Nvidia might throw out the GPU rulebook with rumored RTX 4080 Super and 4070 Super launch
Two cards but four chips?
Nvidia could use different GPUs in its incoming RTX 4080 and RTX 4070 refreshes, as yet another rumor comes flying onto the scene around the purported next Lovelace graphics cards from Team Green.
The theory that the rumored RTX 4080 Super and 4070 Super could mix their chips is a new one, although there has been plenty of chatter about Nvidia switching the GPUs – using AD102, which powers the RTX 4090 currently, in the 4080 Super – already. (The vanilla RTX 4080 uses AD103).
In my opinion there may be chip mixing on 4080S and 4070S. That means 4080S can use AD102 or AD103, 4070S can use AD103 or AD104. Think about EVGA 2060KO.October 25, 2023
Kopite7kimi has proposed a different theory – and note, it’s framed as just the leaker’s opinion, so that’s not a particularly firm assertion – that chip mixing could mean both AD102 and AD103 are used with the RTX 4080 Super.
This means that some models will use a technically much better chip than others, and the same is apparently true for the RTX 4070 Super with it using both AD103 and AD104.
While this kind of move is not unheard of (we’ll come back to that), it’s unusual to think there’d be so much in the way of mixing (of course, there may not be – this is just a rumor).
At any rate, if true, we’d end up with two different spins using two different GPUs for each Super refresh. And one of each spin would have a superior spec to the other – so why would you buy the lesser one? Well, that’s the crux, in that just because some of these Super graphics cards use a higher-tier chip, they can have the same spec.
Let’s dive into this a bit more…
Get the best Black Friday deals direct to your inbox, plus news, reviews, and more.
Sign up to be the first to know about unmissable Black Friday deals on top tech, plus get all your favorite TechRadar content.
Analysis: Super confusing
So, let’s focus on the RTX 4080 Super for the moment, as it’s the more interesting one here as there are already several rumors out there about the graphics card using the AD102 GPU.
If a theoretical RTX 4080 Super is built with AD102 from the RTX 4090 which has many more CUDA cores, and other versions use AD103 with fewer cores, Nvidia can have it so they have the same core count. All the company does is disable cores on AD102 to match AD103.
Exactly this happened with EVGA’s RTX 2060 KO Gaming graphics card back in 2020, as Kopite7kimi points out in the above tweet. This card used TU106 (which was the GPU for the 2060 and 2070 models) as well as TU104 (a higher-tier chip used for the 2070 Super and 2080 models). However, both of those were run with the same core count.
Why do this? Well, it was a way of using up TU104 chips that weren’t good enough for the RTX 2080 (or 2070 Super), with any defective cores simply being disabled. In short, it’s a way of getting more mileage out of imperfect higher-tier GPUs, and this is what Nvidia will be doing with the AD102 chip in the RTX 4080 Super, or that’s the theory.
One existing rumor around AD102 being in the 4080 Super, though, made the case that Nvidia might be using the GPU to up the VRAM to 20GB – because AD102 is needed to get the required memory bus (320-bit) to run that much video RAM.
If Nvidia is plumping for AD102 simply to use up chips that aren’t good enough to make the cut for the RTX 4090, as is the suggestion, that rather torpedoes the 20GB of RAM theory, as Team Green would keep the spec essentially the same with both AD102 and AD103 spins. It hints that it might be the case that the source of the previous RAM rumor may have got the wrong end of the stick from whatever their insider told them, perhaps.
(Although to be fair, none of this definitively rules out that VRAM rumor – remember, this isn’t a firm contention from Kopite7kimi as we already observed).
As for the RTX 4070 Super, the most recent chatter has indicated that this may also come with a different variant – called the Ti Super – so in this case, the AD103 version floated here might be configured to be a touch more powerful than the ‘vanilla’ Super, as it were. Although we still remain unconvinced about that naming scheme.
Take all of this with a large amount of seasoning, mind you, as it’s entirely possible these strands of rumors are just concepts and possible paths forward being explored at Nvidia right now, which may all come to nothing. Given the confusion around all this, we suspect that future Lovelace plans are indeed up in the air right now...
Via VideoCardz
You might also like
Darren is a freelancer writing news and features for TechRadar (and occasionally T3) across a broad range of computing topics including CPUs, GPUs, various other hardware, VPNs, antivirus and more. He has written about tech for the best part of three decades, and writes books in his spare time (his debut novel - 'I Know What You Did Last Supper' - was published by Hachette UK in 2013).