Adobe Premiere CS6 vs. Apple Final Cut Pro X speed test
The 10.0.6 has made Final Cut Pro X noticeably faster at everyday tasks.
Given that, I was interested in seeing how it now compare with Premiere CS6.
This video times two everyday tasks in Adobe Premiere CS6 6.0.0 and Final Cut Pro X 10.0.6:
For over 50 free plugins and more for Final Cut Pro X, visit my FCPX home page.
The source clip is a 1 minute 1080p25 ProRes 422(HQ) movie.
The first task is to render a Gaussian Blur effect on the timeline. The second task is to export a timeline without rendering first.
I’m much more used to using Final Cut, so if I’ve not given Premiere a fair chance, please comment below.
For regular Final Cut Pro X information, follow me on Twitter.
This was on an 3.4 GHz i7 iMac with 16GB of RAM and 2MB Radeon 6970M. The apps were running from an SSD, the files were on a 7200rpm drive. If you have a very different kind of Mac (or a PC), you’ll get different results. Also the brand of video card makes a difference. However, if you have the same kind of Mac as this, you’ll probably get similar results.
Perhaps others can do the same test on their computers to see the difference, it’s a matter of exporting 1500 frames (1 minute at 25fps, 50 seconds at 30 fps) of 1080p ProRes 422(HQ) movie and importing it into Final Cut Pro or Premiere. Maybe alternate video card architectures will make a big difference.
Premiere 6.0.2 update
After making the video I updated my copy of Adobe Premiere to 6.0.2 and redid the tests three times. The update seemed to make a difference.
Timeline render: The times were 80, 61 and 62 seconds (compared with 75 seconds in the video)
Export: The times were 109, 100 and 101 seconds (compared with 117 in the video)
Have you changed any of the original settings for Premiere and Final Cut X? Or are they just as they would be after initial install? You may need to do Premiere at its best settings and Final Cut X at its best setting for a fair test (settings for the entire program, ranging from its use of the computer’s memory to how it handles exports.)
I was using the default application setup for Premiere – I didn’t modify memory or disc settings. There aren’t such settings in Final Cut Pro X. I used the 1080p DSLR sequence settings.
Hey Alex, I love the test. I was wondering if there is any chance you would run the same test on FCP7 just for a baseline comparison between the two.
Final Cut Pro 7.0.3
Some FCP7 effects were implemented using interpreted FxScript. Gaussian Blur is an FxPlug effect that uses the GPU.
Timeline render: 111 seconds (vs 43. seconds for 10.0.6)
Export: 105 seconds (vs. 37 for 10.0.6)
And a Radeon isn’t really fair to Premiere Pro, since it needs an NVIDIA card with CUDA to really strut it’s stuff, or to even show off the Mercury Playback engine.
I’ve heard CUDA makes a big difference.
This test shows the result on one Mac with one kind of video card. However if you have a Radeon card in your Mac, this test is relevant. If you don’t have a production Mac for editing, then you should test other cards as well as the Radeon I have.
Speed difference with Premiere and a CUDA card and software is huge though. Such a difference I would never consider using Premiere Pro without one.
Very interesting test – So much for all the hype around Premiere .. that said I am sure there is more to a difference betweeen these two apps in the real world than the speed of export and render ..
Without using a CUDA supported NVIDIA card you are really robbing Premiere Pro.
interesting … thanks for sharing
so yeah indeed PremierePro Export time can be ridiculously long if u have openCL stuff going on (like MBL) but should be pretty quick if u have a Cuda powers System and i’d might add it got faster with 6.0.2 on MacOSX 10.8.2 with an Nvidia Card (like a GTX480 or GTX570 or Quadro)
so my question is what machine / GPU / OS Version did you do your test on?
I was one person who didn’t jump ship to Adobe or Avid. I decided to stick it out with FCP X as I knew Apple would be adding all the missing features in time. I’m glad I did as the program is becoming a high end product. Apple made a good choice by changing how we edit. This is the future. Great Test Alex since I edit on both a 27″ iMac that is set up almost identical to you. I also edit on a MacPro. Alex, you do great stuff!!
I too stuck with FCPX and am really glad I did. Since 10.0.3 it has been good and with the later updates it has become really good. I edit on a MacPro and MacBook Pro.
Dude, any test like this NOT done on a CUDA is not worth anyone’s time, since any sensible PPro editor will have one. And once you go CUDA things look just a stitch different my friend:
1min Apple ProRes Quicktime (non HQ)
17″ MacBook Pro, OS X 10.8.2, 2.66GHz Intel Core i7, 8 GB 1067 MHz DDR3 RAM, NVIDIA GeForce GT 330M 512 MB with CUDA v5.0.36
Premiere Pro CS6 v6.0.0
FCPX Render 120s
PP6 Render *94s*
FCPX Export 77s
PP6 Export *64s*
Indeed, this test is nor fair at all, should do it on a MacBook Pro Retina Full options 16 Go with Nvida 1 Go, since CS6.0.2 is CUDA accelerated with the 650M, the results will lokk very very different…
Pingback: Final Cut Sunday! | Miguel Cobo
i tried to replicate your test here on my hardware, going with CUDA for Premiere. i have hackintosh:
i7 920 – 3.8Ghz
6 GB RAM
Nvidia 560 Ti
test made on 10.7.4!
PPro CS6 v6.0.0
Gaussian Blur Export ProresHQ:
PPro – 0:49
FCPX – 0:22
to try something else, apart from Prores Codec
Gaussian Blur Export H264 (960x540p25 – “better quality” settings in fcpx):
PPro – 1:00
FCPX – 1:06
One thing to note here can be the quality of encodes, with blur there is not much to compare.
Personally i think (tried many tests for various apps) that CUDA downscaling in PPro (especially for interlaced footage) cannot be matched in quality in any other NLE. Its pretty fast too! FCPX comes very blurry compared to PPro on this task. Therefore, CUDA gives “sometimes” access to better algorithms for certain tasks, so possibly better quality.
PS: I tried activating OpenCL in Premiere and that gave me 4:50 on ProresHQ export (you have to tick Maximum render quality to use CUDA/OpenCL). Might be different on AMD though, because Nvidia doesnt like OpenCL as much as CUDA.
I think this is a totally fair test! A lot of iMacs have been deployed with Radeon chipsets and they make fantastic editorial machines, despite their lack of the ability to swap cards. What Alex is doing here is showing a comparison in speed on a VERY POPULAR machine. If you don’t have this particular machine, then ignore this test. But for the many of us who have bought and deployed iMacs in a professional capacity over the past few years, this was a good comparison.
Oh puh-leeeeeze… don’t be a pathetic apologist. This is not titled “Adobe Premiere CS6 vs. Apple Final Cut Pro X speed test FOR PEOPLE THAT HAVE THE LEAST OPTIMAL HARDWARE”. If you’re going to do a test, you do it “REAL WORLD” to get the most representative results. Especially if you’re running an *FCP X* site to at minimum avoid being suspected of bias and feigning results to your personal favor (which I’m sure Alex didn’t, but actually meant well). And fact is: professional Premiere users will be using NVidia hardware to take advantage of the Mercury Engine, since that is one of it’s FEW advantages. Period. So any other results are moot.
Don’t get me wrong, I absolutely LOVE X and despise PPro for editing! I just see the smart-ass kiddies without a clue roaming the forums now pointing to this article with a nyeh nyeh nyeh attitude, whilst making both themselves and FCP X users as a whole look like morons. Since this article suggests that X is FASTER than PPro, where anyone WITH a clue will recognize that it says nothing. So which one are you then?
Thanks dm, but I think that this post does say something: not ‘Final Cut Pro X is faster than Premiere CS6 on most Macs’. It says that on the best iMac available in the summer of 2011 – with the inherent configuration limitations that Apple imposed then – Final Cut Pro X has at least caught up with Premiere. That’s why I did my amateur test.
In practice the philosophical difference between the applications is much more important than their speed. During day to day editing tasks, all NLEs are fast enough on almost any computer. You can test this point by imagining how much of a speed difference would be needed for you to switch. I think the speed difference would have to be at least 20 times for me, which is probably the same for those happy with Premiere.
Did you enable GPU rendering, as seen in the video by Kanen Flowers?
Mercury Playback on a 27″ iMac with AMD Radeon 6970M (Premiere Pro CS6)
You may even skip that process and just take out the GPU supported cards list before starting Premiere.
It seems only fair to test GPU enabled filters on both apps, if you want to compare apples with apples.
Also, you could use modern codecs, like one of the ones a Sony, or Panasonic camera would output. QuickTime is pretty passe now.
“QuickTime is pretty passe now.”
LOL!! On WHICH planet exactly??! You have no clue what you’re going on about… ouch!
Apple is not developing QuickTime any more. And which cameras shoot QuickTime? I’d like to see a test with AVCHD or P2. Just saying.
When you say P2 are you referring to the card? Or the file type? Because there is no P2 file type. All P2 cameras that record on to P2 Cards record in an MXF file format. But those MXF files are recorded using a wide array of video codecs such as AVC-Intra, DVCPRO HD, DVCPRO 50 or 25. And AVCHD is just h.264 with a 25Mbps data rate limit.
ProRes is still being developed and is an industry standard deliverable format.
Yepp, like I said: NO CLUE (what Quicktime even IS).
Oh, and and cameras that “shoot Quicktime”?? LOL! Are you shittin’ me?? You mean cameras like THE ALEXA or pretty much ANY AND EVERY DSLR??! Is THE DIGITAL CINEMA enough camera for you??
Dude, do yourself a favor and just STFU.
All theses little speed tests are irrelevant, the face that you dont have to transcode DSLR footage with premier, already saves you so much time and not to mention file space. I really cant see a reason to switch back to FCP.
Wow. You actually are moron enough not to know that transcoding in FCP X is
a) completely OPTIONAL
b) happens in the BACKGROUND so why should I even give a flying rat’s ass?
c) only clueless wedding and skateboard video makers are dumb enough to think that editing H.264 or ANY long-GOP format natively is somehow a GOOD thing.
Wear your NOOBISHNESS on your sleeve much?? Go back to trolling YouTube and pirating the newest release of Premiere, won’t you?
dudemeister, your horrible attitude isn’t going to win you any friends… or people wanting to work with you in the long run. Might want to think about the long-term implications of flaming people online. Your words will follow you for decades to come. I’m sure you have more than enough stuff to keep you busy today, and don’t care what anybody thinks… but one day you will.
LOL… yeah, sure. Whatever, DAD.
Since my real name is of course “dudemeister” and everyone I work with knows I posted here it is a SURE THING that it will come back to haunt me… mmmh-hmmm… sure. Gotcha.
New to this “Interwebs” thing, eh? :-)))))
Besides, I have no problem standing behind calling people on their BS. If you want to call me on some actual FACTS, feel free. Otherwise I don’t need any holier-than-thou-morality-etiquette lectures from you, thanks. Don’t like what or how others write? Don’t read it or stay off the internet. Problem solved.
You don’t have to like it – in fact you don’t have to think anything about it!
Any of our opinions against the use of mixed-format editing are fast becoming moot.
Whether you like it or not, mixed CODEC editing is here to stay. Camera companies will continue to refine and expand the use of them whilst adding new ones.
H.265 will be the next in our suites: http://en.wikipedia.org/wiki/High_Efficiency_Video_Coding
So having an NLE (or FX package) that will handle all of them natively on the same timeline is an absolute must unless you want to spend your entire post-life transcoding before you edit.
I read the article with some interest so decided to run a couple of tests myself on one of my systems.
Some here are the results of my tests for FCP 7.0.3 vs FCPX 10.0.7 vs PP6 6.0.2
MacPro 8 Core 3GHz Xeon 32GB RAM 1GB GTX285
Mac OS 10.8.2
CUDA Driver Version: 5.0.37
GPU Driver Version: 8.0.61 295.30.20f02
CUDA/Mercury enabled on PP6
Read/Write to 800MBps SAS RAID 6
Test sequence is a Mixed CODEC edit that is a total of 98 seconds long:
1. ProRes 422 HQ 1920x1080i50 48kHz Linear PCM Stereo (also the sequence setting)
2. 1920x1080i50 XDCAM HD 50 422 48kHz Linear PCM Stereo
3. ProRes 422 HQ 1280x720p25 48kHz Linear PCM Stereo
4. H.264 1280x720p23.98 AAC 48kHz 5.1 Surround Sound
5. H.264 1920x1080i50 48kHz Linear PCM Stereo (Canon 5D Mk2 native)
6. Animation CODEC 1920x1080p25 48kHz Linear PCM Stereo
7. Apple DV PAL 720x576i50 16:9 FHA 48kHz Linear PCM Stereo
FCP7 Mixed CODECs export to ProRes 422 HQ
FCPX Mixed CODECs export to ProRes 422 HQ
(~30 seconds to export and transfer from FCP7 XML via 7toX and load in FCPX)
56 seconds exporting
PP6 Mixed CODECs export to ProRes 422 HQ
(~101 seconds just to import FCP7 XML into PP6)
120 seconds conforming the MP4 before the export would start
304 seconds exporting
Total: 424 seconds
Basically the Mercury playback helps when editing and viewing (color correction etc) but Adobe Media Encoder (AME) which PP6 uses to export does not it seem leverage the GPU as far as I’m aware. Certainly pre-rendering the PP6 made no difference whatsoever to the export time only added render-time on top!
From Adobe literature:
• GPU processing (Mercury Playback Engine GPU Acceleration) extended to use OpenCL on some systems: AMD Radeon HD 6750M and AMD Radeon HD 6770M graphics card with 1GB VRAM in MacBook Pro computers running Mac OSX v10.7
• use of GPU acceleration by additional effects and transitions, including Warp Stabilizer effect and Push transition
• Mercury Transmit, a new interface making it easier for third-party I/O hardware manufacturers to interface with Premiere Pro, allowing for more robust and stable systems”
There was/is no mention of using the GPU for AME…
I also decided to repeat Alex’s test using the Gaussian Blur Export.
Test file: ProRes 422 HQ 1920x1080i50 – 1 minute long with 30px Gaussian Blur
PP6 ProRes 422 HQ export with 30px Gaussian Blur
FCPX ProRes 422 HQ export with 30px Gaussian Blur
I’ll repeat with a GTX580 on the next tests – I imagine CUDA does not improve PP6 over FCPX when you have a faster, higher spec Nvidia card that also boosts OpenCL on FCPX.
However using a lower spec Nvidia card such as the GeForce GT 330M it seems (from some of your posts) to detrimentally affect FCPXs performance yet CUDA shines.
It goes to show it’s ‘horses for courses’ and if you run a lower spec Mac/PC then you might find PP6 is actually faster than FCPX whereas others do not.
I will have to test on the MBP Retina when I get a chance and I would love to test both on a 6GB GeForce GTX Titan in a 12 core and see how that stacks up somehow I’m not sure I’ll get a chance for the latter anytime soon!
I use PP CS6 with CUDA… it really shines in some parts as editing and running real time effects but even though right now I much prefer FCPX. The 10.0.8 version is very stable, and that is precisely the problem with PP CS6, it’s got very annoying behaviors, it hangs randomly sometimes or the whole Mac hangs or when you start ADM than it can also hang or exporting and preparing renders takes for ages… Adobe still has to work in solving all those bugs because they ca be very frustrating and time consuming.
FCPX now is really a very stable system and in general when editing in FCPX everything works faster and smoother than in PP CS6 even with CUDA. Exporting renders in PP CS6 can be a nightmare, as it was for me.
Once you get through the new FCPX Timeline editing paradigm, thinks can really go very fast as everything is so well integrated.
And with the new Adobe’s subscription price policy FCPX can make a big difference when talking about costs, specially if you have several stations.
Relating to the title of this article, it is so, CUDA plays no roll when exporting your videos… and FCPX is clearly faster. The time you may win in PP with CUDA is the time you loose but multiplied when you start to encounter all those PP issues. Those issues btw are not suffered by all users, I guess it depends on your reels and your system. Some are very happy with PP and some not… I’ll try PP again in a few months when they update it again and see how it goes. But FCPX is getting better and better and is being updated very regularly with great new features.
To know major differences between All three leading applications – After Effects, Premiere Pro and Final Cut ProX, you shouldn’t miss this blog http://www.videoeditinginstitute.com/after-effects-vs-apple-fcp-vs-adobe-premiere-pro