What Sony’s PS5 specs inform us about AMD’s Big Navi playing cards

As massive {hardware} spec reveals go, yesterday’s PS5 specs deep-dive was fairly darn dry. Sony’s Mark Cerny spent no less than 50% of the 50-odd minute presentation speaking concerning the console’s SSD, and we nonetheless don’t have any concept what the factor truly appears to be like like. It was simply summary graph after summary graph, a bit just like the one up the highest there, which I feel is supposed to painting how sure games have an effect on the PS4’s fan noise and energy consumption. “Yep, axes are definitely more power intensive than cars, that sounds about right, yep, for sure,” she says nodding sagely.

Nonsensical graphs apart, although, Cerny did fortunately spend a little bit of time speaking concerning the PS5’s GPU, which, just like the GPU inside the Xbox Series X, relies on a customized model of one of many upcoming AMD Navi GPUs. Admittedly, the Xbox Series X reveal earlier within the week had barely juicier details about what we will anticipate from AMD’s “Big Navi” graphics playing cards once they come to PC, however there have been nonetheless a few attention-grabbing titbits in Sony’s PS5 presentation which might be price taking a better have a look at. For starters, it appears to be like like 16GB VRAM playing cards might be very a lot a function of AMD’s 4K line-up. Here’s what it’s essential know.

Our associates at Digital Foundry have much more data on the PS5 specs and Xbox Series X specs if you happen to really feel so inclined, however I’ve summarised the headline specs that concern us PC folks within the desk under.

As you’ll be able to appear, each GPUs can have 16GB of GDDR6 reminiscence at their disposal, which is kind of the step up from the 8GB at present put in on their RX 5700 and RX 5700 XT desktop GPUs. Heck, even Nvidia’s RTX 2080 and RTX 2080 Super playing cards solely have 8GB of GDDR6 reminiscence to their identify, whereas the RTX 2080 Ti has 11GB of the stuff.

Indeed, the one client PC graphics card to have ever had 16GB of reminiscence earlier than is AMD’s Radeon 7, though that had 16GB of HBM2 (2nd Gen High Bandwidth Memory) moderately than GDDR6. Of course, simply because a graphics card has oodles of reminiscence doesn’t imply it’s essentially extra highly effective than people who don’t. The Radeon 7, for instance, wasn’t actually that rather more highly effective than Nvidia’s RTX 2080 once I reviewed it final 12 months, and its 4K efficiency has since been overtaken by the RTX 2080 Super as nicely.

However, the truth that each of the next-gen consoles can have 16GB of reminiscence implies that AMD might be holding quick to this explicit quantity of their Big Navi line-up, and I wouldn’t be shocked if we noticed a number of 4K desktop GPUs with the identical quantity of reminiscence. After all, the truth that each consoles’ GPUs have a special variety of compute models (or CUs) means that these aren’t simply two barely totally different variations of the identical factor getting used right here, however two fully several types of graphics elements.

Here a flop, there a flop, in every single place a FLOP FLOP.

The PS5 GPU is especially intriguing, although, because it truly has a decrease variety of compute models than AMD’s RX 5700 XT. This 1440p-oriented graphics card has 40 compute models, 8GB of GDDR6 reminiscence and a clock velocity of 1650MHz, giving it a reminiscence bandwidth of 448GB/s – the identical because the PS5’s GPU. However, the clock velocity on the PS5 GPU is admittedly fairly nippy. It’s means sooner than the one contained in the Xbox Series X, and it’s significantly faster than the RX 5700 XT as nicely. This, in response to Cerny, truly leads to larger efficiency than a GPU with extra compute models, however a decrease clock velocity.

In the presentation, for instance, he gave the theoretical instance of a GPU with 36 CUs clocked at 1000MHz and a GPU with 48 CUs clocked at 750MHz. Both, he stated, would end in 4.6 Teraflops of energy as a result of means Teraflops are calculated (which is analogous to what the PS4 Pro is able to attaining), however that the 36 CU / 1000MHz GPU could be the one which gave higher efficiency, as a result of different GPU elements are inclined to run sooner when the GPU frequency or clock velocity is larger – offered you’ll be able to deal with the ensuing energy and warmth points related to it, that’s.

Cerny admits that the PlayStation hasn’t all the time excelled on this explicit space – therefore that bizarre weapon graph I used proper on the prime of this text – but it surely appears to be like just like the PS5 have been particularly designed to handle these explicit issues. Instead of working at a continuing, mounted clock velocity frequency, for instance (just like the Xbox Series X), and letting that dictate how a lot energy the console attracts, Cerny says the PS5 will do the alternative, working at just about full energy with a variable clock velocity. This ought to ship a extra constant stage of efficiency throughout the board, in response to Cerny, which is necessary when every console wants to seem precisely the identical as its next-door neighbour.

Teraflops should not an correct indicator of total efficiency, says Cerny, which he would say when the PS5’s T-floppage is decrease than the Xbox Series X…

Cerny didn’t go into element concerning the PS5’s cooling answer, admittedly – they’re saving that for a later date – however these particulars do paint an attention-grabbing image of how AMD’s Big Navi desktop equivalents may perform. After all, he did say early on within the GPU part of his presentation that, “If you see a similar discrete GPU available as a PC card at roughly the same time as we release our console, that means our collaboration with AMD succeeded in producing technology useful in both worlds.” He did then go on to make clear that, “It doesn’t mean that we at Sony simply put the PC part into our console,” but it surely does nonetheless recommend that there might be loads of similarities between them beneath the hood.

So what can we draw from this? It’s doable, as an example, that we’ll see some AMD Big Navi GPUs with a decrease variety of compute models however tremendous excessive clock speeds (most likely with some pretty hefty cooling options), or possibly they’ll observe basically AMD custom of being huge energy hogs however with the brand new, added enchancment of working like a whisper. Or each! As all the time, it’s foolish speculating an excessive amount of till AMD announce one thing concrete themselves, however given what we already learn about AMD’s Big Navi GPUs and their aim of making “uncompromising 4K gaming” efficiency, this could be a really early glimpse of how they’re hoping to realize that in comparison with what Nvidia at present provide. Either means, the race for the final word bestest best graphics card appears to be like prefer it may change into correctly fierce by the tip of the 12 months, and I might be ready on tenterhooks to see the way it all performs out.