Lets get some things straight, /g/, CPUs are pure math barring

File: Screenshot from 2016-08-27 22-15-27.png (70KB, 558x958px) Image search: [Google]

70KB, 558x958px

Lets get some things straight, /g/, CPUs are pure math barring disastrous implementations, and that's final.

The math here shows AMD's Zen core being weaker than a Skylake core in pretty much every metric, integer and AGU output, FPU, load/store/entry/register/retire size.

The only thing that might show promise is the much larger much more complex (4vs8 way) L2 cache, that might make up for the lack of hardware resources, but as said, it's more of a complex option, I can assure you if AMD had more time and money they'd go for larger resources instead of a more complex cache like in this case.
The L3 victim cache might be alarming, they're known to be much higher latencies than inclusive caches but this was obviously known in the beginning of the design phase so there might be a surprise here, AMD mentioned 5x the b/w of BD L3.

On the other hand Zen seems pretty much equal to Broadwell in execution resources, so it's no wonder its IPC is pretty much equal.


Anyone feel like adding in their thoughts?