As spotted by Mac Rumors:
…between the 6-core and 8-core machines, the two unsurprisingly show similar single-core scores of around 3300 in Geekbench 3, while the newly surfaced 6-core machine sees a lower multi-core score of 18309 compared to the 24429 seen on the 8-core version.
Performance on almost all current high-end Intel chips is pretty similar, thanks to thermal limits and Turbo Boost — you can see that the new 2.6 GHz Retina MacBook Pro matches the Mac Pro’s single-core score, and the multi-core score scales almost perfectly linearly, even between the mobile and Xeon parts, with the number of cores.
Based on Apple’s stated specs and process of elimination, these are all of the new Mac Pro’s CPU options:
|E5-1620 v2: 4-core 3.7 GHz
|E5-1650 v2: 6-core 3.5 GHz
|E5-1680 v2: 8-core 3.0 GHz
|E5-2697 v2: 12-core 2.7 GHz
Notably, they left out better 6-core and 8-core options with the same TDP that seem to have no other disadvantages except a few hundred dollars, which hardly matters at the price range of a high-end Xeon workstation:
|E5-1660 v2: 6-core 3.7 GHz
|E5-2667 v2: 8-core 3.3 GHz
On paper, and these Geekbench results from the 6- and 8-core support this, it looks like you really can’t go wrong with any of the CPU options below the 12-core — the deciding factors should only be cost and how much parallelism you can take advantage of.
The 12-core will not only be very expensive, but it will likely suffer in single-threaded tasks with its low Turbo Boost ceiling, so I wouldn’t recommend it unless it will spend most of its life doing highly parallelized CPU tasks that actually max out all 12 cores.
If I end up buying a Mac Pro of this generation, having seen no other benchmarks or Apple’s likely very inflated CPU pricing, I’d probably go 6-core. The 8-core is tempting on paper, but I bet it’ll be a very big price jump for what ends up being very similar performance most of the time.