Quote Originally Posted by demonkevy666 View Post
Ok so pretty simple
taken IPC deficiency, because scaling can't be prefect then putting it at about 60-65% of that pixel fill rate seem about right.

so what's the Nvidia Math that seems more accurate ? if you have any idea....



I've killed one card with a metal fan clip on the back of it over the gpu die area while it was plugged in :/
I might have killed another similarly, power supply didn't thoroughly dissipate it's charge after being unplugged. Battery to the board fell on it. landed in the same spot as the fan clip.


I'm disappoint in the, PC enthusiast, PC gamer's "PC master race" as whole lately just impatient and the unwilling to wait for stock to fill up, but ok with buying things as soon as they release.
people where ok with buying $1,200 RTX 2080 TI's when it's MSRP price was what $800 now they want to complain about it during a world wide a pandemic. after production was stopped for a few months.
All you have to do wait three months and not buy anything, stock volume will pick up by then.
They are the same thing now. In the past Nvidia had a shader clock instead of logical cores so you would multiply the shaders by the shader clock (not core clock.) The ROPs and TMU are the same math. The big change is that NV more physical shaders that can do less threads (not a great analogy but it helps) so amd is counting bulldozer style to get higher flops that games wont utilize. With amd RDNA has fixed some of it but only the main shader in a unit can do things like schedule out of order operations. AMD also has a bottleneck getting data between the ROPs and the shader clusters that Nvidia does not have. Since most games are made a favor NV due to their larger market share it make the IPC look bad for AMD. On properly coded/optimized games that make full use of asynchronous compute the bottleneck is much less.