The implied assumption in this comment is that "AMD CAN do all these things you want, but they're just not doing it" when that couldn't be further from the truth. I'm guilty of posting this ad nauseam, but it truly is THE defining factor in all of this: Nvidia has an R&D budget of over $5.27 billion. AMD has an R&D budget of just $2 Billion, and don't forget, where Nvidia can spend the overwhelming majority if their much larger budget purely on graphics, AMD has to split their between x86 and graphics. Due to the fact that x86 has the larger T.A.M. when compared to graphics, and Ryzen is much more lucrative to AMD than Radeon is, it's safe to assume that of that $2 Billion budget, less than half goes to graphics. So basically, it's a situation in which AMD's graphics division has to compete with Nvidia while Nvidia spends about 6x more on R&D. Based on that reality alone, I think it's actually a MIRACLE and rather impressive that AMD is able to compete at all, and even match or exceed Nvidia in Raster. Can you think of any other example from any other market or industry in which a duopoly exists and the underdog manages to still compete while having around a sixth of the budget? I think instead of complaining about what AMD does or doesn't do, we need to understand that financial reality of the situation and realize that when it comes to things like ray tracing, Nvidia had a generational headstart and basically unlimited funds when compared to AMD and use that realization to temper our expectations.
FYI: Intel's R&D budget is over $15 billion, and while Intel has numerous different markets and products in which to utilize those funds, I think it's safe to assume that Intel is spending a whole lot more than AMD on x86 R&D, probably by a factor approaching Nvidia's 6x advantage. It is therefore yet another extremely impressive feat that AMD has accomplished what it has in x86 as well while being at a severe financial disadvantage, Based on the money spent, Intel should be outperforming AMD by 40+%, but they're not, which testifies to the fact that AMD does a whole lot with very little.
I do agree with the above stuff you post. I am also posting similar replies to people expecting AMD to beat Intel and Nvidia in everything just so they don't have any excuse to try to attack AMD.
BUT. What I was saying is that they where shortsighted here, thinking that RT performance will remain secondary for at least the next 2 years. That they still have time, they can focus on raster for the next 2 years and still be fine. Well, RT performance isn't that much important in gaming yet and it might not be in the next 2 years, but it is in marketing. And that's where they are losing. 4070 Ti will sell better because of DLSS 3.0 and RT performance. AMD might offer something competitive with FSR 3.0, but they wouldn't be able to do so with RT. Even Hardware Unboxed that many Nvidia trolls call it AMD biased (they
are used demand in having tech press at 80%-20% favoring Nvidia over AMD, so anything close to 50%-50% they call it biased) was mentioning in their latest video that as time passes, RT performance will becoming more important, more games will be supporting it. And the RT performance is the main argument from tech press today to paint 7900XT/X as a NON recommended product. If AMD has focuses on RT performance, they could sell their top cards at at least RTX 4070 Ti prices. Now they will be having problems selling RX 7900XT/X even against RTX 4070 Ti.
So, because of their limited R&D, they have to be more careful about their targets. They thought of DDR5 and a long lasting socket and they failed in CPUs. They thought of chiplets and raster and they failed in GPUs. And they have also to change their business model. AMD tries to offer products that are future proof. Intel and Nvidia try to win current benchmarks, ignoring or even limiting the long term value of their products. AMD needs to do the same because consumers
do not appreciate this. If AM5 was a 1-2 generations socket with DDR5 AND DDR4 support, the platform could be (much) cheaper. If 7900XT/X where coming with lower VRAM capacity and narrower data bus, they could be cheaper to produce. And if they where not much better in raster than 6900X but twice as fast in RT, they could have a real chance in the market.
Intel is another kind of monster. They know they will sell even mediocre products, thanks to their strong ties with OEMs and the dependensy of OEMs to Intel products. As with their efforts to remain an alternative to ARM in tablets, by throwing billions, they will do the same with GPUs. They need GPUs for their future servers and they need GPUs to keep having a gaming platform in retail, even if Nvidia in the future decides to go big in ARM and slowly start abandoning X86 platform with a goal to create it's own ARM based gaming platform. Think Valve but with hardware.
PS AMD needs to focus on RT because if they don't they will lose the next generation of consoles. MS and SONY are not married to AMD.
