|
A Creative COW Product Review
|
|
Walter Biscardi
Biscardi Creative Media, Buford, Georgia USA
© 2003, Walter Biscardi and CreativeCOW.net. All rights reserved. |
Article Focus:
A few months back Walter Biscardi had the eye opening experience of testing one of the first Firewire 800 drives to come along. At that time, he did not believe all the hype about FW800 being able to run Uncompressed video with any reliability. But that test sure made a believer out of him. Now Firewire 800 is making the evolution from a single drive unit, to multiple drive units striped together in RAID format with LaCie introducing the d2 Big Disk Extreme series. In this article, Walter takes this new offering from LaCie for a test drive. Walter Biscardi is a leader in both our Final Cut Pro and Pinnacle CinéWave forum communities where Cow members have long appreciated his friendly and insightful input.
|
 A few months back I had the eye opening experience of testing one of the first Firewire 800 drives to come along. At that time, I did not believe all the hype about FW800 being able to run Uncompressed video with any reliability. But that test sure made a believer out of me. (See the review: WiebeTECH Fire800)
Now Firewire 800 is making the evolution from a single drive unit, to multiple drive units striped together in RAID format with LaCie introducing the d2 Big Disk Extreme series. Until now, RAID has mainly been the domain of SCSI and Fibrechannel drives, and some folks stripe together internal drive units for a RAID array. So what’s the advantage of a RAID over a single drive unit? In very simple terms, your files are spread out over multiple drives so each drive is playing a very small part of each file, thus the information can come off the array faster than it would if it was coming off a single drive. This allows you to play back more tracks of video in realtime and also play back larger, uncompressed files.
The Test Setup
PowerMac G5 pictured with (left to right) LaCie Big Disk, LaCie d2 Big Disk Extreme units.
The editing system we’ll be working on is an Apple PowerMac G5, dual 2.0ghz machine with 3GB RAM, Aurora PipePro uncompressed capture card, running Final Cut Pro 4.5. We’ll be capturing both 8bit and 10bit uncompressed material via SDI from a Sony J3/902 multiformat Beta machine.
The drive unit we’ll be testing is the d2 Big Disk Extreme 500GB model which features a 7200 rpm drive and an 8MB buffer. LaCie sent us two units so we’ll be striping them together. Along with the drives, LaCie also sent along a Firewire 800 PCI card (LaCie part # 000321) as this will be necessary to stripe the drives together on the G5, more on that shortly. The FW800 card gives you three more FW800 connections to your computer. In addition, LaCie sent along a Firewire 800 PCMCIA card to allow FW800 connections to a laptop but as of this writing, I have not had a chance to test the drives on a laptop. I will add an addendum to this review when I get the chance.
LaCie Firewire 800 card for desktop computers
For those familiar with the d2 Firewire 400 series from LaCie, these Big Disk Extreme units are a little taller, a little wider and a little deeper than the standard d2 models.
Installation
Installation of the LaCie elements were very straight forward. Because this test will involve running the units in a RAID setup, we’ll be installing the FW800 PCI card that LaCie sent along. But if you simply want to purchase and run one unit as a stand-alone drive, you can do that as well by connecting the unit to any available FW800 port on your machine.
Now one thing to note is that the drive units do NOT have an on/off switch so you really want to attach these units with your machine off. They will suddenly kick on if you connect the firewire cable and then the power cable. They are designed to turn on and off automatically with the computer.
First, I installed the FW800 PCI card to Slot 4 of the G5. Per instructions received from LaCie, I then connected one d2 Big Disk Extreme unit to the PCI card and the other unit directly to the G5 FW 800 port. I powered up the G5 and both drive units mounted up on the desktop as individual drives. I then used Apple’s Disk Utility to stripe the two units together as one RAID array and the array popped up on the desktop. All in all, the entire installation took about 5 minutes.
The first thing I noticed when the system powered up was the noise, or I should say, the lack thereof. Those of you who run SCSI or Fibrechannel arrays in your edit suites know how loud the fans and drives are when they are powered up. Generally you need to purchase some expensive cabinetry to muffle or cancel out the sound while still getting them the air they need to stay cool. The Big Disk Extremes do not use fans and were essentially silent. That caught my attention right off the bat.
Drive Speed Testing
The big first question we need to answer is “how fast is this array?” Well let’s find out.
On my system I happen to have the CineWave Drive Performance test application which I’ve found to be a very accurate benchmark for video applications. Here are the results:
Now one thing you may be wondering is why did I use HD 1920x1080 as the test format? Because I’ve found that this gives a much better read on the drive speeds than running a test using Standard Def formats. These speeds are quite good and definitely capable of uncompressed capture and playback and probably some real time effects/filter playback.
How does this compare to “traditional” SCSI or Fibrechannel arrays? I also have a Medea RTRX 1440 unit which has 10 drives striped together in a dual channel format which is currently connected to a PowerMac G4, Dual 1.25ghz machine and connected to an ATTO UL3D SCSI card. Here are the results of the same disk performance test on that unit:
The speeds are about double, but then we’re comparing a 10 drive unit vs. a 2 drive unit. So it’s really apples and oranges, but you do get a sense of how the FW800 unit compares to the SCSI. It’s interesting to note that the RTRX is a 1.4TB array with 10 drives and the Big Disk Extreme is a 1TB array with only 2 drives. So from a storage standpoint, they’re pretty close.
I should also note that this same RTRX array attached to the G5 with an ATTO UL4D gives results in the 210-220mb/sec range so the drives are even faster than this test represents.
Technical Testing
To simplify this review, we’ll look at all the 8bit work first, then we’ll look at how some of the same tests ran with 10bit material.
The first thing I did was to load up a previously completed Final Cut Pro project which had all the media offline and recapture all the media using Aurora’s 8bit eXtreme UC codec. There was approx. 30 minutes of footage off 8 DigiBeta tapes and the capture process was flawless.
Let’s look at real time playback first. I started by stacking 4 video clips only (NO audio) on a 10 second timeline, then scaling each clip down to 45% and placing them in the four corners, and finally set FCP to loop the playback. Not all the clips ran the full 10 seconds as I wanted to see if these files dropping off would have any effect on playback. This timeline ran no problem looping for a while with no dropped frames. After adding a 5th clip at 30% scale, the video playback finally dropped frames.
Now I went back and placed the same 8bit video clips in a timeline, but this time added the audio tracks as well. Playback worked perfectly fine with 3 video clips and 6 tracks of audio, but once I added the fourth video track with audio tracks 7 & 8, I got consistent dropped frames after 4 seconds of playback.
One last little test with the multiple streams of real time playback was to step back to the 3 video tracks / 6 audio tracks and make the timeline longer with multiple clips in each track. So I grabbed a bunch of raw clips and ran the timeline out to approx. 1 minute and let that loop for a while. Video playback was perfect with no dropped frames.
Now how about working with 10bit files? Ummmm, it’s a lot like working with 8bit files.
I started by capturing about 15 minutes of footage from another project via SDI using the Aurora 10bit codec and not a dropped frame or hiccup to be seen. The Log and Capture controls were very snappy with no delay as when I captured using 8bit.
I laid all 15 minutes worth of footage onto a 10bit timeline and pulled the playhead back and forth across the entire timeline for 30 seconds or so. No issues and the drives stayed right with me displaying the correct image on the screen as I scrubbed.
Now I created the same timelines described in the 8bit section and….. well about what I would expect. Some drop in performance with the larger 10bit files. In creating the same timelines as described above, with and without the audio tracks, I was able to run two tracks of video in real time, both scaled down 40%. So I can’t get a lot of multi-track real time video playback, so I tried out some filters, text and graphics with 10bit.
A Photoshop generated Name Super laid over a 3 Way Color Corrected clip played out just fine in real time.
I then added some animation by having the colored bar and the name slide in from opposite sides of the screen over a 3 second duration, no problem. I then added a Gaussian Blur to the video clip starting at a setting of 10 down to 0 over the first 10 seconds of the clip. That dropped frames, but if I turned off the 3 Way Color Corrector, it worked just fine.
Ok, Blur was too much with another filter, but some other filters stacked up just fine like the 3 Way Color Corrector and Flop so it just depends on the speed of your machine as to how many filters you’ll be able to stack up. For regular day to day editing, being able to apply a graphic over a color corrected clip is a real necessity for speed and I was able to do this just fine on the G5 with Final Cut Pro 4.5.
“Real World” Testing
So now you know how the drives perform from a technical standpoint, but what about that all important “Real World” test? You know, that test when you’re charging the client by the hour and they’re sitting right there next to you paying attention to every little thing you’re doing and they especially notice every time the system hiccups, crashes, or otherwise misbehaves and you start sweating just a bit and nervously hope that each time you ask the machine to do something it won’t crash or hiccup again and………. breathe, Wally, breathe.
I installed the Big Disk Extreme onto my system on a Saturday and had a client coming in for a 5 day edit starting on Monday. Normally I wouldn’t put something so un-tested into production with a paying client, but the installation over the weekend went so well that I broke my own rule.
So my client showed up Monday and we cut two projects over the 5 day period. What happened with the Big Disk Extreme? Nothing. Absolutely nothing.
Dropped frames? Zero
Unexpected quits or crashes? Zero
Render problems? Zero
UFO sightings? Zero
It was the most stable system I’ve cut on since we opened our facility three years ago. Halfway through the first day my client commented on how quiet it was in the suite and asked how I got the drives to be so quiet. I pointed out the two little silver boxes alongside my equipment rack.
The Verdict
In all, I’ve probably run the Extreme for about 15 days of production so far with absolutely zero issues. Is SCSI and Fibrechannel faster? Yes. Does Firewire 800 RAID work in the day to day operations of 8bit and 10bit editing? Yes. Now one of the quirks that LaCie did tell me is that the drives are actually faster right now on Apple G4 machines than G5’s. I did not test this out as I could not free up my main G4 for testing, but that’s something to keep in mind.
Now I’ve got to say I’m really excited about the possibility of moving away from SCSI/Fibre in my Standard Def suites and moving to the quieter FW800 arrays. Sure I give up some speed on the units themselves, but from a real-world editing position, I really don’t up anything but the fan noise.
I’d have to give the LaCie Big Disk Extreme a clean 5 out of 5 Cows as they have delivered on everything I expected and more. It’s a solid performer and quiet to boot!
Walter Biscardi, Jr.
Creative Director
Biscardi Creative Media
Creative Cow Final Cut Pro, CinéWave and Atlanta FCPUG Forum Host
### |
Digg it
