AMD’s next-generation graphics cards are rumored to differ somewhat from what was previously expected in terms of specs, as a new leak occurs – and while there’s a bit of disappointment about some of the confirmations made in this new spill, there’s really nothing to worry about. In fact, Nvidia will probably be concerned about some of the information presented here…
Before we get into why this happens, let’s examine the leak itself, which comes from Angstronomics (via VideoCardz (Opens in a new tab)), a publication covering the semiconductor industry we’ve heard about before. (Even if it’s not one of the more regular sources of information revealing on Twitter, we consider Angstronomics to be reliable enough — plus a major YouTube leaker more or less corroborates this in-depth detail, and we’ll get back to which — which).
Angstronomics draws the line on RDNA 3! We detail the key specifications of the Navi31, Navi32 and Navi33 that were completed in 2020 and have not changed since then! We also highlight some of the architectural changes including OREO😋WGP, Cache, and template sizes inside! https://t.co/n2qB7KiiBgAugust 12 2022
Key points of this speculation include that AMD is focusing on “area, area, area,” meaning smaller chips that are still powerful enough to meet Team Red’s performance goals for RDNA 3, as well as focusing on performance per watt (efficiency). Of course the company already boasted a 50% increase over RDNA 2 with the latter (the same RDNA 2 leap made over the original RDNA architecture).
Freely spice up your spice with all this, but Angstronomics tells us that the Navi 31 – the flagship model for the RX 7000 graphics cards – will be rumored to use a single graphics chip die (GCD) as well as six memory chip blocks (MCDs), with 12,288 cores (known as ALUs). ). The main GCD size will obviously be 308mm², and another big change from the previous speculation mentioned is the Infinity cache size which appears to be 96MB for the top dog model (Navi 31, most likely the non-XT version, will use 80MB).
The rumor mill mentioned much larger sizes than that, like 192MB; Indeed the current Navi 21 uses 128MB of Infinity Cache, so it will actually be smaller than that. Angstronomics believes that AMD tried twice as much cache as the Navi 31, but that didn’t work to provide enough extra performance that it was worth doing.
Hence some slight disappointment about this rumored spec, and the second point that caused a bit of concern comes with the alleged configuration of the Navi 32, which Angstronomics confirms will work with 7,680 cores instead of the 8192 as previously thought. Apparently the Infinity Cache was set to 64MB, and possibly 128MB for a 3D stack model (but then again, it looks like AMD might have thought about this idea, and then abandoned it because it didn’t make sense. Performance gained. additional).
As for the Navi 33, the mid-range display, that would be a lot smaller if Angstronomics were for the money, as the chip was previously rumored to have a size of perhaps 400mm², but in fact it weighs as much as 200mm². It is supposed to work with 32MB of Infinity cache.
Regarding the cooling system of the RDNA 3 main graphics card, AMD is believed to keep a triple fan system similar to the current main one, but it will be a little longer. The GPU will use a pair of 8-pin PCIe power connectors, according to the leaker.
Analysis: Worried about all this? Well, maybe it’s Nvidia that should be worrying…
As mentioned earlier, some people have been concerned about elements of this purported spec, some parts that appear to have been trimmed a little – or a lot in the case of the Infinity Cache – compared to previous chatter from grapevine. However, you really don’t have to worry, and in fact, it’s probably Nvidia that should be doing that worry, and here’s why.
It is interesting that Moore’s Law died (Opens in a new tab) (MLID), a prominent leaker on YouTube, quickly got into the matter and, after speaking to multiple sources, pretty much supports everything Angstronomics puts forth here. There is one major difference, however, which is that for the Navi 32 GPU, MLID still hears it will be 8,192 cores, instead of the 7,680 cores mentioned in this new leak. Of course, MLID still admits that its sources may be wrong, so it could go either way so far (and 7,680 kernels sure are possible).
MLID sources back up to a wide range of spec fidelity, smaller template sizes mentioned here, and fewer Infinity Cache loads – but that’s the main thing on that front. MLID notes that the performance ratings it hears haven’t really changed, and that it still expects AMD to maintain the rumored performance jumps we’ve heard about in the past — enough to compete with Nvidia’s RTX 4000 range — even if the specs are tweaked along the lines above.
Even if performance ends up being a touch weaker than Nvidia’s, there shouldn’t be much difference in MLID theories. And we have to remember that given the smaller die sizes mentioned, it would be cheaper for AMD to make these 3 RDNA GPUs — and thus pricey to attack the Nvidia RTX 4000 models. That might be the most exciting thing about next-gen cards, if AMD gets aggressive with pricing. Nvidia could be in trouble—particularly seeing power-hungry GPUs (at least through the rumor mill) could mean other complications like a PSU upgrade.
Not to mention, Nvidia should have a hard time deciding on its next-gen Lovelace launch schedule, which could allow AMD to get there first with RDNA 3 graphics cards, and take advantage of the sales lead above all these other factors. We shouldn’t get carried away by any of these rumors circulating, of course, but this latest spill should be worrying for Nvidia on several fronts, we imagine.
For Intel – Angstronomics reports that the Navi 33 beats its flagship Arc Alchemist GPU while being more power efficient and costs less than half that amount, which would obviously be a torpedo to sink Team Blue desktop GPUs (and they’re already struggling severely even as it is, without next-generation competing products).