Trimesh
Banned
Electrical is different from electronic!
How does your equation apply to a GPU?? Where did you factor the number of compute units in your equation? Doesn't Voltage increase along with the number of CUs? Doesn't Capacitance too? Isn't the increase in CUs bigger than the frequency increase????? Wouldn't XSX's polynomial increase be bigger than PS5? And again, electrical is different from electronic
Well, at least where I live the degree is called "Master of Science in Electrical Engineering", despite the fact that the course is overwhelmingly focused on electronics. Maybe your country is different, but here the only qualifications that you can get that are specifically called "electronics" are vocational ones.
I was simply taking exception to the claim that a smaller but higher clocked chip can't take more power than a larger but lower clocked one - because it's just not true.
As for the polynomial, that's something you would have to determine experimentally - if you can hold V constant (say because the clock is relatively low for the process) then it will contain only first-order terms since it's basically a factor of the reactance of the device gates at the frequency they are expected to transition at.
The problem is that this may not be the case - each gate input effectively contains an RC network where the R is formed by the Rds(on) of the drive FET and C is determined by the gate capacitance - with a given gate threshold voltage (determined by the process) as the clock increases you eventually reach the point where the voltage can't rise to the threshold voltage in time, and the device can no longer work. This can be addressed by turning up the supply voltage (since the higher voltage allows more charge to be pushed though the FET in a unit time) - but at this point the x^2 term starts to kick in and the power increases drastically.
Again, I have no idea if this is or will be the case with the PS5 - but claiming "it can't happen" is simply factually wrong.