fbpx
fcp.co logo transparent
fcp clapperboard
Welcome, Guest
Username: Password: Remember me
{JFBCLogin}
25 Jan 2021
New boarders will have their posts moderated - Don't worry if you cannot see your post immediately.
Read More...
  • Page:
  • 1

TOPIC:

Looks like FCP X has a tiny bit of room to grow. FCP X vs. Motion 5 benchmarks 16 Aug 2011 10:00 #842

  • tpayton
  • tpayton's Avatar Topic Author
  • Offline
  • Platinum Member
  • Platinum Member
  • Posts: 929
  • Karma: 25
  • Thank you received: 131
FCP X render times just feel slow don't they? I did a little testing and I think that Apple has a bit of optimization to do that will make some dramatic improvements.

Test Machine: 2009 2.2 ghz iMac, 8 gb ram, 256MB GForce 9400, 10.6.8.

Test Clip: 10 second h264 from Canon 7D. 1080p. Filters Applied: Bad TV. Sharpen. FCP X set to playback of Better Performance. Motion set to Full Res, Normal Quality.

Measurement Tool: iPhone Stopwatch, ATMonitor App

Note: I learned by another poster that FCP X always renders in Best Quality, so Motion 5 needs to be tested in Best, not Normal. Plus the Bad TV filter was different in Motion and FCP X. Therefore these Motion numbers are not accurate.

Playback:
FCPX: GPU 75% CPU 60%. 15 fps (approx)
Motion 5: GPU 76% CPU 67%, 6 fps

Render/RAM preview:
FCP X: GPU 0%, CPU 99% 1:33
Motion 5: GPU 60%,CPU 65% 1:27

Share ProRes 422 (no prerendering):
FCP X: GPU 30-40%, CPU 65% 2:19
Motion 5: GPU 50, CPU 80%, 1:28

Share h264: (no prerendering)
FCP X: GPU 30-40%, CPU 85% 2:22
[b]Motion 5: GPU 60-90%, CPU 95% 1:12[/b]


Conclusion: Interesting that FCP X doesn't use the GPU when rendering, just exporting. Perhaps we will see some improvement or at least less tax on the CPU. But look at the speed of Motion! It can can export with render ProRes almost twice as fast as FCP X. In fact h264 export is twice as fast! It looks like Apple can do a little tweaking on FCP X and we will see some spectacular speed gains.

Please Log in to join the conversation.

Last edit: by tpayton. Reason: Typo

Re: Looks like FCP X has some room to grow. FCP X vs. Motion 5 benchmarks 16 Aug 2011 16:01 #853

  • tap5a
  • tap5a's Avatar
  • Offline
  • Senior Member
  • Senior Member
  • Posts: 48
  • Karma: 1
  • Thank you received: 17
One good thing about background rendering working only with CPU is that there are more possibilities for network rendering etc. because GPU is not needed. Also machine doesn't halt if Final Cut Pro X renders while you're working on some other GPU intensive program. But definitely there's place to improve for Apple.

Another thing is that Final Cut Pro X always plays Motion files + Effects with NORMAL QUALITY but renders with BEST QUALITY. So rendering test should be done in Motion with Best Quality.

Best
Tapio Haaja

Please Log in to join the conversation.

Re: Looks like FCP X has some room to grow. FCP X vs. Motion 5 benchmarks 17 Aug 2011 21:50 #877

Why use the GPU if you have so much power in the CPU?
Using both would be cool...

Please Log in to join the conversation.

Re: Looks like FCP X has some room to grow. FCP X vs. Motion 5 benchmarks 18 Aug 2011 08:32 #894

  • tpayton
  • tpayton's Avatar Topic Author
  • Offline
  • Platinum Member
  • Platinum Member
  • Posts: 929
  • Karma: 25
  • Thank you received: 131
Tapio-
Thanks for your insight, and I apparently I ran the test completely wrong. Plus after double checking the settings, I found that the "Bad TV" filter in FCP X wasn't just the Motion 5 Bad TV effect, but Bad TV plus Add Noise. Then as you stated it wasn't rendering in Best Quality.

So apparently I need some egg on my face! Here are new numbers:

Motion (Best, Bad TV, Add Noise, Sharpen)
ProRes 422: 1:48 (average)
H264: 1:51 (average)

FCP X (Bad TV, Sharpen)
ProRes 422: 1:52 (average)
H264: 1:52 (average)

Conclusion: My first test was incorrect and misleading. Motion might be a tiny bit faster, but nothing as extreme as I saw before. However, there is still the issue with the CPU and GPU relationship. Motion is using the GPU for Ram Preview while FCP X doesn't use it for rendering. So there is some more that can be squeaked out of FCP X if I can take more advantage of the CPU.

The other interesting outcome is the h264 render times are not any different than the ProRes.

Again I apologize for the shortcoming on my test.

Please Log in to join the conversation.

Last edit: by tpayton. Reason: typo

Re: Looks like FCP X has some room to grow. FCP X vs. Motion 5 benchmarks 18 Aug 2011 14:51 #899

  • tap5a
  • tap5a's Avatar
  • Offline
  • Senior Member
  • Senior Member
  • Posts: 48
  • Karma: 1
  • Thank you received: 17
Thank you very much Tpayton for new tests!

Another small thing to note from Motion's Tech Specs is that Motion renders material with GPU and 16-bit floating point and Final Cut Pro X with CPU and 32-bit floating point.

From Motion's Tech Specs:

Color bits per channel: 16-bit float for GPU rendering; 32-bit float for background software rendering of Motion projects in Final Cut Pro (float processing requires a graphics card with 256MB of VRAM; 512MB or more recommended)

-Tapio

Please Log in to join the conversation.

  • Page:
  • 1