Intrinsic said:
Lol..... wishful thinking much? Ok. lets just put something in perspective. If PS4neo has a GPU that is twice as powerful as the one in the PS4, there id only two ways for that to happen. 1. Double the GPU compute units from its current 18 - 36 or, 2. Keep everything as is and instead double the GPU clock from its curremt 800mhz to 1600mhz. Both of those are IMPOSSIBLE to do at a 28nm process and and put it in a box the size of the PS4 and dofr thr latter, jist flat out impossoble cause the chip wont be stable. For the PS4neo to be using a 36CU GPU, that can only happen with Polaris. Only rraly question here is what clock Sony chooses to run it at. 900mhz/1200mhz+ is really all the difference between 4TF and 6TF as far as GPUs go. Well more like 4TF and 5.5TF but you know, 6TF sounds cooler. |
Well. Almost.
You see for every process node like 28nm, 14nm etc'. - Fabs can have multiple types of processes, low power and low leakage, but low clock, high power and high leakage, high clock etc'.
Fabs market that to chip designers too.
GPU's take advantage of the fact that their workload is highly parallel, so it makes sense to go with more hardware and go with a more energy conservative process that in turn also tends to limit clockrate.
With that said a 28nm 18CU part running at 1.6ghz could be entirely possible, but it wouldn't be a simple case of grabbing a random Radeon off the shelf and overclocking it, there would need to be some work done under the hood... But then you may also introduce bottlenecks into the chip design itself, yields could suffer. etc'.
I have legit seen people on PC OC enthusiast websites overclock Radeon 7850's (PS4 Equivalent GPU) to 1.5ghz, but keep in mind with that clock and voltage, it's a chip that will suffer from the effects of electromigration pretty rapidly, the transisters were never designed to be pushed that hard.
www.youtube.com/@Pemalite
