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

Real world D300 vs D500 vs D700 FCPX Benchmarks

^ Wow, the 6 core rendered faster. Pretty amazing that the 6c D700 is the best for FCP X performance. Duly noted! Interesting too considering the 8 core has a higher 64 bit multi core Geekbench score... I guess the clock speed explains it!

One clarification! Does adding Romantic and rendering happen with the title still attached? That's what I did. Or did you disable it and then render Pages with Romantic? That would explain why some people have Romantic rendering slightly longer than clouds, and why others have Romantic rendering in about half the time as Clouds.

I did my machine for fun (specs in my sig) 2010 gen with HD5870 (clearly the biggest bottleneck!) Looking forward to upgrading my GPU and would like to try these tests again. I'm amazed how much faster the iMac did than my machine!

"Pages" - 1:33
"Clouds" - 7:25
"Romantic" - 8:16
ProRes 422 - 8:16
h.264 (compressed master file, didn't "Send To Compressor") - 6:13
 
i'm astonished by that fact as well. I think it requires more testing though, i can'T believe this is the answer, since Apple specifically designed fcpx and the Mac Pro to 'build a perfect unity'.

Yes, i just kept adding the effects, nothing was removed or disabled. The test was done on the internal (500gb) drive, another test on our promise pegasus2 was slightly slower - that should change though under heavy workload.

I also tested my 2013 macBook Pro, the results were between 30% and 70% slower.
 
AidenShaw said:
That's certainly the story that the Apple marketing machine is pushing....
Click to expand...

I must say with the above results the machine really does well with OpenCL and FCP X performance. I would like one but already have too much into my 2010 to do that for awhile :D
 
I put in an MSI Gaming 3G R9 280X tonight. (Using ePower 450W external GPU PSU.)

Pages: 1:46 (about 14% slower than the HD5870)
add Clouds: 5:33 (about 33% faster)
add Romantic: 6:10 (about 34% faster)
Pro Res 422 Export: 1:03 (over 687% faster)
h.264 encode: 6:04 (about 2.4% faster)

Consensus: Scrubbing in FCPX seems snappier. If you are exporting long projects, this card is for you! The investment was under $400. I have yet to do an actual editing session on the card since I installed it tonight, but I expect it should have increased performance to a noticeable degree. Yes, it works OOB with no boot screen. If anyone has questions about how I hooked up the ePower feel free to ask. Was it worth the money? Yes, for me. I can recoup the cost of the card easily, and unless I put in a second one, this machine is tapped out for performance at this point.
 
Last edited:
Interesting results there, isn't the 280 performance supposed to be the same of D700(according to somebody here)?
 
This is a great thread.

One note though: the lack of Quick Sync will usually mean encodes to H.264 are faster on Haswell MBPs and iMacs and so on, but if you do a multi-pass encode for quality (as I do for all my clients), the nMP will always be faster 6 cores and up.
 
Good thread!

Im looking to update my 08 pro and I use a Nikon D800E and a Phase one with (capture one) program.
 
N19h7m4r3 said:
There's two D700's in the system, compared to a single 280X.
Click to expand...

280x:
add clouds: 5:33
add romantic: 6:10

2x D700:
add clouds: 1:41
add romantic: 1:47
even doubling the processing time(considering just one D700):
add clouds: 3:22
add romantic: 3:32

Still much faster;)
 
sirio76 said:
280x:
add clouds: 5:33
add romantic: 6:10

2x D700:
add clouds: 1:41
add romantic: 1:47
even doubling the processing time(considering just one D700):
add clouds: 3:22
add romantic: 3:32

Still much faster;)
Click to expand...

That is interesting, but then we have to remember the nMP has a newer Processor that's faster clock for clock, and faster RAM, and SSD. It's a better system overall. One of the reason I got one and sold my old MP. :)

Although it'll be interesting to see how the new GTX 980 fairs in these tests, given that in OpenCL it beats even the R9 290x now.
 
CouponPages said:
....you can't tell if the D300 won simply because it was a single threaded program that runs better on a 4 core than an 8 core...
Click to expand...

Why not just restrict active CPU cores to 4? That makes a level playing field. It eliminates the uncertainty over the performance contribution of additional CPU cores.

This article describes how to limit the number of active CPU cores for testing purposes. It can be done from the terminal command line and doesn't require any special utilities. I just tested it on OS X 10.9.5, and it works fine.

(scroll down to heading: "Setting the Number of Active Cores from the Command Line") https://developer.apple.com/library/Mac/qa/qa1141/_index.html
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back