[Scott Reynolds]

Here’s an update.

We’ve been looking at memory consumption by plug-ins. Some are clearly better behaved than others. Here is the breakdown. Draw your own conclusions.

Cheers,
Scott

FCPX Plug-in Memory Analysis
———————————–

Setup:
Mac Pro 5,1 (Mid-2010)
24GB DDR3 ECC SDRAM
ATI 5870 1GB
500GB SSD (System drive)

FCPX 10.2.2
Library: 6.5TB

Plug-in Used1 [1] Used2 [2] Memory Compressed Time [3]
GB GB min:sx
None NA NA 14.68 0 1:50
Basic 3rd [4] * * 14.75 0 2:00
PROStar * * 14.75 0 2:00
mLooks/mLooks2 * * 17.00 3.5 2:10
CoreMelt Free * * 17.02 3.55 2:10
CoreMelt TrackX 17.02 3.55 2:10
mFlare * 21.15 9.08 2:39
mGlitch * * 21.21 9.83 2:45
FxFactory [5] * * 21.36 10.46 2:59
BCC [6] * * 21.39 10.19 3:00


Notes:
[1]: Plug-in used in currently loaded project
[2]: Plug-in used in project in library
[3]: Time from application launch to scrubbing through footage
[4]: Basic 3rd party package includes:
– Alex4D
– neatVideo
– ColorFinale
– MyFCPEffects
– CineFlare Free
– XEffects Free
[5]: Includes 16 purchased plug-ins
[6]: Only Lights FxPlug package

[Scott Reynolds]

Noah, I appreciate the perspective.

Maybe the machine is not bright and shiny and new, but it’s used week in and week out cutting episodic with Media Composer, also with terabytes of footage with no real performance issues – nothing like we’re seeing here. So I don’t believe everything we’re seeing can be chalked up to an old, slow machine.

We’ve thought about an upgrade, but have a large investment in PCI-e cards, and not a fan of the new Mac Pro architecture. Getting an external PCI cage connected by firewire, um sorry, USB, oopps, I meant USB3, oh gees, I mean Thunderbolts, seems a little inelegant and adds to the cost. I had a Mac Cube, not sure whether I’m ready to jump on in with a trashcan. Either way, there needs to be a business case made for an upgrade, and we’re not there yet.

And all this actually ignores the point that the problem we’re seeing a memory consumption issue, not a CPU/GPU performance, memory or IO *bandwidth* issue. With a memory leak a new machine would simply run out of memory faster!

So, my original question still stands, are there any tools available that allows me to profile the memory consumption of a library, event and/or project in FCPX? Are there any tools that would allow me to validate the internal consistency and/or validity of the library? I’d love to know where more than 40G+ is going in a 22 minute timeline.

Cheers,
Scott

[Scott Reynolds]

Noah,

That’s what I ended up doing. 3 chunks, each exported as ProRes422 HQ, re-imported, concatenated and then re-exported.

The system is a mid-2010 MacPro (5,1). 6-core 3.33GHz, 24GB RAM, ATI HD 5870 graphics card. It should have enough beef to handle a 22 minute timeline.

The footage is all 1080p, 29.97, ProRes422 (it’s transcoded before I get it).

Cheers,
Scott

[Scott Reynolds]

Jeremy, thanks for sticking at this this. Yes, this was a fairly obvious step that I neglected to include in my original description. Sorry. I have been bit by that before, so it was checked and enabled in the Window pull-down menu (A/V Output).

However, in a fit of frustration and grasping of straws, I’ve just re-installed OS X Yosemite. As part of this, I seemed to have picked up an update to the Apple Professional Codecs, plus I had to re-install the BMD Desktop Video Utility once again.

Now it’s working. I have no idea why, as nothing “changed” per-se. I’m not sure which one or combination of these actions was responsible, but the answer is it’s back up and running.I suppose I just have to chalk it up to “spooky action at a distance”, but I’m going to be on pins and needles for a little while, that’s for sure.

Thanks again for your suggestions Jeremy. It’s so valuable to have another set of eyes saying “Well have you checked this? done that?”. Much appreciated.

[Scott Reynolds]

yep. it’s shown (and selected) with the correct properties (@29.97).

It’s a weird one.

[Scott Reynolds]

Thanks for the reply and suggestion, but I have indeed done that – multiple times, with multiple different driver sets (10.4, 10.4.1 and 10.4.2).

Just to reiterate the BMD’s own application (Black Magic Media Express) and Final Cut 7 correctly use the installed hw and drivers, and correctly display images on the A/V screen. The problem is limited to FCPX using the display.

[Scott Reynolds]

Thanks to all for your comments. Valuable tips and much appreciated.

Scott

[Scott Reynolds]

With some amazing detective work by T. Payton, there is a work around for this problem. The first step in the work around is to synchronize all the externally recorded audio tracks into a single clip. The second stage is simply sync’ing the original video clip, including the guide track and the sync’d clip created in step 1. The resulting clip includes all the audio – both guide and externally recorded tracks and their references in the audio pane of the inspector.

Again, thanks to everyone who contributed suggestions, and especially T. Payton who did the heavy lifting investigation into the issue.

Regards,
Scott

[Scott Reynolds]

Not included at all unfortunately.

[Scott Reynolds]

Here’s a couple of screen grabs.

This one shows the video clip with the stereo guide track.

This one shows the multi-mono channel audio, just before I syncrhoized.

Here, I created the sync’d clip.

And here are the audio properties of the sync’d clip

As you can see channels 2 and 3 were NOT added to the sync’d clip.

I have tried multi-cam clips and they do work OK, and a few posters have pointed out some advantages of going this route, so I might bite that off and see how it works.

Doing this project in FCP X is more of an experiment on my part, after reading some encouraging posts indicating that the product has matured sufficiently to be considered for more advanced work. I must admit, I’m still a bit skeptical, but I’ll keep my mind open.

Thanks everyone.
Cheers,
Scott