In Part One of this review of the Scan 3XS PowerDAW Windows PC, I took a look at the makeup of this system and the reasons behind the choices made.
In this part I have run a series of real life performance comparisons, between the 3XS PowerDAW, my own "home build" Windows PC and my Mac Book Pro. I could have just stuck with the standard "DVerb" test, but I've found that this often bears little resemblence to how well a system can cope with real-life scenarios.
Here are the specs for each system -
Machine | Operating System | CPU | Cores / Threads | GPU | RAM | Thunderbolt |
---|---|---|---|---|---|---|
3XS PowerDAW 2017 | Win 10 Pro | i7 6900k 3.2GHz | 8 / 16 | NVidia GTX1050Ti 4GB | 32GB DDR4 | 1 x TB3 |
8dB Self Build 2015 | Win 10 Pro | i7 5820k 3.3GHz | 6 / 12 | NVidia GTX660Ti 2GB | 16GB DDR4 | 2 x TB2 |
Mac Book Pro 8.2 2011 | OS X 10.10.5 | i7 2.2GHz | 4 / 8 | AMD 6750 512MB | 16GB DDR3 | 1 x TB1 |
On first appearances, this may seem like a really unfair comparison, particularly for the Mac Book Pro. However, keep reading - there are areas where the operating system makes a world of difference. These tests reflect my day-to-day working environment. A growing trend among the industry is to have a "super system" back at base, but also a "mobile rig" to allow flexibility.
Test One - The Pro Tools Expert Stress Test
Finding inadequacies with the standard "DVerb" test, the team at Pro Tools Expert made their own "stress test" session. This consists of 200 VI tracks, each with the Avid Boom! plug-in, plus a further 128 audio tracks, each with three plug-ins on. For this test, my Mac Book Pro would not even play and took an age just to open the session, so I have just compared the Scan system against my existing setup
The obvious winner here is the Scan system, by quite some margin. My own self build system almost peaked out, indeed rising to almost 95%, but then it plateaued out at around 83%. Intriguingly it took longer to load up it's processing than the Scan PC, which went straight up to around 61% and stayed there. Having double the memory, six times the system drive access speed and four extra threads available makes a huge difference to being able to cope with VI's and large amounts of audio tracks. The more powerful GPU also makes a big difference with the scrolling smoothness - on my self build system both the edit screen and mix screen were almost unusable, whereas on the Scan system the smoothness of scrolling and metering made it look like this was effortless.
Test Two - System Usage While Playing A Busy 5.1 Feature Film Mix
I have my own standard post-production centric "stress test" Pro Tools session. It has around 200 tracks, lots of live noise reduction, EQ, dynamics and 5.1 reverbs and is running it all natively - no Avid or third-party DSP. This test measures the displayed system usage of the three systems. I have used the same section of the mix on all three, as in native, plug-in allocation is dynamic.
With this test it is more about how much headroom you have, to be able to cope with suddenly having to rack up the processing, or cope with a large amount of simultaneous track activity, such as in the sequence I used. With DSP the plugins and mixer are allocated when you open the session, so you get a good idea of whether the session will play. With Native Dynamic Allocation, you don't find out until you hit a busy section. So, it's always good to have as much processing headroom as possible.
On the MacBook Pro, we're looking at around a 60% peak, which doesn't leave much headroom. Indeed on this machine you quite often get processing spikes which bring the session to a halt. It is also pretty much impossible to have any other software running on this machine at the same time as Pro Tools. On my self build system we're looking at around 30%, but on the Scan system it romps home with around 16%, which leaves a massive amount of processing headroom. Things become really interesting when you add a pair of HDX cards into the system and attempt the same test. Come back for part three to see what happens.
Test Three - Time Taken To Render Audio Using zplane elastiqueAAX
In my review of the zplane elastiqueAAX pitch and time plug-in, I compared how long it took to AudioSuite render a piece of 5.1 audio, against the Avid PNT plug-ins. For this test I compared how long it took each of the three test systems to do the same task.
As you would expect, the sheer amount of power available makes AudioSuite processing much faster on the 3XS system. The speed appears to increase by a larger percentage than the specification, which shows that the combined power is more than the sum of its parts.
Test Four - Time Taken To Offline Bounce A 30 Minute 5.1 Mix
When at the end of the day, the client asks you, "could you quickly export the mix as is, of this reel, so we can take a look at it with the grade tomorrow?", having a good amount of Native grunt to power offline processes makes a huge difference to how late you'll be working that night.
So we can see that the Scan 3XS PowerDAW can bounce out the same 30 minute section, two minutes quicker than my own system, but more than twice the speed of the poor Mac Book Pro. The time saving implications are obvious. Along with speed of offline bounce, the massive amount of processing headroom available means that you can also safely undertake other tasks while you're waiting for the offline bounce to finish.
Test Five - Time Taken To Export A Premiere Pro Sequence
A lot of Workstation PC owners don't just use it exclusively for Pro Tools. Among the Expert Team, most of us also do video editing and among the Windows users, Adobe Premiere Pro is the go to software. I've been noticing Premiere's use rising in Post Production, mainly because of the excellent integration with graphics packages like PhotoShop and After Effects.
I recently made an overview video for a plug-in manufacturer. Here's how long the three systems took to export my final edit to H264.
Here's where the choice of GPU makes such a difference. We saw in the Pro Tools Expert Stress Test the difference it makes to the Pro Tools GUI, but here Premiere can work with the NVidia Cuda cores to farm out the video rendering to the GPU. No such luxury with the AMD GPU that's in the Mac Book Pro.
Test Six - Time Taken To Reboot and Restart A 5.1 Feature Film Mix
We've all been there. For one reason or another - maybe you've had to update a plug-in or another piece of software and the installer demands that you do a restart to complete the installation. There are generally good reasons for this - the new software will often have to embed itself within the menu context of the operating system or the software you're using it with. So I decided to test how it took to reboot the system and re-open and play a busy feature film mix with HD video.
This isn't the easiest of tests to film, as you obviously can't run screen capture software while you're rebooting the system. So, you'll have to take my word for it. Here's how the systems compared -
MacBook Pro | 8dB Self Build | 3xs PowerDAW | |
---|---|---|---|
Reboot Time (mins / secs) |
3:00 | 3:40 | 2:15 |
This is the really surprising thing. Although the Mac Book Pro has the slowest processor, less RAM than the 3XS and much slower system drive access speeds, it still performs this test faster than my own self built system. Before you see this as too much of a victory though, my own system has a raft of software that kicks in on boot-up - media backup software, sync software and various other things going on, plus it has to mount up network drives and three times as many drives as the MBP or the 3XS. I feel that if I disabled all those (which I really don't want to) then tried the test again, it could well have the speed of the 3XS system. I'm making excuses I know. Imagine how fast OS X could perform this test with the 3XS hardware though...
Conclusion
A lot of you may have thought that the results of these tests was a forgone conclusion. However a lot of the high performance of the 3XS system is not just down to the specification of the component parts. What makes this system perform so well is how those parts have been put together.
I mentioned in Part One, the various pitfalls that I've experienced, building my own system, and the various problems that I avoided by having Scan build and tune this system. Using this system has been an absolute joy, as it just simply works. No fuss, no complaints, no further tweaking needed and no further "optimisations" required.
It may cost a little extra, but the savings in time alone during your day-to-day workflow, make it entirely worth it. I can't recommend enough, going for a pre-built solution like this. Especially when it's been put together by the experts at Scan.
Come back for Part Three, where I'll be attempting to use this 3XS system to replace the Mac Pro which powers a real working post production rig.
The machine reviewed is a Scan Computers 3XS FWX99 PowerDAW, the full specification of which can be found in part one of this review. The cost, including delivery, at the time of writing was £2600.