drkohler said:
No I'm not and I don't think you see the big scope. You are still pretending that an x% clock increse in the main loop increases framerate by x%. That is NOT true for an optimised system. Let's make an extreme example. Our software has 2 tasks that run at around 60% of a core at 1.6GHz. This requres 2 cores that may either be idle for the remaining 40% or at best we can squeeze in some low% task. With 30% more clock, we can squeeze both 60% tasks into one core, and so we have freed an entire core. Again the point is that a 30% clock increase with reoptimisation of the whole tasking system that results in much more than a 30% framerate increase. Whether any developer actually does this remains to be seen. A completely different optimisation in the Neo would be a hardware revision of the generic dual cpu design of the PS4. In this case, they should have gone with 8 Jaguar cores on a common 4M L2 cache instead of he silly dual 4 core on 2M L2 cache Jaguar design. I cannot fathom which idiot went for that design choice. |
Sheesh..... no one is sayomg x% cpu increase = x% framerate boost. You are arguing with yourself.
what has been said and has been agreed on is pretty simple.
there are gsmes that are CPU heavy(A). Today these are games that struggle to even maintain a 30fps framerate. Then there are games that are locked to 30fps (B) not out of necessity but due to stability. What you are saying applies to case B. Where with a slightly more powerful CPU on the easy side of things, or with a little more elbow grease in engine optimization you can have gains of going from a locked 30fps to 60fps.
In the case of A, even with the more powerful CPU what they end up with is just the case B. Basically an all round better performing engine on the CPU side, but not one that hits 60fps but rather averages 35-45fps internally and this gets locked down to 30fps.
