Switch 2: Nvidia processor brings huge increase in DLSS and performance

A larger aspect of nintendo switch 2 is your hardware upgrade compared to the previous switch. The combination processor comes again from Nvidia. First detail on technology Nintendo revealed on its product pageWith reception innovations such as variable refresh rates (VRR) for performance.

This means that the screen does not represent its maximum of 120 Hz from hardening, instead the GPU dynamically adjusts the fresh rate of the game into a second (FPS) frame. Practical for game developers: You will no longer adapt to 30, 60 or 120 FPS.

Nvidia speaks in her blog Meanwhile, ten -time graphic performance when jumping from the first switch on switch 2. Neither Nintendo nor Nvidia goes to the processor specifications. With the rumors of the last few months and years, however, a fairly accurate picture can be taken.

Big GPU, more memory

Switch 2 is probably a small version of car and robot processor Nvidia Jetson Oin AGX. Eight Cortex -A78C arm core switch -2 are for seating in the processor. The integrated graphics unit allegedly consists of 1536 Sheder Core. There will also be 12 GB RAM, possibly as LPDDR5-6400 or even rapid LPDDR5-7500.

With high clock frequencies, the GPU blanket should be about eight times more with the previous switch. At that time, 400 GigaFlops of the selected Tegra-X1 processor now make almost every PC processor with an integrated mini-GPU.

There are also architectural reforms and quite fast and large memory. Hence the factor 10 can also be without AI-optscaler.

Meanwhile, a high growth in performance is bitterly essential, as the Tagra-X 1 processor of the old switch was already old and slow. Even in Switch 2, however, there should be no current technology, but at least some that are more suitable for game console.

Fight Spam: WhatsApp wants to more regulate messages

DLSS, but perhaps without a frame generation

The GPU should belong to the ampere generation, namely 2020, the position of the RTX 3000 GeForce series. It is the second generation with retrizing and tenser core for AI algorithms. In view of moderate computing power, very clear retrusting graphics effect is not expected. Even a Geforce RTX 3050 should be much faster than switch -2 GPU.

The 2020 technology stand will explain why Nvidia has mentioned the support for deep learning super sampling (DLSS), but does not respond to any accurate tasks. NVIDIA AMPERE DLSS is independent of 3.5, yet a frame for launched launched frames without generation. DLSS 4 is available only from Ada Lovelace (GeForce RTX 4000); Multi-frame generation (MFG) with multiple intermediate frames from Blackwell (GEFORCE RTX 5000). Switch 2 should mainly use AI-Apscaler to reduce resolution and thus increase the frame rate. For comparison: Switch 1 comes from the GPU Maxwell generation, which began in 2014 as a Gefforce GTX 900 with AI hardware.

If you want to play on a television with switch 2, you are limited to HDMI 2.0 with 60 Hz in 4K resolution. After all: The console can produce 2560 × 1440 pixels (WQHD) with 120 Hz.

The Cortex-A78C core, meanwhile, comes from 2020, like Ampere GPU. More comprehensive CPU core, double number and possibly high clock frequencies bring a significant jump.

Four Cortex-A57 of the old switch came up with barely one gigaherts. In the case of Switch 2, the 2 GHz or more CPU clocks are close to the frequencies-Jetson Oin AGX also makes it, and ARM has also adapted the A78C core towards this price. It is unknown that Samsung or TSMC produces processors. Jetson Oin AGX and all RTX -3000 GPUs come from the old 8 -nenometer of Samsung. But there were also rumors that NVIDIA could switch to a process of 7 Nm class with Switch 2 branch.

(MMA)

Also iOS navigation app default: Google Maps instead of Apple Maps