Geekbench 2 vs. 3

[StephenCampbell]

I've done some experimenting, and it seems that any score that a machine gets in Geekbench 2 will be about 110% of what it will be in Geekbench 3.

This means that while the Ivy Bridge i7 iMac says 13,772 on everymac.com and the Haswell i7 shows 14,449, that 14,449 is in Geekbench 3 and actually is a 15,893 relative to the 13,772 of last year's machine. This is a 15% performance increase. Not bad!

 

[onekerato]

and actually is a 15,893 relative to the 13,772 of last year's machine. This is a 15% performance increase. Not bad!

Geekbench measures raw processor and memory performance. Macworld publishes benchmarks based on application tasks and may be a better gauge of the % improvement you're likely to see.

Macworld: http://www.macworld.com/article/205...boost-from-pcie-flash-storage-faster-cpu.html

 

[Tutor]

I've done some experimenting, and it seems that any score that a machine gets in Geekbench 2 will be about 110% of what it will be in Geekbench 3.

This means that while the Ivy Bridge i7 iMac says 13,772 on everymac.com and the Haswell i7 shows 14,449, that 14,449 is in Geekbench 3 and actually is a 15,893 relative to the 13,772 of last year's machine. This is a 15% performance increase. Not bad!

For two of my systems running Linux, that percentage would be a bit low:

(1) AlphaCanisLupus1 (32-cores) - Geekbench 3 = 2,986 Single Core (SC) / 71,691 Multicore Cores (MC) [ http://browser.primatelabs.com/user/9077 ]; AlphaCanisLupus1 - Geekbench 2 = 58,027; 71,691 / 58,027 = 1.235 or about 24%.

(2) AlphaCanisLupus0 (32-cores) - Geekbench 3 = 2,990 SC / 71,367 MC [ http://browser.primatelabs.com/user/9077 ]; AlphaCanisLupus0 - Geekbench 2 = 56,997; 71,367 / 56,997 = 1.252 or about 25%.

Some of that percentage increase with my systems could be due to Primate's having made Geekbench 3 more aware and responsive to systems with a large numbers of cores. Changes like that made by benchmark authors can be flies in the comparison oinment.