Creative Communities of the World Forums

The peer to peer support community for media production professionals.

Activity Forums Creative Community Conversations FCP X hardware performance

  • FCP X hardware performance

    Posted by Oliver Peters on June 2, 2012 at 2:04 pm

    There have been some posts about hardware optimization and X. Here’s a little comparison I just ran.

    :65 ProRes 1920x1080p/23.98 media clip. Same timeline. I applied a MB Looks “blockbuster” preset. This internally uses diffusion+vignette+4 color correction filters. This was left unrendered and then exported, so rendering occurred on export. Exported (Share – Media Export) as current setting.

    1. 2009 MacPro 8-core (2.26GHz), 16GB RAM, ATI 5870 (1GB VRAM) GPU, 2 internal RAID-0 media drives.
    8 min. export

    2. 2011 MacPro 12-core (2.93GHz), 12GB RAM, ATI 5870 (1GB VRAM) GPU, Fibre Channel SAN media drives.
    6 min. export

    3. 2012 iMac 4-core i7 (3.4GHz), 16GB RAM, HD 6970M (2GB VRAM) GPU, internal SSD, Thunderbolt Promise media drives.
    4 min. export

    Also unrendered playback was reasonably close to real-time (or at worst it looked like around about 15fps) on the iMac, yet rather “stuttery” on the two MacPros.

    Clearly it looks like: a) X is optimized for the i7 processor, b) actual processor speed is more important than the number of cores, c) the more VRAM the better.

    – Oliver

    Oliver Peters Post Production Services, LLC
    Orlando, FL
    http://www.oliverpeters.com

    Elias Huch replied 13 years, 10 months ago 18 Members · 60 Replies
  • 60 Replies
  • Olof Ekbergh

    June 2, 2012 at 2:18 pm

    This is very similar to my findings.

    FCPX seems to really depend on new hardware to work well. I was really disappointed with how FCPX ran on my 2009 8core, this station runs FCP7 and M100 very well. But really stutters playback and has frequent beach balls from Hell when running FCPX.

    Using the new iMac i7 maxed out with a TB RAID, it works very well using Matrox MX02. It helped make me start to use FCPX instead of just “testing” it.

    Olof Ekbergh

  • Oliver Peters

    June 2, 2012 at 2:41 pm

    [olof ekbergh] “FCPX seems to really depend on new hardware to work well. I was really disappointed with how FCPX ran on my 2009 8core,”

    Fortunately, I really haven’t hit many “beach balling” issues on the Mac Pros. Maybe just lucky. The main thing I see is that once you move away from the built-in and Motion-template-based effects, performance is really challenged on the Mac Pros. In general, I find unrendered performance to be tolerable with the built-ins, Tonalizer, the DH filters and most of the FxF filters. Playback isn’t always smooth, but you can work with it and not lose the creative flow.

    I think it also depends on the complexity of the filter itself. For example, a filter that does one simple task (like the color board or a single vignette) is pretty light on the gear. A filter like MB Looks, DFT Film Stocks or even DV Shade Easy Looks is really applying the equivalent of 5-10 filters all as part of one plug-in. Therefore the “pipes” get clogged.

    Hence the importance of a fast machine (regardless of number of cores) and VRAM. Note the card in the tested iMac has double the VRAM of the ATI 5870.

    – Oliver

    Oliver Peters Post Production Services, LLC
    Orlando, FL
    http://www.oliverpeters.com

  • Jeremy Garchow

    June 2, 2012 at 3:10 pm

    Nice one, Oliver.

    It is certainly apparent that whatever NLE is next, we are going to need some new hardware.

    I guess the plan is working? 😉

  • Mathieu Ghekiere

    June 2, 2012 at 3:30 pm

    I’m surprised the MB looks filter still needs approximately 4 times real-time export even on your fastest machine. I haven’t run the numbers, but I’m wondering if Red Giant can optimize their Magic Bullet Suite more to speed up the render times.
    I have the suite in its newest version, bought it for compatibility for FCP X, but I was a *bit* disappointed with the exporting speed. Maybe I was expecting too much, with all the talk of the 64-bit FCP X support.

    I do love the quality of the Magic Bullet filters though. I find they give very nice results quickly (sometimes with a bit of fiddling, and their newer UI in comparison with their first ones, is much better and user-friendly for changing presets) and although it can dramatically change the colors in your video, I find it still keeps a lot of detail in the image.

  • Craig Seeman

    June 2, 2012 at 3:54 pm

    [Jeremy Garchow] “It is certainly apparent that whatever NLE is next, we are going to need some new hardware.

    I guess the plan is working? ;)”

    Although some would disagree with me (us?) FCPX is designed to sell hardware. It eats resources . . . probably another reason why I believe Apple will replace the MacPro with powerful computer.

  • Craig Seeman

    June 2, 2012 at 3:58 pm

    [Mathieu Ghekiere] “I’m wondering if Red Giant can optimize their Magic Bullet Suite more to speed up the render times.”

    Saphire Edge also needs a lot. If you really want to push things try using the Neat Video noise reduction filter. This on my 2008 Dual Quad MacPro. Things seems much more zippy on my 2011 15″ MBP. The desire for a MacPro replacement is palpable. FCPX, once you start heavy lifting with some filter, really demands it.

  • Oliver Peters

    June 2, 2012 at 4:19 pm

    [Mathieu Ghekiere] “I’m surprised the MB looks filter still needs approximately 4 times real-time export even on your fastest machine. I haven’t run the numbers, but I’m wondering if Red Giant can optimize their Magic Bullet Suite more to speed up the render times.”

    Playback performance IS better with PProCS6, but I haven’t yet done a direct export or render comparison. Maybe tonight. In the past when I did direct render comparisons on the MP between the various NLEs, X was always the slowest.

    – Oliver

    Oliver Peters Post Production Services, LLC
    Orlando, FL
    http://www.oliverpeters.com

  • Craig Seeman

    June 2, 2012 at 4:25 pm

    Dates of the models?
    2011 MacPro? 2012 iMac? I assume you mean 2010 MacPro and 2011 iMac. This based on when released.

  • Oliver Peters

    June 2, 2012 at 5:35 pm

    [Craig Seeman] “Dates of the models?”

    These aren’t my machines, so I was going by when they were purchased from Apple by the owners.

    Oliver

    Oliver Peters Post Production Services, LLC
    Orlando, FL
    http://www.oliverpeters.com

  • Craig Seeman

    June 2, 2012 at 6:05 pm

    [Oliver Peters] “These aren’t my machines, so I was going by when they were purchased from Apple by the owners.”

    Although it’s easy to determine the actual model year from the specs. Based on date of purchase it might be confusing to call a MacPro purchased currently as “2012”

Page 1 of 6

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy