Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.


macrumors member
Original poster
Mar 1, 2016
I received several new processors and delidded all of them this weekend. I also installed all of them into my 2009 Mac Pro to ensure they all worked. As long as they were installed I figured I'd also benchmark all of them.

  • All testing was done within 60 minutes from start to finish so ambient temperature in the room was approximately the same. For the record the thermostat was set to 71ºF in the house.
  • I tested X5675 (3.06GHz Hex), X5680 (3.33GHz Hex) and X5690 (3.46GHz Hex) processors all in dual CPU configuration.
  • All testing was done on the same CPU tray, with the same sticks of ram in the same slots, in the same 2009 Mac Pro, with only the current benchmark software running.
  • I tested each processor with 8 sticks of RAM, removed the ram in slots 4 and 8, PRAM reset to be safe, and retested with 6 sticks of ram.

The raw data can be found in the following Google Sheet.

Let me know if there is any trouble accessing the file. I might have to edit the permissions.

You'll notice additional tabs across the bottom of the document. If anyone can come up with a better way to analyze the data I'd be happy to implement all of your suggestions. For lack of a better way, I simply divided the respective score of each configuration by both the TDP (in watts) and also by the current lowest price I could find on eBay.

This information has been requested by several people to help make better choices when upgrading their systems. The prices on the X5690 3.46 Hex have plummeted in the last month so hopefully more people can afford these beasts. The X5675 is still an attractive option since the TDP is only 95w vs 130w on the other two processors.


macrumors G3
Sep 21, 2010
Wow, Geekbench places a gigantic weighting on RAM throughput. Almost 25% in a couple of cases. Way too much IMHO.
  • Like
Reactions: Fooze


macrumors P6
Apr 3, 2014
Hong Kong
Thanks for all the work, the benchmark is pretty much reflecting the expected CPU processing power. And the performance difference is well within the reasonable predicted range. (well, except GB 4.1 multi core score. That's really hard to understand because of the unknown non linear scale)

CPU speed (multi core) - [(3.46-3.06)/3.06] x 100% = 13.07%

GB 3.4.1 (multi core) - [(32080-28848)/28848] x 100% = 11.2%

R15 (multi core) - [(1627-1438)/1438] x 100% = 13.14%

*GB 4.1.0 (multi core) - [(25124-23576)/23576] x 100% = 6.57%*

CPU speed (single core, turbo) - [(3.73-3.46)/3.46] x 100% = 7.8%

GB 3.4.1 (single core) - [(2804-2635)/2635] x 100% = 6.4%

GB 4.1.0 (single core) - [(2955-2791)/2791] x 100% = 5.8%
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.