I own a 2010 8 core x 2.4 Mac Pro. OSX Mavericks 10.9.5 with Metric Halo ULN-2 Interface.
I’d like to upgrade my processors to 12 core 3.06 or 3.46 Ghz.
Anybody here using such a configuration with success? I am experimenting crackling @ lower latencies on original setup and want to avoid that on bigger sessions especially when running templates with Ivory II, UVI synths, Kontakt etc…
Would this upgrade help my case? Would I be better off with a more modern computer (Mac Pro, BacBook Pro retina or Imac?
from 2.4 to 3.06 it’s supposed to be quite faster. There are 3.46 CPU’s available to upgrade to but am affraid of heat since they work at 130W instead of 95 W each. X2 this is way more power requested…
RdRm is correct, clock speed matters a lot for low latency. That’s why I just built an 8 core 5960X i7 instead of a 12 core Xeon (PC of course). I am running the 8 core at 4.3Ghz, liquid cooled, but you cannot overclock the Xeons, so the 12 core is limited to 2.7GHz. In this case, the clock speed difference is so great that the 8 core outperforms the 12 core in all settings, but especially more at 32 samples.
Keep in mind that for the newer machines (including nMP) these cores are all on a single CPU, so heat is more of an issue and somewhat limits clock speed, versus the dual CPU on an older Mac Pro. However, there is some performance penalty involved with dual CPU (memory access, timing, etc) so the older duals need more clock speed for the same performance as a single CPU machine.
My 8 core at higher clock speed outperforms a 12 core Xeon at it’s lower clock speed when using 32 samples (or even 64 to a lesser degree) under load. It actually outperfoms the 12 core in all cases, because of clock speed, but more pronounced the lower the latency is set. Buy as much clock speed as you can afford for very low latency performance. Of course, the more cores the better. But for very low latency, CPU power per core becomes more and more important.
You would be increasing clock speed by 25% and number of cores by 50%. You would definitely improve performance and the more plugins you would like to load at low latency, the more benefit you would see.
For higher latency mixing, etc you would see a large improvement. At 32 samples, it’s hard to say, but I would expect at least 25% improvement, maybe as much as 50%. Depends on your specifics. Just a guess, but maybe a close one.
One easy way to figure. Multiply number of cores by clock speed.
2.4 X 8 = 18.4 GHz total
3.06 x 12 = 36.72 GHz total
For mixing at high latency, about double the performance. For low latency, 25 - 50% is my estimate.