Apple released the latest version of Compressor (4.4.5) last Monday. One of the key features in the update was improved performance. Since all of us rely on compression to deliver our final projects, I wanted to see how much improvement Apple made to this essential software.
Then, because Compressor speeds have often lagged the competition, I also decided to compare the compression speeds of the latest version of Compressor to the current version of Adobe Media Encoder
While both Compressor and AME support a wide variety of codecs, I tested two of the most popular delivery codecs: H.264 and HEVC.
The test movies used three different source codecs: XDCAM EX, ProRes 422 HQ and ProRes 4444.
Based on these findings, you’ll find Compressor performance for the new version essentially the same as the older version, while Adobe Media Encoder will be significantly faster compressing material for H.264.
You’ll find a link to a PDF containing all my timings in the Extra Credit section below.
NOTE: Apple’s release notes indicate that many of these performance enhancements are designed to optimize Compressor for the upcoming 2019 Mac Pro. As that system has not yet been released, I have no way to verify the performance difference.
Here’s the system I used for testing: a 2017 27″ iMac with an i5 CPU. This isn’t top of the line by any means, but it is typical of the systems of many of us.
NOTE: I first began testing compression speed in 2014. Here’s a link to my most recent test before this one.
I made it a point to use the same three test files for the last several years so that we could compare results. However, this year, I transitioned from a 2013 i7 iMac to an 2017 i5 iMac. While the CPU speed is faster, the i7 supports hyper-threading, while the i5 does not. So the speeds of this test can’t be directly compared to past years. (When it comes to compression, an i7 is significantly faster than an i5.)
One other note. I made an error in testing the earlier version of Compressor (4.4.4) in that I had other applications running while I was timing the test. Since Compressor actually uses a background process to compress media, the more applications that are running, the slower the background process will run.
I didn’t discover this error until after I upgraded to the new version, at which point it was no longer possible to go back. The result of this is that Compressor 4.4.4 reported slower times than it would if it were the only application running.
Keep in mind that in any real-world situation, you would have Compressor running along with a variety of other applications. So, the timings for Compressor 4.4.4 reflect real-world use, while Compressor 4.4.5 timings represent an ideal environment.
I compressed the same three files – XDCAM EX, ProRes 422 HQ and ProRes 4444 – into H.264, HEVC 8-bit and HEVC 10-bit formats. Compression timings were reported automatically by the software.
All files were stored on the internal Fusion drive of the iMac. All compressed files were stored back to the same drive. This prevented any delays caused by external storage.
Here are the settings I used for Compressor – all based on the default settings in Video Sharing Services > HD 720p. To change from H.264 to HEVC, I changed the Codec selection. Because HEVC 8-bit is hardware accelerated, I selected the Faster Encoder type to make sure that hardware acceleration was used. To change from 8-bit HEVC to 10-bit, I changed the Profile. I made no changes to bit rate or other settings.
The benefit to HEVC 10-bit is that it preserves the highlight detail in HDR media. All images were compressed to 1280 x 720.
These are the settings I used for AME – all based on the default settings in Social Media > YouTube 720p HD and HEVC > 720p. Both H.264 and HEVC 8-bit use hardware acceleration. However, the default bit rate for H.264 is double the rate of Compressor, which will create larger compressed files. While I tracked file size in the PDF they are not reflected in my charts, below.
The bit-rate for HEVC 8-bit is one-quarter the size of H.264, resulting in much smaller files.
To switch from HEVC 8-bit to HEVC 10-bit, change the Profile from Main to Main 10. This also means that compression is not hardware accelerated, resulting in much longer compression times. The benefit to HEVC 10-bit is that it preserves the highlight detail in HDR media.
All images were compressed to 1280 x 720.
Here’s the comparison between the two versions of Compressor. (Shorter bars are faster.) The new version is slower compressing the same files into H.264, slightly faster compressing files into HEVC 8-bit and clearly faster compressing SOME Files into HEVC 10-bit.
IMPORTANT NOTE: Both versions of Compressor failed to compress the ProRes 4444 test file after working on it for 2.5 hours. For this reason, I don’t recommend using Compressor for any HEVC 10-bit compression.
Comparing Compressor 4.4.5 to AME 2019 when compressing H.264, AME is faster across the board – almost 40% faster. (Shorter bars are faster.) The fastest results occur with XDCAM EX material, but even with ProRes 422 HQ, AME is 10% faster than Compressor.
With HEVC 8-bit, Compressor is 45% faster; and, remember, both Compressor and AME use hardware acceleration for 8-bit HEVC.
However, the change to HEVC 10-bit is striking. AME is about 70% faster and Compressor failed twice to compress a ProRes 4444 file, which AME compressed in 24 minutes.
The latest version of Apple Compressor is about the same speed as the prior version. It still has problems, like the earlier version, compressing HEVC 10-bit files.
If you are looking for faster performance, while there are exceptions, in general, AME will provide faster results.
8 Responses to Compression Speed Test: Apple Compressor vs. Adobe Media Encoder
“AME is about 70% faster and Compressor failed twice to compress a ProRes 4444 file, which AME compressed in 24 minutes.”……wow !
Thanks again for the detail work Larry
Yeah, I was pretty struck by that as well. Both Compressor 4.4.4 and 4.4.5 failed to compress the ProRes 4444 file. AME had no problems.
To be honest, I’m a little surprised with your findings. On my system (MacBook Pro 15” 2018, 6-core i7 2.6 GHz) I’ve witnessed at least a 22% performance boost on Compressor 4.4.5 than on Compressor 4.4.4. For example, what used to take around 30 min for export now finishes in around 23 min (I always export to 480p H.264 for Video Podcasting, from a UHD source material encoded with AVCHD). As for CPU clock speed, on Compressor 4.4.4 it was averaging a measly 2.0 GHz (due to thermal throttling I guess) which was way below the nominal clock speed and now on 4.4.5 it’s averaging at least 2.9 GHz and the temperature is hovering stably around 80 Degrees Celsius, which is, if you ask me, a pretty satisfactory performance boost compared to the previous version.
Do you think that the i7 CPU is behind this discrepancy in results? Maybe if you try the same tests on an i7 system it would yield different results?
Thanks for writing – I appreciate your contribution.
The short answer is, yes, an i7 makes that big a difference. The longer answer is that I did not realize the significant performance improvement an i7 makes on media handling when I bought this system. Going forward, I would not buy an i5 again, even though it is much less money.
I’m in the process of getting a system with an i7 and will rerun my tests using that as soon as I can.
I’d love to make such a comparison myself, only a little inconvenience that prevents me is that Compressor 4.4.4 fails to commence by showing me a nice warning “Restart your computer or verify that the installation is correct” with the option “Repair installation” which does nothing. It’s unusable on me.
Larry, are you aware of the actual fix for this problem that works?
I’ve run into this before. Give Apple Support a call.
I’m running Mac OS 10.15.4 with 3.6 GHz 8-Core Intel Core i9 and 664 GB RAM with a Radeon Pro Vega 48 with 8 GB
And I just crashed Adobe Media Encoder trying to compress an Apple Pro Res 4444 file with a modified Vimeo preset that is 1hour and 30 minutes. But yesterday it didn’t crash but took about an hour or so to compress. I did have it on render at maximum depth VBR 2 pass, Maximum Render quality. I’m on Encoder 14.0.4 (Build 16)
Now I’m trying Apple Compressor 4.4.6 on its Vimeo preset , also double pass (but I don”t have maximum depth type of controls that I know of). Its cutting through like better in just 25 minutes!
Thanks for your comment. Several notes:
* In AME, if you have a system that is more recent than 2016, don’t enable Maximum Render, it doesn’t apply.
* In both AME and Compressor, don’t select 2-pass. Select 1-pass, which enables hardware acceleration. Quality is better with hardware acceleration (1-pass).
Follow those two tips and your media should compress in less than 15 minutes.