Switch 2 Speculation

The rumored T239 chip is based on the Orin SoC that is manufactured on the 8nm process, correct? Hypothetically speaking, what would be the R&D costs to adapt the design to a different process?
Generally, each new generation of process technologies will feature incompatible design rules with the prior generations and design rules between different foundries with comparable technologies are likely incompatible as well. In IC design, you would only be able to reuse the RTL code between the different physical design implementations across the targeted set of process technologies ...
 
The rumored T239 chip is based on the Orin SoC that is manufactured on the 8nm process, correct? Hypothetically speaking, what would be the R&D costs to adapt the design to a different process?
Not insignificant. Though if they worked with Nvidia and Samsung, there's no telling what kind of business deal(s) they could drum up to make it happen anyways. I think sales expectations for Switch 2 will be pretty high, so Nvidia and Samsung might be more inclined to make a good deal on a port to a better Samsung process.

Especially if there's any expectation of doing this process port eventually anyways, like they did with Tegra X1 on the existing Switch, mid-life.

That said, Nintendo is notoriously cheap.
 
Especially if there's any expectation of doing this process port eventually anyways, like they did with Tegra X1 on the existing Switch, mid-life.
That was due to a hardware design flaw's boot ROM exploit. It was mostly a coincidental decision that the modified Tegra X1 design was implemented on a new process technology ...
 
That was due to a hardware design flaw's boot ROM exploit. It was mostly a coincidental decision that the modified Tegra X1 design was implemented on a new process technology ...
I doubt that was the sole reason for it. They could have certainly just done a modified design on the same process, but decided to go for an upgrade anyways. And it's not like process ports aren't entirely normal and common in the console world, especially if they're gonna be around like 7+ years. Granted, the economics of it aren't as appealing as they used to be, but we're talking Samsung here, not TSMC, so it may still make sense.
 
I doubt that was the sole reason for it. They could have certainly just done a modified design on the same process, but decided to go for an upgrade anyways. And it's not like process ports aren't entirely normal and common in the console world, especially if they're gonna be around like 7+ years. Granted, the economics of it aren't as appealing as they used to be, but we're talking Samsung here, not TSMC, so it may still make sense.
I think an unfixable hardware problem which easily lead to major circumvention of anti-piracy measures was a strong reason enough alone to change the design either way since no software-based patching solution could be employed to protect their business. The console vendor in question did not take advantage of the property of a new process in the form of a mid-life performance upgraded SKU like they did with prior portable systems being another indicator of their original intent ...
 
The rumored T239 chip is based on the Orin SoC that is manufactured on the 8nm process, correct? Hypothetically speaking, what would be the R&D costs to adapt the design to a different process?
Just keep your expectations in check .This is a Nintendo console, not the latest iPhone. Switch 2 will most likely use 8nm process and target PS4 graphics with more ram to allow PS5 ports. This will be enough for it to be a success anyways as Nintendo consumers will be more than happy. Nintendo don't need to do more.
 
Apologies if this was discussed already but how does Switch 2's GPU compare to the GPU in the PS4 Pro?
Less than, pretty similar to Steam Deck.

I'm more interested in what the controller is, how much storage there is, etc. I'm going to buy one, tens of millions of people are, probably over a hundred million, and it'll cost <= $400 or less, because it's Nintendo of course it will. We'll get to see it sometime this month.
 
Ampere is much better than GCN. At the same theoretical number level Ampere delivers better performance. To get to PS4 pro performance the Switch should have around ~2,5 TFLOPs.
 
Really? I know the Pro has way more shaders, but the switch 2 has a more modern and efficient architecture than the GCN that made up the Pro's GPU, no?

Yeah, RDNA2 and Ampere are pretty similar in a lot of respects outside RT, and their was an entire shipping manifest of parts for the Switch 2 someone found showing the ram is only a bit faster than Steamdecks. All else being equal then they'll both get bandwidth capped at something similar.

The good news is Steamdeck is solidly faster than a PS4, able to play Forbidden West without any of the no doubt extreme optimization done for PS4. So while not as fast as a PS4 Pro, still probably "good enough".
 
Really? I know the Pro has way more shaders, but the switch 2 has a more modern and efficient architecture than the GCN that made up the Pro's GPU, no?
The computational power isn't the only important parameter. The upgraded PS4 SKU will very much come out on top in factors such as fixed function unit and memory performance ...
 
Yeah, RDNA2 and Ampere are pretty similar in a lot of respects outside RT, and their was an entire shipping manifest of parts for the Switch 2 someone found showing the ram is only a bit faster than Steamdecks. All else being equal then they'll both get bandwidth capped at something similar.

The good news is Steamdeck is solidly faster than a PS4, able to play Forbidden West without any of the no doubt extreme optimization done for PS4. So while not as fast as a PS4 Pro, still probably "good enough".
Steam deck gets around the same framerates of a PS4, but at lower resolutions. For example god of war gets slightly better framerates than a PS4 at half the resolution.
 
Steam deck gets around the same framerates of a PS4, but at lower resolutions. For example god of war gets slightly better framerates than a PS4 at half the resolution.

That's why that "without the extreme optimization for the specific platform" part matters, a lot. At this point at the end of life there's an insane amount of platform specific optmization for huge first party titles like Forbidden West. That the Steam Deck can just run the same game through brute force provides a good estimate of how much faster it is.

Also, maybe no reveal this month after all :cry:
 
That's why that "without the extreme optimization for the specific platform" part matters, a lot. At this point at the end of life there's an insane amount of platform specific optmization for huge first party titles like Forbidden West. That the Steam Deck can just run the same game through brute force provides a good estimate of how much faster it is.

Also, maybe no reveal this month after all :cry:
The deck just would never reach the power of the PS4 GPU, period. It's too power limited, the bandwidth isn't close so even with the same optimization for both, the PS4 would still pull ahead.
 
I think an unfixable hardware problem which easily lead to major circumvention of anti-piracy measures was a strong reason enough alone to change the design either way since no software-based patching solution could be employed to protect their business. The console vendor in question did not take advantage of the property of a new process in the form of a mid-life performance upgraded SKU like they did with prior portable systems being another indicator of their original intent ...
Again, they could have done a simple design fix on the same process as before, but decided to do a more extensive port to a whole new node(which also went from planar to FinFET, correct me if I'm wrong).

And yes, they absolutely did take advantage of the new process by significantly beefing up battery life and efficiency. Same as most other consoles do when they make such a change. Mid gen Pro consoles like we had last gen were not the norm and were completely different chips entirely, so not relevant to what we're talking about.
 
Back
Top