Nintendo Switch 2 rumored to outperform Steam Deck but lag Xbox Series S with 3.1 TFLOPS

[Redaktion]

The Nintendo Switch 2 could be officially unveiled in just a few days time. New rumors now suggest that the hybrid console is significantly slower than an Xbox Series S, even when docked. In handheld mode, the gaming system could perform similarly to the Steam Deck.

https://www.notebookcheck.net/Nintendo-Switch-2-rumored-to-outperform-Steam-Deck-but-lag-Xbox-Series-S-with-3-1-TFLOPS.947160.0.html

[Wrath7007]

Yes it's a bit less in teraflops. But the switch 2 has something that the xbox  series s doesn't. Nvidia Dlss. That alone can make something that does 4k 20 frames a second ,automatically do 4k 120fps

[Mr. DLSS]

You realize that DLSS has a cpu overhead cost right? The switch 2 is already going be running a power starved processor. Your 4k 120 is a pipe dream. 4k 30 is reasonable. Plus even if you could take enough load of the gpu through upscaling their is zero chance that the CPU processing side of the chip could do 120hz in anything modern

[GeorgeS]

ROTFLMAO!!!

Disclaimer: I have a SwitchOLED, Ally & countless other gaming devices

It is rather comical the fanboi swooning over rumors and conjecture on the Switch2!!

While YES there might be a specification or two that are better than a lowly system like a 'Steam Deck' HOWEVER realistically the sum total of whatever the entire Switch2 becomes will likely only outperform the SwitchOLED and many other gaming systems based on OBS or otherwise antique technology.

[Swizzy]

More "leaks" lmao

[Mario Kassar]

if it runs Dark Souls 3 in 33 fps, I will buy it

[Jordon]

After converting from ampere TFLOPS in the switch 2 to RDNA 2 TFLOPS in the steam deck, the switch 2 will barely outperform the steam deck when its docked. When its in handheld mode it will be 40-45% slower than the steam deck

[X2Jason]

Yes it's a bit less in teraflops. But the switch 2 has something that the xbox  series s doesn't. Nvidia Dlss. That alone can make something that does 4k 20 frames a second ,automatically do 4k 120fps

This is inaccurate. Firstly it wikl be a very rudimentary version of DLSS, focusing on just the upscaling part, and not even at DLSS 3 or 3 levels of accuracy one would imagine.

Frame generation is almost certainly a pipe dream.

Secondly, and most importantly, upscaling technology has diminishing returns when coming from a low performance base. 4K20 can not magically turn into 4K120 as you suggest. To go from 4K120 to 4K200 is more likely.

I would suspect that the console will start target a 1080p base resolution for a AAA title, with 4K supported for the system dashbkard when docked, and for the occasional lower end game, or titles like Mario etc where the low texture details are masked by vibrant colours and a clean design, capable of hitting 4K when upscaled.

[Dinklebot]

Yes it's a bit less in teraflops. But the switch 2 has something that the xbox  series s doesn't. Nvidia Dlss. That alone can make something that does 4k 20 frames a second ,automatically do 4k 120fps

This is inaccurate. Firstly it wikl be a very rudimentary version of DLSS, focusing on just the upscaling part, and not even at DLSS 3 or 3 levels of accuracy one would imagine.

Frame generation is almost certainly a pipe dream.

Secondly, and most importantly, upscaling technology has diminishing returns when coming from a low performance base. 4K20 can not magically turn into 4K120 as you suggest. To go from 4K120 to 4K200 is more likely.

I would suspect that the console will start target a 1080p base resolution for a AAA title, with 4K supported for the system dashbkard when docked, and for the occasional lower end game, or titles like Mario etc where the low texture details are masked by vibrant colours and a clean design, capable of hitting 4K when upscaled.

You don't know your hole from an @ $$ in the ground.

[Kenan]

The expected performance of the Nintendo Switch 2 easily reaches the performance of the Playstation 4 Pro, which delivers almost 4.2 TFLOPS! Switch 2 can also easily surpass the performance of the Playstation 4 Pro!

1. -- Expected GPU of the handheld Switch 2: Nvidia T239 Ampere , at 1007.25 MHz (clock is adjustable with software) = 3.09 TFLOPS performance in FP32.
-- GPU of another handheld, Rog Ally: AMD 780M (in Z1 Extreme APU), at 2700 MHz (clock specified as Boost Clock, also adjustable) = approx. 4.1 TFLOPS performance in FP32.

2 – An example of change in performance as TFLOP with two graphics cards, both with the same graphics processor of the other, both as GTX 1080, with different boost clocks:
-- ASUS ROG STRIX GTX 1080 GAMING: 8.873 TFLOPS in FP32 at Boost Clock 1733 MHz.
-- Colorful iGame GTX 1080 Fire Ares X-TOP AD LE: 9.784 TFLOPS in FP32 at boost clock 1911 MHz.

In the example with the grapics cards with the GTX 1080 by increasing the boost clock by "only" 178 MHz, the performance of the GTX 1080 was increased by 0.911 Tflops!

Nintendo had the clock speed of the Switch 1 increased later, after the release, for more performance gain!

3. What many people don't seem to know: If two identical graphics cards (GPUs) are placed on two different main boards, on which ram memory and processors (CPUs) are also located, whereby the distances between the first board on a stationary console are much greater than on the second board as on a handheld, then the following happens: The handheld's system performs much better overall because the information has to travel much shorter distances between the CPU and GPU and the storage medium! Therefore, a smaller handheld objectively performs significantly better than a larger console, even if the GPUs and CPUs would have the same performance individually.

-- In short: The Switch 2 can "easily" achieve the performance of Rog Ally, which uses the Z1 Extreme chip, by increasing the boost clock by a few hundred MHz, as well as the performance of the Playstation 4 Pro, which is specified with almost 4.2 TFLOPS.