The New Apple Mac Pro

[N19h7m4r3]

How often/quickly do the d700's drop from the 850mhz state? I recall a discussion in the original M-P thread involving some hefty reasoning over just how much if ever they would get to 850. I figured if they used the 7990 chips/bin they came from (it was discontinued) they would get a combined tdp near 375w, which would allow a decent use of near max boost.

The primary GPU is always at 850Mhz, it's the second one that down clocks when not used.
The configuration is GPU #1 is Video, #2 is rendering( openCL ) in general.

It's up to the application to effectively use both. For FCPX, and Luxmark they'll gladly run at 850Mhz until the job is completed, and then it'll down clock again. This is only the case in OSX.

In Windows both run at 850Mhz from what I've seen, hence the better performance, and scores. Especially for 3Dmark

On average the D700's are around 10-15% slower than a W9000 depending on application and use. Although many applications are still not using them or are using older versions of OpenGL/CL, which handicaps performance



In Luxmark a 7970 in OSX will get you 1222 in Sala, while the D700 nets you 1755.
So it's a clock of 925Ghz Vs 850Mhz with the D700 getting out ahead.



The Ars review by a professional user was very enlightening, and highlights the strengths and weakness very well.
Developers need to actively work too take advantage of both GPUs to their fullest, and AMD seriously need to work on their lacklustre OpenGL performance.

http://arstechnica.com/apple/2014/01...-2013-mac-pro/

It's shameful that a GTX660 can match the D700 & 7970's in OpenGL performance simply down to shoddy drivers.

Thanks, I think this will get better for more apps, hopefully.

It's getting there. Even Aspyr the folks that ported Civilisation to the Mac have managed to let the second GPU be used for computation work. Netting better performance and quicker calculations, and nicer 4K support.