Digital Foundry Article Technical Discussion [2024]


Interesting discussion so far. I'm about 55 minutes in and Richard says something along the lines of "the backlash against ray traycing continues but then you listen to the Mark Cerny talk...... basically the people that matter, the architects of the hardware of the future realizing that rasterization is kind of reaching its limits......"

While I understand the sentiment of what he's trying to say, it'd be best if he phrased it better. If your company or business exists to sell products and services, the people that matter are the potential buyers of said products/services. A business, product or service without a buyer does not matter at all. It's completely irrelevant.

Yep it’s the buyers that matter in the end. I haven’t watched the video yet but I assume Richard meant that someone like Cerny matters a whole lot more than some random person on the internet. He has a lot more influence on those buyers and directly decides what products are available for them to buy.
 
don't think AMD engineers are less capable, they know what they would have to do, that's more up to AMD's funding, so Maybe Sony is helping financially.
Why would you assume they’re as capable? There’s a point where the big money gets the best people and if those best people spend years and years collaborating under competent leadership, they straight-up become better than the competition.
 
Why would you assume they’re as capable? There’s a point where the big money gets the best people and if those best people spend years and years collaborating under competent leadership, they straight-up become better than the competition.
By this logic no one stands a chance against nvidia, and if anyone does it must have to be another trillion dollar company like Apple or whatever.
 
By this logic no one stands a chance against nvidia, and if anyone does it must have to be another trillion dollar company like Apple or whatever.
That may be true though. Right now, Nvidia has the money and resources to stay ahead of the competition for quite a while. AMD's bread-winner is their CPUs. I've recently upgraded my computer and will always get an AMD-CPU and Nvidia-GPU. I don't think AMD is going to compete with Nvidia on the high end anymore. That doesn't bode well for consoles.
 
That may be true though. Right now, Nvidia has the money and resources to stay ahead of the competition for quite a while. AMD's bread-winner is their CPUs. I've recently upgraded my computer and will always get an AMD-CPU and Nvidia-GPU. I don't think AMD is going to compete with Nvidia on the high end anymore. That doesn't bode well for consoles.
As long as this guy makes the next Xbox, we'll be happy.


He and his team got the most out of the One X and Series X consoles at those prices. Apparently he has an even freer hand now.
 
As long as this guy makes the next Xbox, we'll be happy.


He and his team got the most out of the One X and Series X consoles at those prices. Apparently he has an even freer hand now.
Most out of the one x and series x while they were competing with Sony in the console space. I think the days of Microsoft using the console as a loss leader are over. The series x lost them a lot of money and what did they get out of it? If we listen to Tom Warren they are going a completely different direction after another lackluster generation of selling consoles. Best bang for your buck consoles are done with Microsoft.
 
By this logic no one stands a chance against nvidia, and if anyone does it must have to be another trillion dollar company like Apple or whatever.
Not at all because others can also become better or new talented guys with no previous experience who will make a difference might also join AMD or Intel, but this will take a little while. At the moment, NVIDIA's engineers are clearly a cut above the rest. It's certainly not due to luck, so I'm not sure why you question this when the results speak for themselves. If Team A constantly wins championships, no one questions their superiority over the other teams. On the flip side, it's also foolish to assume this dominance will never be challenged.
 
That may be true though. Right now, Nvidia has the money and resources to stay ahead of the competition for quite a while. AMD's bread-winner is their CPUs. I've recently upgraded my computer and will always get an AMD-CPU and Nvidia-GPU. I don't think AMD is going to compete with Nvidia on the high end anymore. That doesn't bode well for consoles.
Yeh I wasn't saying it was not right, and I was thinking the bolded looking at things from the point and tried to step around the platform wars, but I may have not understood what company the poster was meaning and trying to be vague has done me no favours, I thought he was saying Sony has better hardware engineers in the gpu/cpu space than AMD.

Not at all because others can also become better or new talented guys with no previous experience who will make a difference might also join AMD or Intel, but this will take a little while. At the moment, NVIDIA's engineers are clearly a cut above the rest. It's certainly not due to luck, so I'm not sure why you question this when the results speak for themselves. If Team A constantly wins championships, no one questions their superiority over the other teams. On the flip side, it's also foolish to assume this dominance will never be challenged.
I thought your post after the posts above it was saying Sony has better cpu/gpu engineers than AMD, although the bold somewhat contradicts the point about big money buying all the top talent. Seems your riding both sides of that argument because that is the logical counter to your first point.
 
Yeh I wasn't saying it was not right, and I was thinking the bolded looking at things from the point and tried to step around the platform wars, but I may have not understood what company the poster was meaning and trying to be vague has done me no favours, I thought he was saying Sony has better hardware engineers in the gpu/cpu space than AMD.


I thought your post after the posts above it was saying Sony has better cpu/gpu engineers than AMD, although the bold somewhat contradicts the point about big money buying all the top talent. Seems your riding both sides of that argument because that is the logical counter to your first point.
No, I meant NVIDIA has better engineers than Sony and AMD.
 
That may be true though. Right now, Nvidia has the money and resources to stay ahead of the competition for quite a while. AMD's bread-winner is their CPUs. I've recently upgraded my computer and will always get an AMD-CPU and Nvidia-GPU. I don't think AMD is going to compete with Nvidia on the high end anymore. That doesn't bode well for consoles.
Nvidia is quite easily surpassable. We're almost at the great equalizer which is the rapid slow down in performance gains via node shrinks. We have like 2 left after 3nm and then it'll be stagnation for a while. That will give competitors a long time to catch up. As for the effect on consoles, I don't think it matters too much at all. Consoles never use the high-end. Amd just needs to meet the performance targets set by Sony and Microsoft. If they do that, the lack of high end doesn't matter.

Amd's problem is that they're kind of like Spirit airlines but, they're trying to charge prices like Lufthansa/Delta(Nvidia). There's nothing wrong with being spirit airlines but they need to evolve their business model to be more like Ryanair. Learn how to be profitable by offering cheap products at large volumes. If they do this, they can be wildly profitable. I don't know if it's an ego thing with them or what, but their path to success in the gpu space is so clear. They need to aggressively press Nvidia at the prices points with the most volume and offer significantly more value. Nvidia likes 60% margins but Amd needs to learn to be happy with 20% or less. Let Nvidia do the R&D, then just copy their philosophy without infringing on the patents. At this stage, Nvidia cannot afford to get into a pricing war with other gpu vendors because their stock will collapse. Since they already have a majority of the market, there's nothing to be gained through lower priced volume sales other than lower annual revenue.
 
AMD can easily compete and take market share from Nvidia if they focused on it.

The average gamer doesn't give a shit about the high-end GPU's and technology we talk about in here.

If AMD focused on making a GPU with the performance level of the 7900XTX but at half the price it would give Nvidia a right shafting.

That wouldn't be possible on every process node but it would be a good target to aim for and would generate more sales than any Nvidia GPU.

Not to mention such performance at that price would do wonders for PC gaming.

They did it with the HD4870, HD5870 and HD7970.

So they can do it......
 
Nvidia is quite easily surpassable. We're almost at the great equalizer which is the rapid slow down in performance gains via node shrinks. We have like 2 left after 3nm and then it'll be stagnation for a while. That will give competitors a long time to catch up. As for the effect on consoles, I don't think it matters too much at all. Consoles never use the high-end. Amd just needs to meet the performance targets set by Sony and Microsoft. If they do that, the lack of high end doesn't matter.

Amd's problem is that they're kind of like Spirit airlines but, they're trying to charge prices like Lufthansa/Delta(Nvidia). There's nothing wrong with being spirit airlines but they need to evolve their business model to be more like Ryanair. Learn how to be profitable by offering cheap products at large volumes. If they do this, they can be wildly profitable. I don't know if it's an ego thing with them or what, but their path to success in the gpu space is so clear. They need to aggressively press Nvidia at the prices points with the most volume and offer significantly more value. Nvidia likes 60% margins but Amd needs to learn to be happy with 20% or less. Let Nvidia do the R&D, then just copy their philosophy without infringing on the patents. At this stage, Nvidia cannot afford to get into a pricing war with other gpu vendors because their stock will collapse. Since they already have a majority of the market, there's nothing to be gained through lower priced volume sales other than lower annual revenue.
NVIDIA realized that node process shrinkage would come to an end long ago, hence the tensor and RT cores.
They divergered from raster and AMD has not been able to adapt yet.
The way I see it NVIDIA set the tone, AMD tries to catchup, but before they catch up NVIDIA changes the narrative. (perhaps "neural rendering" (new hardware blocks?) is this generations change)
With a 10% (possible soon sub 10%) marketshare, AMD has limited R&D funding to thow at catching up.

If it was easy, AMD would have done something different than seceeding the high end.
 
NVIDIA realized that node process shrinkage would come to an end long ago, hence the tensor and RT cores.
They divergered from raster and AMD has not been able to adapt yet.
The way I see it NVIDIA set the tone, AMD tries to catchup, but before they catch up NVIDIA changes the narrative. (perhaps "neural rendering" (new hardware blocks?) is this generations change)
With a 10% (possible soon sub 10%) marketshare, AMD has limited R&D funding to thow at catching up.

If it was easy, AMD would have done something different than seceeding the high end.
It doesn't matter if Nvidia sets the tone or not. AMD could easily add rt cores if they wanted but, they insist its not efficient use of die space. Even if Nvidia drops new technology, adoption is very slow. It's usually a time lag of a generation or two or 3 till we see increased adoption. You need enough of the userbase to have the hardware before you see rapid adoption in games. RT showed up in Turing and only in rtx 4000 are we seeing increased adoption of rt. Even then, very few devs cannot go all in because they risk alienating a segment of the market. Even DLSS that everyone raves about is supported in 600ish games since it's release during the turing generation. Guess what? 19k games were released on steam this year alone. If we count all the games released since 2018, DLSS adoption is probably sub 1% and DLSS is adopted at a faster rate than RT.

All of this is to say that the tone Nvidia sets is not insurmountable at all. At the end of the day, all that matters is performance per dollar. If AMD released a gpu with rtx 4090 performance in raster and rtx 3080 performance in rt for $599, most people are not buying rtx 5000. Amd need to release products that make it hard to justify the purchase of Nvidia gpus at any price point under $599. If they can do that, they'll be extremely successful.
 
It doesn't matter if Nvidia sets the tone or not. AMD could easily add rt cores if they wanted but, they insist its not efficient use of die space. Even if Nvidia drops new technology, adoption is very slow. It's usually a time lag of a generation or two or 3 till we see increased adoption. You need enough of the userbase to have the hardware before you see rapid adoption in games. RT showed up in Turing and only in rtx 4000 are we seeing increased adoption of rt. Even then, very few devs cannot go all in because they risk alienating a segment of the market. Even DLSS that everyone raves about is supported in 600ish games since it's release during the turing generation. Guess what? 19k games were released on steam this year alone. If we count all the games released since 2018, DLSS adoption is probably sub 1% and DLSS is adopted at a faster rate than RT.

All of this is to say that the tone Nvidia sets is not insurmountable at all. At the end of the day, all that matters is performance per dollar. If AMD released a gpu with rtx 4090 performance in raster and rtx 3080 performance in rt for $499, most people are not buying rtx 5000.
Again if it was easy, AMD would have done it 🤷‍♂️
And marketshare (NVIDIA at ~90%) tells that whatever they did, they did it right.
 
Back
Top