Apple FCP and Compressor vs. Adobe CC, Timeline-to-Transcode Workflows, Part 1: Methodology

Standalone File Tests

For these last two tests, we wanted to see what would happen if we used Adobe Media Encoder and Apple Compressor as standalone transcoding tools. So we devised two more, relatively simplistic tests, from which to test a baseline of a single finished file being transcoded in each tool.

The file we’ve chosen is the longform, multicam content with timecode burn though mention in Test 3, above.

Test 4: Single File, Single Transcode

In this test, we dropped the finished clip into a watch folder. With one parameter or profile, this yields one output file.

Test 5: Single File, Three Transcodes

In this test, we dropped the finished clip into a watch folder. The output uses three separate parameters or profiles, meaning three output files, as a way to see whether any additional performance gains are yielded by the internal queuing mechanisms.

The Machines

On each of our two test machines, we started from a completely new user (“BraintrustTest”) so as to eliminate potential System Preferences persistence or any issues surrounding other applications running in the background.

Speaking of hardware, we also decided to throw ourselves a hardware challenge and do the tests on today’s “average” MacBook Pro and MacBook Air models.

The MacBook Pro we chose (ME249LL/A) is a 2013-era laptop with a 2.3 GHz Intel Core-i7 (quad-core) processor and 512 GB of PCIe-based flash storage drive. In addition, this MacBook Pro has 16GB of 1.35v (low-voltage or DDR3L) RAM running at 1600MHz and an NVIDIA GeForce GT 750M with 2GB of GDDR5 memory. It’s priced at approximately $2,540 at Amazon. [http://amzn.to/JDCvbW]

This flash-based storage, connected directly to the PCIe bus rather than via M-SATA or the slower SATA, means that we’ll eliminate any bottlenecks in disk I/O for our testing.

The only other two Macs to have the PCIe-based flash storage drive are the new Mac Pro, which we will talk about below, and the 2013-era MacBook Air laptops.

So to make it a fair comparison, we chose the MacBook Air 11” laptop (MD712LL/A) with a 1.3 Ghz Intel Core-i5 (dual-core) processor and 256 GB of PCIe-based flash storage drive. This MacBook Air has 4 GB of 1.5v (DDR3) RAM running at 1600MHz and an integrated GPU called the Intel HD Graphics which uses 1 GB of built-in memory. It’s priced at approximately $1,140 at Amazon. [http://amzn.to/1eOnuA6]

Why laptops?

As with any methodology, there are bound to be questions. One we can already anticipate is, why did we choose laptops instead of a desktop machine?

When Jan and Tim both did their respective tests, laptops didn’t really have the horsepower in terms of central processing unit (CPU) and graphics processing unit (GPU) to provide a fair comparison against desktop machines or server-grade solutions.

We still don’t think it’s a fair fight between laptops and servers, but what has changed is the ability to test both discrete and integrated GPUs in a laptop form factor. In addition, laptops now have robust i5- and i7 CPUs, with the latter offering quad-core performance.

Jan did his tests in 2012 on a 2009-era Mac Pro, which used dual 2.93 GHz Quad-Core Xeon processors and 12 GB of RAM. He also used an NVIDIA Quadro FX 4800 graphics card with 1.5 GB of onboard RAM, which had CUDA-based instruction sets. The machine ran on the Mac OS X 10.7.4 operating system.

We could go to a new Mac Pro, that shiny trash-can-like cylinder, but we felt it best to stick with something that might be used in the field. Plus we’ve heard that the Crossfire linkage between the dual AMD-based GPUs in the Mac Pro does not yet work well in Mac OS X 10.9 (Mavericks) although it does work in Boot Camp using Windows on the same Mac Pro desktop computer.

So laptops it is, for this snapshot in time, although we’d like to run the same tests on a Mac Pro if anyone can spare one for a few weeks.

Now that we've outlined our methodology and rationale, check back next week for the results in part two.