AMD's new RDNA 3 graphics cards have been the subject of controversy following their release, with the company directly addressing a point of criticism raised by the rumor.
Namely, that Team Red released Navi 31 with non-functional shader prefetch hardware (on A0 silicon; we'll get back to what that means), a theory advanced by hardware leaker Kepler_L2 (opens in a new tab) on Twitter (among others).
AMD told Tom's Hardware (opens in a new tab): "Like previous generations of hardware, shader prefetching supports RDNA 3 in accordance with (*3*). The code in question controls a Experimental feature that has not been selected for inclusion in these products and will not be enabled in this product build.
"It is standard industry practice to include experimental features to allow for exploration and tuning for deployment in a future generation of products."
In other words, nothing is broken here as the hardware leaker suggested, but the relevant code is for an experimental feature that shouldn't be enabled yet (but will be used later with a future generation of Radeon graphics cards).
Analysis: Get it right the first time, but what about the graphics driver?
In short, what AMD says is that there is nothing to see here, go your way.
This controversy has also been wrapped up in the suggestion that there is something wrong with the (alleged) A0 silicon used in these early RDNA 3 boards, with "A0" meaning it is the initial factory silicon of the physical version. A1 would be the next (minor) revision (to fix problems found in the first version), then A2, and so on (with a move to B for bigger changes, then C, etc. - other versions may appear over time and lapping or fixing is applied).
There are a few points to keep in mind here. First of all, that AMD has not confirmed that RDNA 3 cards use version A0 (this comes from industry sources and live conversations, so we have to be careful there). And the second point is that A0 silicon can be, and is, used by chipmakers with finished products, that just means the silicon works fine and doesn't need to be re-rotated, which did an impressive thing (the engineers got it right). the first time, in other words).
Sure, one could argue that AMD might have wanted the aforementioned shader prefetch feature to work in RDNA 3, but they didn't quite make the cut, which is why Team Red is now calling it experimental, and some people have claimed so, but obviously We can't tell what happened behind the scenes, and that's pure speculation.
That controversy aside, there's still the issue of seemingly unstable clock speeds with huge discrepancies on the RX 7900 XTX and 7900 XT, as other people on Twitter (and reviewers), like us, have pointed out. We discussed it recently. According to these reports, clocks can vary from something like 2,4 GHz to 2,9 GHz from game to game.
This could be related to increased power consumption and decreased clocks as a result, but some observations of similar power consumption have been made between these much higher and (strangely) lower speeds. It's hard to judge what's going on here, so at this early stage in the game, AMD said it's looking into the issue with the help of the user who originally discovered it.
And in any case, generally speaking, gaming benchmarks point to solid performance for the 7900 XTX compared to its rival, Nvidia's RTX 4080, and driver fixes could smooth out frequency variations. of the most important watch, perhaps. Still, concerns about power consumption and boost speeds are still a concern, but hopefully anything that isn't right is software fixable enough.
In fact, there's more than small talk about how AMD could have sped up RDNA 3 to meet its long-running year-end release schedule, to prevent Nvidia from dominating holiday sales and to keep its investors happy. So it's quite possible that polishing the Adrenalin drivers could add a considerable amount of performance in the (quite a lot) near future.
Releasing a product without fully working drivers isn't a new situation, of course, but it's not ideal either; away from. That said, we're guessing the (kind of) good news for RDNA 3 buyers is that their cards could get faster in the next couple of months, and maybe even considerably faster. This could lead to a situation where the 7900 XTX actually outperforms the RTX 4080 and competes more effectively with the 4090 (for rasterization, which means performance without ray tracing); but we are ahead here.
Today's best graphics card deals