Well, as already stated by DF, devs will most likely throttle one to maximize the other. The thing is, doing it this way will allow devs to extract more performance than having set clocks. Otherwise we would likely have a PS5 running at say 3.2 GHz CPU and 1.9 GHz GPU flat, for example. But this way, you can boost your CPU or GPU at the times where the other is idle.
no it was stated that the reason it throttle back is because the games dont use as much CPU, why have a high clock just to be idle?
“Several developers speaking to Digital Foundry have stated that their current PS5 work sees them throttling back the CPU in order to ensure a sustained 2.23GHz clock on the graphics core.
It makes perfect sense as most game engines right now are architected with the low performance Jaguar in mind - even a doubling of throughput (ie 60fps vs 30fps) would hardly tax PS5's Zen 2 cores. However, this doesn't sound like a boost solution, but rather performance profiles similar to what we've seen on Nintendo Switch. "Regarding locked profiles, we support those on our dev kits, it can be helpful not to have variable clocks when optimising. Released PS5 games always get boosted frequencies so that they can take advantage of the additional power," explains Cerny.”
current games use GPU to calculate things that used to be calculated on CPU, for example around the time uncharted 4 was released the devkits for PS4 were updated to calculate physics on GPU by default that means more free time for the jaguar CPU so naturally it doesn't require as much on a more powerful CPU and you can throttle back the clocks it doesnt imply you cannot use both at max clock in fact it was stated
"There's enough power that both CPU and GPU can run at their limits of 3.5GHz and 2.23GHz, it isn't the case that the developer has to choose to run one of them slower."
and yes if they run at that speed is because they will use a workload that requires it otherwise they can simply down clock, what is the point of having a high clock if you will be idle? if you can downclock do it
Say for example you have to render 50 people on the street with high polygons. That is high on the CPU, but a street itself is not that taxing on the GPU. You lower the GPU power usage and give the CPU its max performance.
If you have a scene with ray tracing but with barely any polygons in it, the CPU will not need to use that much power but the GPU is under a heavy workload. You then shift the power to the GPU.
If you had set clocks, you would have lower performance in either of the above scenarios.
NO, games used to be like that like 20 years ago, back then you gave the vertex data to the GPU calculated by CPU to render like the direct mode of opengl, that is why carmack spent lot of time batling to send the absolute minimum required geometry in quake for rendering, it changed, we store vertex data in vertex buffer objects and vertex array objects and those are on GPU memory, the CPU now only tell GPU which one to use to render a specific character or object scene, to tax the CPU using lot of characters require you to run certain calculations on that are taxing to CPU for example some physic boxes(those are calculated on GPU now) and AI calculations and making lot of drawing calls(on PC) as those have to be "negotiated" with the OS and driver so they consume CPU time and you prefer to draw many objects with one single drawcall or use something like Vulkan/DX12