bitsandchipsNo Dual and Four Cores Zen CPUs, at least initially

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Oct 10, 1999
11,915
0
76
Amdhal law is not at play in CB R15, the test is supposed to scale at almost 100% even with 64 cores.
But it does not. From anandtech.com/bench

Code:
Ivy Bridge
E5 2697 v2 24c/48t 3.0 GHz turbo - 2988 - 41.5/c/GHz
E5 2687 v2 16c/32t 3.6 GHz turbo - 2508 - 43.5/c/GHz
E5 2687 v2 8c/16t 3.6 GHz turbo  - 1297 - 45/c/GHz
i7 4960X 6c/12t 3.7 GHz turbo    - 1097 - 49.4/c/GHz

Haswell

E5 2697 v3 28c/56t 3.1 GHz turbo  - 3994 - 46/c/GHz
E5 2687W v3 20c/40t 3.2 GHz turbo - 3060 - 47/c/GHz
E5 2650L v3 24c/48t 2.1 GHz turbo - 2433 - 48/c/GHz
i7 5960X 8c/16t 3.3 GHz turbo     - 1337 - 50/c/GHz
As the the core count increases, the score/core/GHz decreases. Take the E5 2697 v2 24c/48t for example. If scores scaled linearly with cores, then it should score 49.4*24*3=3556. On a quad processor system, the scaling should be even worse since code and data are local to one processor only (unless you're implying that Cinebench is NUMA aware).

Anandtech says Cinema 4D is NUMA aware. I guess core scaling would be significantly worse if it wasn't.
 
Last edited:

Similar threads



ASK THE COMMUNITY