's your problem. It doesn't mean that Nvidia will replace/give back money for all returned GPUs.

 

[EarthDog]

If I'm right, then Nvidia is not forcing on their partners' power design.............They only provide reference designs and suggest how the product should look/work (firmware, max voltage/current/clocks etc.).

Right, so, AIBs won't put out anything less than the reference spec that NV provides. Same as AMD?

 

[Woomack]

Right, so, AIBs won't put out anything less than the reference spec that NV provides. Same as AMD?

Nvidia is not forcing ... so AIBs can use something less than the reference spec and sometimes they do but usually not much below the specs. There were single cases of graphics cards with one power phase below the reference design. Most AIBs won't release high graphics cards with weaker power design as they are asking for trouble.
AMD has it in warranty terms for the RX6xxx if I'm right.

 

[EarthDog]

Wow... I haven't come across such a thing, that's good to know if that's actually how it works for them. Noted!

 

[Niku-Sama]

hence the reason for starting this.

i've read/seen reviews about some cards having absolute crap power delivery. Some manufacturers are more likely to do this than others so i figured i'd ask and see what experiences people have had with them to kinda get a rough idea.
of course its ever changing but still something to consider

 

[Nebulous]

Asus doesn't have a good track record when it comes to their Top tier cards. Their Strix line, tho sexy looking cards, have a poor record for being reliable. [H] has a bunch of threads discussing it. I was never a fan of Asus anyways. I had several of their boards back in the day and 9 out of the 10 boards I had, had to be sent back for RMA. Their RMA is atrocious. Every board I got back never worked right and had to be sent back. I washed my hands of Asus since that time.

If the selection of Nvidia AIB cards keep dwindling, I won't hesitate to jump back to team red. I was a team red member wayy before I decided to try Nvidia and it won't stop me to go back. IMHO AMD cards had better image quality over Nvidia anyways. I remember having my very first Radeon 9800 PRO and it was da bomb diggity. Never had any issues with their catalyst drivers either.

 

[Niku-Sama]

good to know, i stopped buying asus products a long time ago but it doesnt sound like anythings been fixed in that time.

 

[EarthDog]

Ya'll scare me.

I suppose it's really bad if the average is 1.5% and a brand doubles it (zomg threads, anecdotes! Oh my)... but if you look at like 97% success vs 98.5% is it really different? I'd go all in with both of those odds every. single. time.