There are LOADS of different brands (makers) of graphics cards: I can think of MSI, Gigabyte, ASUS, Palit, PYT, Sapphire and Zotac off the top of my head. There’s probably a dozen or so more, too. Then you have NVIDIA’s direct GPUs too, which they call the Founder’s Edition cards.
Do all these brands actually offer something different though? Or can you just buy any GPU in a range (such as the RTX 4070 Super range) and know that you’ll always receive the same sort of performance?
Well it’s ‘kinda a difficult answer, because there are some SMALL performance differences between GPU makers – but some of the ‘peripheral’ differences (like warranty terms and cooling configuration) can be a bigger difference – as I explore in this video.
If you prefer text over video, please read on for the guide/transcript version of this video.
Video Transcript And Guide
Hey everyone, buying a car is SIMPLE. If you know you want an F-150 pick-up, for example, you just go straight to Ford. BUT buying a graphics card is… NOT simple. Let’s say that you have spent countless hours of research and FINALLY decided to buy an RTX 4070, you THEN have to decide whether to buy direct from NVIDIA (which is called the Founder’s Edition variant) OR buy one from MSI OR Asus… OR from Zotac, Gigabyte, Palit or PNY, apparently:
Yeah, so it’s confusing – especially when Amazon COMPLETELY mislabel things, and tells everyone that a white 4070 only supports HDMI (compared to the black version that supports both HDMI AND DisplayPort):
Spoiler alert: every single RTX 4070 card on the market supports both monitor connectors: Amazon is just rubbish.
How GPUs Are Released
So why are AMD and NVIDIA making this all so confusing? Why are there so many different brands? And does the GPU brand ever really matter to consumers? Well before answering those questions, let me back-up for a minute and explain how (and why) graphics cards get released. So making the specific graphical chips that power GPUs is HARD. Not every company can do it, especially because it costs so much money to research, design and manufacture the specific chips.
THIS is really where AMD and NVIDIA can make a lot of money. Here’s the thing though: creating some awesome new GPU microarchitecture with 84 RDNA 3 Compute Units is only half the battle. These chips then need to be sold to us consumers, so that we can play Fornite… I mean, awesome, manly games like Sniper Elite 5. Yeah, Fornite sucks. Moving on.
The process of actually getting a chip onto a consumer grade PCB (a circuit board), installing all the VRAM, cooling and everything else and THEN selling it to end users is not fun.
Firstly, while many of us are probably fairly decent human beings (I’m guessing!), some consumers SUCK – they abuse and misuse products, and then try returning them. NVIDIA and AMD often don’t want the hassle of dealing with this. Equally while enterprise hardware earns AMD and NVIDIA lots of money, consumer boards like graphics cards have much lower margins. There’s more risk there. So it kinda makes sense for the work of turning a GPU chip into a final product to be outsourced. That’s where MSI, Asus, Sapphire, XFX, Palit and all the rest come in: they are called “board partners” and they take specific chips and specifications from AMD and NVIDIA, and then create the final graphics cards to sell.
Yes I know that NVIDIA sells GPUs directly too – calling them Founder’s Edition – and I’ll discuss this point later on in the video, but NVIDIA makes (and sells) a lot less of these than the GPUs they sell via their board partners so I wanted to focus on that point first.
What Stays The Same Between GPU Makes
When you look a specific GPU model – like the RTX 4070 from NVIDIA or the RX 7600 XT from AMD – you’ll notice that the specifications give a specific value, like having 5888 Cuda cores or offering 12 gigs of GDDR6X VRAM (in NVIDIA’s case) or the AMD card that says it has 32 compute units and 16 gigs of VRAM (well actually it says “maximum”, but this is pretty much what you will get with that specific card, no matter who you buy it from).
If you know that you want a certain graphics card model, then all the key specs will be the SAME – and that’s because NVIDIA and AMD will sell the graphical hardware (like the chips) to their board partners with MINIMUM board and cooling specs. In other words, all the different GPU brands have to deliver a certain base level of performance for each GPU model. You’re certainly not going to get 60 fps with Sapphire but only 41 fps with an XFX card, for example.
Now you might be wondering about graphics cards that say “OC” in their overly long and confusing product name. “OC” stands for “overclocked”, and this is where the board partners make the chips run a bit faster – although this is often just a few percent better performance (than the baseline case). The cards don’t use different chips (or anything like that) – it’s mainly just marketing by the brand partners, to be honest.
There WAS a time when overclocked graphics cards were worth purchasing, but nowadays both AMD and NVIDIA offer automatic boosting technologies – and this is something that their brand partners must deliver as part of the minimum specs I mentioned earlier. NVIDIA has GPU Boost for example, which allows the GPU to automatically run faster than the specified clock speed when it’s able to – basically, as long as the graphics card doesn’t get too hot and it doesn’t draw too much power. If a GPU is only rated for 200 watts (on its TDP rating), then a board partner wouldn’t be able to constantly boost the GPU to the extent that it uses up 250 watts – for example.
But despite everything I’ve just said, don’t blindly think “great, I can now go and buy a GPU from ANY brand” and hop onto AliExpress to buy an AMD RX 6090 Arc Super… that’s totally not Photoshopped.
Where GPU Brands Do Have Differences
There are actually SOME differences between GPU brands that are worth knowing about. Firstly if you view the detailed specs for a particular GPU, you might see that they advertise having Boost Clocks but that “Boost clock achievability will vary based on several factors including thermal conditions”.
I’ve shortened that quote a bit, but that’s an interesting point. The cooler the graphics card runs, the easier it is for a GPU to hit the frequencies that AMD or NVIDIA specify. That brings us onto our first main difference between different GPU brands.
The WAY that a GPU is cooled is one point that can vary a decent amount between the different graphics card brands. Yes, AMD and NVIDIA set out minimum specs for cooling, and this leads to ZOTAC’s 4070 card saying things like “advanced cooling” and “IceStorm 2.0”, discussing how air is “360 degrees”.
I mean… yeah… that’s kinda how most graphics cards are cooled nowdays – and how air works too! They have fans on them, and the fans spin. Cool. But marketing jargon aside, the ZOTAC card doesn’t SEEM to have anything too unique here. Compare that to MSI’s card and it does look like more thought has gone into the cooling set-up.
It discusses having multiple fans that work together, and having a copper baseplate as well as various heatsink pipes that run along the full length of the card to help dissipate heat. It says a few other things too and YES I know that this might just be marketing, but in general this specific MSI card does have a better cooling system than the Zotac card.
So this difference between brands does matter… sort of. We know that cards can automatically “boost” their performance when thermal conditions allow – in other words, when a GPU is running cooler. So the fact that some GPU models have better cooling set-ups than others can result in greater performance. BUT let’s not get carried away here. AMD and NVIDIA still give specific performance ranges for their cards, meaning that one board partner’s GPU can’t be 20% faster than another partner’s card – or even NVIDIA’s own ones. Yes the cooling set-up matters, especially if you value a quieter PC when gaming, but it’s not going to lead to 10% or even 5% performance gains.
The second difference I wanted to cover though DOES really matter if you have a smaller case – like a micro ATX or mini ITX. Some graphics cards have two fans, or even just one – making them fairly short, and easier to fit inside small cases. Whereas other cards are “full fat” and have three fans, meaning that they are longer – a LOT longer in some cases. I personally needed to investigate this point a lot when buying my RX 6700 XT card because my old Corsair Carbide 88R case only fits GPUs up to a certain length, and a three fan variant probably wouldn’t have fit. So if you’re working on a smaller form factor build, then this IS a factor that you’ll need to look at – some brand partners just make longer cards (due to their cooling set-up).
The third difference between board partners can also be crucial, and it’s the fact that the partners themselves handle support, returns and replacements. Basically if you have a problem with your GPU and potentially need to RMA it (to send it away), then a company’s customer service levels IS really important here. ASUS, for example, don’t have the best reputation with many bad reviews all over the internet, and some big tech channels like Linus Tech Tips actually cutting ties entirely with them when it comes to sponsorship money. This compares to companies like XFX that have a SLIGHTLY better reputation, and they even used to offer a lifetime warranty for some of their graphics cards – although they did sadly end this program in 2019.
Different companies might also offer slightly different warranty terms too. For example both NVIDIA and AMD specify that their GPUs should have 2 year warranties, and this is generally what XFX now offers in the UK – for example.
Asus and Zotac then offer 3 years of warranty, with Zotac sometimes extending this to 5 years when you register a compatible GPU through their portal. So this is probably the biggest difference between brands of graphics cards. But of course, you should always speak to friends (if you have any!) and check review sites to see whether a particular company offers good levels of service too – there’s no point in having a super long warranty if the company makes it really, really difficult to actually RMA a broken card.
Before moving onto discussing NVIDIA’s founder’s edition, I wanted to quickly point out two other differences between GPU makers. Firstly the aesthetics will vary between cards, of course. Some have a pure white design, others have cool looking fans, some cards are really short while others are super long. THAT’s one benefit of NVIDIA and AMD farming out the GPU delivery to their partners, because it allows for more customization and competition options in the market – which is great!
Finally, some GPU makers offer fairly unique features – like how some Asus cards have a dual BIOS switch that allows you to run the GPU in a lower power mode so that it makes a LOT less noise than the performance mode. So that’s kinda a cool feature, and it sets Asus apart (a little bit) from the other brands.
What About The Founder’s Edition?
So far I’ve discussed how Asus, MSI, Sapphire and all the rest are DIFFERENT to each other, but what about NVIDIA who sell cards directly to the public – known as Founder’s Edition cards? Well firstly NVIDIA don’t manufacture these directly in Jensen Huang‘s oven or something. They ARE still outsourced cards – Foxconn and BYD often have the contract to build these reference boards.
The main difference between the FE cards and the board partner ones is that the FE ones often look “simpler” – but they’re actually fairly nice looking. They’re clean and minimalist, which many of us actually value ABOVE sparkly GPUs with “flashy” looks. Secondly, NVIDIA won’t bother with fancy cooling systems or “overclocking” their cards – you will get the base level of performance, but not the extra 1 or 2% of performance that board partners sometimes squeeze out of their “overclocked” cards. Of course, there’s nothing stopping you from removing NVIDIA’s own cooler, switching to liquid cooling, and then overclocking the card in GeForce Experience. That’s actually why some people buy Founder’s Edition cards – because they don’t have the extra cruft that the partners add in – they want to buy directly from NVIDIA and then customize it.
In short, there was (sort of) a time when the BRAND of GPU mattered – but nowadays, due to things like automatic boots and minimum specifications from NVIDIA and AMD, the main differences come down to “extras” like cooling systems and warranty terms.
If you have any questions, feedback or suggestions about this article, please leave a comment below. Please note that all comments go into a moderation queue (to prevent blog spam). Your comment will be manually reviewed and approved by Tristan in less than a week. Thanks!