Digital Foundry: Nintendo Switch CPU and GPU clock speeds revealed

[Alby_da_Wolf]

SvennoJ said:

Alby_da_Wolf said:
Same CPU speed in both modes is quite obvious, as game logic must always run in the same mode (at most they could cut it by a few tens MHz by suspending unessential non-gaming services that they possibly chose to run in background while docked). GPU speed and so graphics level is obviously where big power saving can be achieved when in portable mode, and this too shouldn't be matter of arguments.
What can really disappoint people with sound reasons is the choice of top speed when docked.

Is it that obvious? PS4 pro has a modest boost in CPU speed. Scorpio will have a much bigger difference in cpu speed compared to xbox one. Does this mean that the only difference between undocked and docked will be resolution? More effects, physics, detail, draw distance, all require more attention from the cpu as well. Or will the game run slower while undocked.

It seems likely the undocked specs are the main target, with docked only adding a resolution boost. Yet if games are tailored to 720p regarding detail, draw distance etc, won't they look a bit sparse in 1080p? Like playing a crossplay ps vita game on ps3.

I wonder how the switch will be marketed. Home console $300, $350 deluxe with larger internal storage, like the wiiU release, with $60 games. Or will it launch as a handheld at $250 with $40 games.

If the device is mainly a portable, gameplay must not be gimped when it's undocked, this means that even undocked, viewing distance must be the same and the gaming world must be kept active within the same distance in both modes, but this doesn't prevent devs from boosting graphics detail when docked.
Obviously devs with less resources will make the undocked specs the main target also for graphics, but games with res and detail lower than what HW could allow happen also on HD consoles with just one use mode.

Pemalite said:

[...]

haqqaton said:

After seeing some OpenGL vs Vulkan comparisons and considering that Switch supports Vulkan and that Vulkan is great for ARM mobile chips like Tegra X1, I think we can be confident to say that Switch, even undocked, will be fairly better than Wii U. To be fair, they are comparing Vulkan to OpenGL|ES in the videos below but I think that the point stands.

Vulkan is irrellevent, it's not going to be some magical "Secret Sauce".
The Wii U has a low level API that is higher performance than Vulkan if a Developer wishes to build for it.
Same goes for the Xbox and Playstation platforms.

An API like Vulkan does offer better performance than higher level API's like Direct X 11 and OpenGL.

Alby_da_Wolf said:
Same CPU speed in both modes is quite obvious, as game logic must always run in the same mode (at most they could cut it by a few tens MHz by suspending unessential non-gaming services that they possibly chose to run in background while docked). GPU speed and so graphics level is obviously where big power saving can be achieved when in portable mode, and this too shouldn't be matter of arguments.
What can really disappoint people with sound reasons is the choice of top speed when docked.

The annoying part is... Nintendo could have thrown out the Big cores in the Tegra chip and just kept the slower, more power efficient cores and then used that extra TDP to throw more clock rate at the graphics chip.

They could have saved money as well due to less die-space.

About APIs, well, being able to develop for just one HW config allows console makers to go to lower level even in the dev tools and standard libraries and get higher optimisation without entering into a nightmare of countless different low level optimisations as this approach would cause on PCs, efficiency will always be higher.

About Tegra cores, indeed, underusing the power of a big chip doesn't look like the best possible design approach for a portable. Early killing of Wii U forced Ninty to take some rushed design choices, maybe...