Quick note: This post (and this entire blog) is targeted at an audience of post-production pros. Within that, this post is targeted more at Mac users considering their options than those already on Windows. For me, this decision has been building behind the scenes for awhile, but I’m starting to get more serious about it. Also, please note that everything in the post below is my present opinion and outlook. There are factors I may not have considered and I have much to learn about the road ahead. If I’m wrong about something, break it to me gently, and I’ll update the post.
Current state of affairs in Mac post land: I don’t know about you, but I’ve grown weary of Apple’s lack of attention to the needs of production professionals (on the hardware side…the software side is still an open question-sort of-but let’s not open an FCP-X debate in a post about hardware). It should come as news to no one that video post-production has ever-increasing demands in computing power, and the greatest performance gains have been seen in recent years by modern software that leverages more CPU cores/threads, GPU, more RAM, fast bulk video storage, and even faster, separate, storage caching techniques. Of course, having software that can take full advantage of that hardware grunt is the other required component. I will, however, concede that while we have reached the point where basic video editing is now possible on lower-performing hardware, a full-on, high-performance, high-productivity soup-to-nuts post production system is the goal here.
Where I think Apple is going: Despite Tim Cook’s carrot-dangling (which probably references Intel Haswell), I am not convinced whatsoever that Apple is going to be building machines that support multiple GPUs, storage, etc. in a high performance configuration. I think they have a different vision. I believe Apple is on a trend toward increasing miniaturization and modularization of computing hardware. This is a mass market approach, and perhaps what is right for Apple. The mass market has reached a point where more mainstream, modest hardware satisfies their needs. Apple knows this, and is leveraging Thunderbolt to get a lot of that stuff outside the box, where it is an optional accessory only for those that need it. That is a fine strategy and opens up exciting possibilities for me and other video professionals. I envision a number of exciting configuration options for specific production purposes, especially mobile ones, where the priority is portability over peak performance.
Where I get stuck: The point where all this falls apart for me (and other post pros I’ve had this conversation with) is my “Big System”, the main editing/animating/coloring/mastering suite. You know that place…a place where you happily trade tiny screens and mobility for a clean, comfortable space, big screens, big sound and high performance hardware…a place where you can sit and do client-driven work without issue. That kind of system is driven by powerful hardware. Its exact configuration can take many forms, but high capability and performance are the priorities in this environment. You want power, and the ability to handle all sorts of work. This is where I do most of my work, and certainly my BEST work.
A system like this- built today- would utilize two or more GPUs, video RAID, other storage, video capture/display card and GUI Display bandwidth. Thunderbolt can handle one or two of those functions, but in total, the sheer volume of storage and PCI-E bandwidth a high performance system is capable of using will present substantially higher bandwidth load than today’s Thunderbolt can support. This problem will exist for quite awhile: It will still be true when Thunderbolt moves to PCI-E 3.0 speeds, which could almost double data throughput (that’s better, but still not enough for all of the above). Beyond that, there are faster Thunderbolt implementations on the roadmap, but they are YEARS away, 100gbps end of decade. So, while I think we will get to that point someday, the reality of today precludes a big, highly capable system built entirely around Thunderbolt.
Now add to this the fact that Apple are woefully behind the times in building machines with high performance CPUs…both at the “enthusiast” level (where is the 6-core i7 machine?), and of course the enterprise level (where are Sandy Bridge-E Xeons?).
Quick point on iMac: I actually like the higher-end iMac as a nicely configured cutting and basic utility station. A nice package for a slightly premium price. A lot of contributions to a project can be made from a machine like this, and in larger facilities I could see some cutting stations being configured this way, but due to the limitations of hardware I’ve noted above, do not think the iMac can serve as the basis of a fully capable “Big System”.
Forcing my hand: So (on the Mac)…CPU grunt isn’t there, and bandwidth isn’t there to add all the GPUs and other necessary hardware to build a robust modern system. This situation has gone from bad to worse as the hardware and software marketplace are marching forward, further enlarging the delta between the power and capabilities of the Mac platform vs. the equivalent Windows-based counterparts. I have held on to my current Mac Pro tower longer than any other before it. Part of this delay was driven by leaner economic times, but also by lack of a compelling business case to purchase newer Mac hardware as the performance gains of the available options didn’t seem high enough to justify the cost of re-tooling. Meanwhile, I have absolutely maxed the upgrade possibilities of this system and am pretty much at the end of that road, needing to move forward soon before this thing totally ages out. I’ll just come right out and say it: For a post professional in my situation (I know a lot of them), there is simply no compelling Macintosh option to move forward that would provide that substantial bang-for-buck performance boost that is required of any major upgrade. Those options DO exist in hardware, you just can’t run a supported Mac OS configuration on it. You can, however, run a supported Windows 7 platform on it. I’m setting aside the Hackintosh issue right now, because, while people get it to work, I’m not comfortable putting my production business in that precarious support position.
Me: mac-centric, but not a complete stranger to Windows: Having some familiarity with the Windows platform, and already experienced in building several custom PCs, I’m not exactly a stranger to Windows. Having said that, I have long-preferred the Mac platform for production work, and I vowed to stay as long as Macintosh computing hardware kept reasonable pace with the rest of the world. Look around my office and amongst the 14 computers and devices in daily use, 12 are from Apple. With one exception, all primary post-production since I opened the doors on my own facility in 1997 has been done on the Mac.
For those curious about the exception, it was in the era between the Power Mac G4 and G5 where Apple fell way behind on hardware performance…meanwhile the Pentium 4 surged. At that time I moved all of my After Effects and video encoding to Windows boxes due to sheer performance advantages, while continuing to edit and master all programs on Macs. The extra hassle of using PC-friendly codecs and moving data between systems was more than offset by computing grunt, but it was a pain. When the G5 was released, performance on Mac returned to levels where staying on the one platform was preferable.
As you probably have already gleaned from reading the above, I have absolutely no patience for platform wars and fanboy-ism from any side. I have my preferences, but I’m a businessman. Being a good businessman requires pragmatism, not platform ideology. These tools are a means to an end. I must satisfy both the creative and technical requirements of my client’s projects. Performance is critical in my world. This is not just about faster render and encoding times. Performance impacts the creative phase of a project substantially more than the tangential benefit of speeding up the creation of deliverables. I have determined that the personal pain of getting comfortable doing post production on a new platform is a lesser evil than the severely limited options the Mac platform currently presents. That has been a huge hurdle for me, one that I’ve resisted a long time and still have a great deal of apprehension about, but at this point I feel my hand is being forced. I know that I am not alone in this, and many will come to this realization after me, so I hope my forthcoming posts on this topic provide value as I work through this process.
Support: While the biggest factor in wanting to stay on the Mac is familiarity, I think it also goes to the issue of support. The more limited choices in the Mac hardware space may have kept prices higher and compromised a few high spec options, but this was offset by generally better support from vendors as well as more common system configs within the user community. Supporting the Mac is generally a simpler affair due to these factors. That’s my humble opinion of course, spoken as a career-long independent operator. The driver of that opinion is this: The tremendous choice in configuration options in the Windows world is both a blessing and a curse. Generally, more options and more competitive pricing are good things, but this typically results in a given users system configuration being totally unique. This complicates both user-to-user and vendor tech support as there are more variables when troubleshooting.
Applications drive hardware choices: This part is relatively easy to figure out. While I run about 20-30 production applications, I have my core post production activities: Editing, Motion Graphics and Color. Plus encoding…more than ever, I deliver encoded files for broadcast and web. Also, I drive an external SDI Reference monitor and also a plasma client monitor.
Editing: I use Adobe Premiere Pro CS6, Avid Symphony , FCP 7 and X. Of those, Adobe and Avid are on Windows. Adobe Premiere is my primary editor as of this writing. It makes very high utilization of CPU threads for codecs while GPU handles image and effects processing in the course of playing back clips on the timeline. CUDA preferred over OpenCL currently. Fast bulk RAID storage (spinning disks are still the only affordable option for large amounts of video data) . A quick note about Smoke, just as it’s making it’s big push as a mainstream product, is sadly Mac only. This fact is currently limiting the amount of attention I am dedicating to this product in light of my impending platform shift.
Motion Graphics: Adobe After Effects CS6. I use lots and lots of plugins. GenArts, Red Giant, Trapcode, Digieffects, Digital Anarchy, you name it. CPU utilization high, GPU less so, but Ray-traced rendering is CUDA-powered, and many plugins OpenGL, CUDA and other GPU processing. After Effects now has a great new capability in it’s Global Performance Cache. This is best directed at a very fast disk such as an SSD or better yet a Fusion IO device. For max performance it should be a fast disk that is separate from your video RAID containing your source media. I also use Adobe Photoshop and Illustrator extensively in the creation of these elements, but these apps and many other supporting apps are not primary drivers of hardware performance decisions. Pretty much everything in the AE universe is equivalent on Mac and Windows.
Color: DaVinci Resolve 9. With the exception of in-app plugin favorites like Magic Bullet Looks and Colorista, Davinci Resolve is my color grading application of choice. I am watching the progress of Speedgrade with interest-and for that matter other systems too-but Resolve continues to rock my color grading world at this point in time. Resolve uses a combination of CPU and GPU performance but skews more heavily to GPU. For a nice Resolve setup you are really looking at a minimum of two GPU cards (OpenCL supported but less performance). As with editing and pretty much everything else, Resolve leverages fast video RAID storage for read and write operations. Resolve is available on Windows as well, but I still need to research how well it performs on Windows, and any platform differences.
Encoding: This is typically all CPU, the more threads, the more clock speed, the faster the encode at any given spec. Fast video storage plays a role too. As far as I know, not many encoding options currently leverage GPU, although when you export from within Premiere Pro, any effects that leverage CUDA on playback will also be CUDA-accelerated during the encode, but the primary encoding function is all multi-threaded CPU work.
Pulling it all together: This is a starting point of sorts, a certain set of priorities that will drive decisions moving forward. I am open to configured systems as well as a roll-my-own approach. I have a ton of research to do in selecting components, as this applies even to pre-configured systems to a point. Bang for buck is a huge factor as I’m simply a business owner with limited resources and need to maximize ROI and that includes upgrade potential.
But all of that is hardware talk. Hardware does nothing without software. I know I’m good on primary application support, but need to find equivalent procedures and apps for all the little utilities and bits of software that are so essential to making my Mac post production business hum. There is a lot of work to do on this front. I plan on seeking advice here on the blog, the IMUG list and on Twitter as well as sharing what I find as I move forward. In fact, I’m not going to pull the trigger on the hardware until I have reasonable confidence that I have all the software bits sorted.
You read that right…at the end of this, the possibility remains that I will choose to remain on the Mac platform. If so-at that point-I’ll certainly have my reasons! This is going to be quite a journey.
I really hope you’ve stayed with me to this point, and have something to say or contribute. Leave a comment and let me know what’s on your mind…