Switch 2 Speculation

Console API are rather close to the metal and changing architecture often means you need emulation which is not good for power consumption

If the Switch 2 emulates the 1 with the same battery draw as playing natively that's kind of ok?

(I don't known if that's where the performance falls)
 
Let's be real, nothing has been confirmed. If it's the old Tegra there's no way it's hitting anywhere close to XB/PS levels
You're right. But we have multiple sources corroborating the fact that demos with DLSS enabled were shown during gamescom. At this point, it's as close as official as it can get without an offical statement that the thing is running a Tegra chip offering real good perfs.
I'm not sure what the old Tegra is. The actual rumored SoC is the Tegra 239, which is not old, and which should be able to run the Matrix demo just fine. Not PS5 level of fidelty, but close enough for sure.
Yes, there is still a slight possibility this one got canned and replaced with smth else, but that's unlikely given we have no leak at all regarding this supposed better SoC.

Switch 1 came out in 2017 on TSMC 20nm, which was a slightly older node nobody else really wanted to use and was hardly that great even for its time. But it was probably cheap, had the manufacturing capacity available, and was 'good enough'.

I dont think Samsung 8nm is out of the question for the same reasons. Especially if we start walking back any ideas that it's gonna be some super powerful platform, which I just dont see. That's just not what Nintendo does anymore and not only will they have a very 'affordable' price target in mind, but they also dont seem to like taking losses on hardware at any point like others will accept.
Samsung 8nm is not out of the question because we don't have the chip to check... but same here, it's mostly ruled out as T239 would be too large/hot/power hungry to get good performances out of it. And we now know that the chip offers very good performances, so we can assume it's not clocked super low as we feared in the first place.
I'm still expecting "low" clocks, but not Switch 1 levels low clocks given the perfs seem to be there.
Anyway, the leaks are now mostly coming every day, we should get a clear picture soon :)

Nintendo used Tegra X1 because it was the most advanced graphics IP in the mobile space at the time. And it was more advanced than GCN 1.1 in the PS4 and Xbox. It was only cheap, because 20nm was the last process without FinFet.
Honestly, my take on "why did Nintendo used the X1" would be smth like: because they dropped the ball, they had no faith in this project and wanted something cheap and already available. Maybe it was the most advanced thing for a mobile device at the time, but with more confidence and more time, Nintendo would have paid for a customized chip, with a better node process, more modern CPUs, and no useless transistors on it.
That's what T239 is actually, a customized chip dedicated to a hybrid console.
 
Honestly, my take on "why did Nintendo used the X1" would be smth like: because they dropped the ball, they had no faith in this project and wanted something cheap and already available. Maybe it was the most advanced thing for a mobile device at the time, but with more confidence and more time, Nintendo would have paid for a customized chip, with a better node process, more modern CPUs, and no useless transistors on it.

Nintendo and Nvidia both dropped the ball with the WiiU and mobile markets respectively. There's an absolute element of right partner at time.

If we get a nice good Nvidia gaming SoC as part 2 of this happenstance, it'll quite cool.

(Although part 2 is arguably Valve going 'what a good idea!')
 
Seems to be using Samsung 8nm. The is a base and a pro variant but seems to be cancelled.. thing looks siper powerful
Disappointing, but I guess that's expected as it's likely very cheap. Samsung's 8 nm process is a terrible fit for such a power constrained device, as it's inferior in power efficiency compared to TSMC's 7 nm process used in the current gen consoles. They should have used 5 nm. This will run hot and eat battery life in no time.
 
Last edited:
Disappointing, but I guess that's expected as it's likely very cheap. Samsung's 8 nm process is a terrible fit for such a power constrained device, as it's inferior in power efficiency compared to TSMC's 7 nm process used in the current gen consoles. They should have used 5 nm. This will run hot and eat battery life in no time.
The difference is ~10% in efficiency between TSMC 7nm DUV und Samsung's 8nm process.
 
The difference is ~10% in efficiency between TSMC 7nm DUV und Samsung's 8nm process.
If we compare Snapdragon 865(TSMC 7nm) with Exynos 2100(Samsung 5nm), the SD865 is actually more efficient, and the 2100 only beats it in peak performance through much higher power draw.


Obviously there's different CPU designs at play here, but it does seem a stark enough example to suggest that Samsung 8nm, a full generation+ behind Samsung 5nm, is probably not as close to TSMC 7nm as you're suggesting.

On the last page, I was mentioning how I dont think Samsung 8nm is out of the question, but I cant take any victory lap here as it's still somewhat disappointing news if true. Not the end of the world, but obviously most people will have hoped for something better.
 
Ha, same source as mine above posted just a couple minutes before you. Didn't think many people knew about Geekerwan. He does good stuff, though only limited amounts in English.
I'm surprised someone else used them as a source but they appear to be the only ones to go so in-depth with power efficiency.

I originally posted those exact same images back in march in this very thread :p
 
If we compare Snapdragon 865(TSMC 7nm) with Exynos 2100(Samsung 5nm), the SD865 is actually more efficient, and the 2100 only beats it in peak performance through much higher power draw.


Obviously there's different CPU designs at play here, but it does seem a stark enough example to suggest that Samsung 8nm, a full generation+ behind Samsung 5nm, is probably not as close to TSMC 7nm as you're suggesting.

On the last page, I was mentioning how I dont think Samsung 8nm is out of the question, but I cant take any victory lap here as it's still somewhat disappointing news if true. Not the end of the world, but obviously most people will have hoped for something better.
We had this discussion in this forum with Ampere and RDNA2 three years ago. Qualcomm ported SD845 to TSMC's 7nm DUV process while Samsung used their 8nm for S10. Benchmarks from the S20 review from Anandtech: https://www.anandtech.com/show/1560...xynos-snapdragon-review-megalomania-devices/9

Starting with 7nm EUV TSMC has run away from everyone. But when it comes to cheap processes Samsung's 8nm is not as bad in comparision to 7nm DUV.
 
Seems to be using Samsung 8nm. The is a base and a pro variant but seems to be cancelled.. thing looks siper powerful
garbage reporting. the T234 isn't canceled, it's literally the Jetson Orin and the Orin AGX in vehicles. it has little to do with the Switch 2's chip other than sharing the same CPU and GPU architecture. a lot of the github code for the T239 is copy and pasted from the T234 because of that
 
Back
Top