current systems throttle down when things start to overheat, this is "non-deterministic" and depends on environmental factors etc. this is harder for developers to optimise for
whereas the PS5 knows it's own power use and power budget, as long as it stays within that budget then it won't overheat. and this means the throttling is 'deterministic' - the same code running will always throttle in the same way at the same time, so this is easier for developers to test and optimise for
> send any unused power from the CPU to the GPU so it can squeeze out a few more pixels
What it sounded like to me (also a layman) is that they have a fixed energy budget that's tied to the thermal properties of a baseline unit in a neutral environment. As long as they're under that budget, they can boost up the clocks of individual components that are under load. So heavy-rendering cycles can use relatively more GPU and heavy-compute cycles can use relatively more CPU.
That might be totally wrong, though, and I would also appreciate an expert's explanation.
There should be power saving modes but that's not what they're talking about. Sony is saying that you can run the CPU at 3.5 GHz xor the GPU at 2.2 GHz but they can't both run at full speed at the same time. So maybe you can have CPU @ 3.3 and GPU @ 2.2 (favoring GPU) or you can have CPU @ 3.5 and GPU @ 2.0 (favoring CPU). I guess most games will favor the GPU but I'm not a game developer.
(As an aside, I coined the term "power shifting" in 2005 to refer to this concept and I'm happy to see it being incorporated into basically everything, even if they probably independently reinvented it.)
My guess is it has something to do with cost savings (cheaper to manufacture silicon that has to perform “better” some of the time as opposed to all of the time).
You can allow one game to use more gpu while reducing cpu and vise versa.
Just imagine a cpu heavy rts vs a graphic intense rpg