-
The FCPX 10.1 is all but slow
Since the interwebz is full of whining babies and cat killing dooshbags I decided to do som real-world FCPX 10.1 tests versus the old version.
Testing hardware
iMac 2012 / i7 / GTX680MX / 32GB / Fusion Drive + TB SSDI used the latest version of OSX 10.9.1 for both FCPX 10.0.9 and 10.1.0
My tests
Rendering and exporting of two separate projects. The export (i.e. “Share”) was done with an unrendered project since FCPX does use system resources very differently when doing either or.Both projects are relatively effect-heavy time lines, where the project “Catinka” differs a bit in that I’ve used two “adjustment” type layers with FX from Motion for my final look.
Both projects also show a huge difference in how they tax the CPU in the machine. The “Catinka” project uses at most 20% of the CPU where the “Young” project during exports maxes it out to full load.
Result
FCPX 10.1 shows very impressive performance gains. My two projects got between 13% and 48% faster render or export-times. Clearly impressive.I’d also say the best workflow for FCPX is work in realtime, export when ready, render only in-app if required for playback.
Yes, I would have loved a few features in 10.1 that didn’t make the cut – batch export, more flexible GUI, role lanes, better audio-mixing, keyframable color correction, curves for color correction, multiple scopes and heaps more. But one thing is for sure – FCPX went from fast to faster and I think, sadly, Apples primary goal with FCPX at this point in time is to demonstrate the roaring speed of their new MacPro baby. This is something it does very very well and it seems everyone gains for that.
The library update was also a much needed feature. Organizations was a mess before, this should bring far more order and easy of project sharing between people.
FCPX 10.0.9 vs 10.1.0 Render and Export
CATINKA RENDER
CATINKA SHARE
YOUNG RENDER
YOUNG SHARE
-
11 Replies
-
Thanks for sharing. The performance increase in 10.1 is impressive. These complainers blow my mind, it’s an update focusing on performance and the new Mac Pro, for X sake!
I posted some of my testing results this this thread- Anybody HAPPY?
On my slower, older systems a free performance increase is more useful to me than a scrolling timeline or whatever.
Have you done Alex4D’s benchmark “BruceX”? I’d be interested in your results (even though FCPX 10.1 won’t export it to h.264, I went ahead and did it to Prores just to get some numbers). I posted my results in the thread below and asked for others results. Nobody is interested in these things I guess… https://forums.creativecow.net/thread/344/26062
BruceX-
https://blog.alex4d.com/2013/10/30/brucex-a-new-fcpx-benchmark/ -
‘But one thing is for sure – FCPX went from fast to faster and I think, sadly, Apples primary goal with FCPX at this point in time is to demonstrate the roaring speed of their new MacPro baby. This is something it does very very well and it seems everyone gains for that.’
This was on an iMac though right? So given this result are you more or less inclined to go for the tube?
-
Did the BruceX test now and got 80 seconds on average compared to the same spec machine on their site. I would however claim the BruceX test is a poor performance test.
a) It only taxes the GPU. From my two sequences it’s very clear some operations are GPU-bound and some CPU-bound.
b) It’s only based on generators and effects, i.e no i/o comes in to play as you would 90% of the case have running FCPX in real life.
c) Outputting to H264 is a very odd choice. No sane person works that way in post-production.
-
I was playing with the thought of a Tube at home yes but the cost probably outweighs the performance increase and I don’t do that heavy lifting work here normally. But it’s defiantly not off the table.
At the studio I’d imagine anyone who charges for FCPX-work will enjoy a MacPro and can easily motivate it. Are the required? Of course not but seeing the performance this machine get (with a blood mobile GPU!) I can just imagine the creative freedom a new MacPro has.
I by the way wonder if Apple in Mavericks can so some magic with the Core-based systems as well. Some models come with both a dGPU and iGPU, couldn’t they simply route the OpenCL-processing tot he dGPU and have the iGPU handle screen-redraws? Much like DaVinci-systems have done it for a few years?
-
But to follow up on your other comments I fully agree. One can live with out a scrolling timeline or a send to motion command. But if their system is so slow they can’t use it, well there is no work-around for that.
As boring as it is I think Apple is laying solid ground to FCPX. I’m very keen on learning more about the added functions to the XML-structure and share-API. Maybe third parties can do some clever magic to / from After Effects / Nuke / Mocha for FX work.
Scrolling in the timeline and event browser seems way smoother now. Much appreciated.
-
I think native FCP X stuff, generators, color board, transforms, etc. are all gnu. Things like slicex and other plugins may be cpu depending on their function. I thought I read something along those lines somewhere.
-
Not 100% sure thats true though. My “Young” project consists of ProRes 1080p-footage, various transfer modes and effects. During a share it maxes the CPU-usage.
My “Cotinka” project however uses very little CPU. I’d have to test it more but I have a theory there about the adjustment-layers with motion-created effects that covers the entire edit.
-
[Bret Williams] “I think native FCP X stuff, generators, color board, transforms, etc. are all gnu.”
I see what you did there.
-
I typed with my phone is what
-
My biggest issue with FCP X was with larger projects. Been working on a doc for a year, and it has well over three thousand clips. Scrolling the clips in the event, waking from sleep, at any moment it could choke up. After converting the events to 10.1, there were no more beachball moments. After putting it through the paces for a few hours I was amazed how smooth everything was running until I realized I no longer had the proxy setting selected after the update.
That made it go to eleven for me.
Reply to this Discussion! Login or Sign Up