“As AMD assures us, the shader pre-fetching code is experimental and thus not expected to work.”
AMD’s new RDNA 3 graphics cards have sparked some debate since their release, and the company has directly addressed one of the rumours.
According to hardware leaker Kepler L2 (opens in new tab) on Twitter, Team Red released Navi 31 with non-functional shader pre-fetch hardware (on A0 silicon – we’ll come back to what that means) (among others).
“Like previous hardware generations, shader pre-fetching is supported on RDNA 3 as per [gitlab link (opens in new tab),” AMD told Tom’s Hardware (opens in new tab). The code in question [highlighted by Kepler L2] controls an experimental function that was never intended to be included in these products and will not be enabled in this generation.
“Inclusion of experimental features to enable exploration and tuning for deployment in a future product generation is a common industry practise.”
In other words, nothing is broken here, as suggested by the hardware leaker, but the relevant code is for an experimental feature that isn’t supposed to be enabled yet (but will be used down the line with a future generation of Radeon graphics cards).
First time right – but what about the graphics driver?
In short, AMD is saying that there is nothing to see here and that you should move along.
This controversy has also been marred by the suggestion that there is something wrong with the (alleged) A0 silicon used in these initial RDNA 3 boards, with ‘A0’ referring to the silicon’s initial physical version from the fab. A1 would be the next (minor) revision (to address issues discovered during the first outing), followed by A2, and so on (with a switch to B for bigger changes, and then C, etcetera – more versions may appear as time rolls on, and honing or additional fixing is applied).
There are a few things to consider here. To begin, AMD has yet to confirm that the RDNA 3 cards use the A0 version (this comes from industry sources and chatter on the grapevine; so we must be careful there). And the second point is that the A0 silicon can – and is – used by chip makers with finished products; this simply means that the silicon works properly and does not require further re-spin, which is quite impressive (the engineers got it right first time, in other words).
Of course, you could argue that AMD wanted to get the mentioned shader pre-fetching feature working in RDNA 3, but it didn’t make the cut, and so Team Red is now calling it experimental, and some people have indeed asserted this – but obviously, we don’t know what happened behind the scenes, and this is pure speculation.
Aside from this controversy, there is still the issue of apparently wonky clock speeds with large variations on the RX 7900 XTX and 7900 XT, as some other people on Twitter have pointed out (and reviewers),
According to those reports, clock speeds can range from 2.4GHz to 2.9GHz depending on the game.
This could be due to increased power consumption and slower clocks, but some observations have been made about similar power draw between these much higher and lower speeds (oddly). At this early stage of the game, it’s difficult to tell what’s going on. AMD has stated that it is investigating the issue with the assistance of the user who discovered it.
In any case, game benchmarks show that the 7900 XTX outperforms its rival, Nvidia’s RTX 4080, in general – and driver updates may be able to smooth out larger clock frequency variations, possibly. Concerns about power usage and boost speeds remain, but hopefully anything wonky is fixable enough on the software front.
Indeed, there’s some speculation on the grapevine about AMD rushing out RDNA 3 to meet its long-held end-of-year launch deadline, in order to prevent Nvidia from dominating holiday sales and keep its investors happy. As a result, polishing the Adrenalin drivers could add a significant amount of performance in the near future.
Of course, releasing a product without all of the drivers on board isn’t a new situation, but it’s also far from ideal. That said, we believe the good news for RDNA 3 buyers is that their cards will become significantly faster in the next month or two. This could lead to the 7900 XTX outperforming the RTX 4080 and even rivalling the 4090 (for rasterization, or non-ray tracing performance); however, we’re getting ahead of ourselves.