Recording

The Outer Limits of Portability

PUSHING THE POWER OF THE FLAGSHIP INTEL-BASED MACBOOK PRO 10/01/2006 8:00 AM Eastern

Back in February of 2004, Mix put the Apple G5 to the test, including the playback of DSP-intensive instruments while recording a live band, support for separate live headphone mixes for the band, playback of a QuickTime movie, quad playback of audio and even running the lights in the venue. The recent release of the new Intel-based MacBook Pro whetted our appetite for another test. So this article was reborn with a new goal: to find the recording limits of Apple's flagship laptop with the newest version of Logic, Apple's native DAW. We made a “logical” choice, as native apps are becoming more and more powerful as a result of beefier chipsets, faster throughput and drivers that impressively reduce latency.

For this stress test, we decided to use our system for a live recording of a five-piece rock band. Because we were using a laptop, the focus of the test was to keep things small. In fact, the footprint was so compact that, apart from our Mackie fader and plug-in controllers, the entire rig could fit comfortably into two small computer bags. To keep the system working hard, we recorded bass, two guitars and a vocal live. We had a drummer triggering samples while our pianist was on a controller playing a modeled keyboard in Logic. We also ran a video while we recorded, and had two instances of Apple's Space Designer reverb going, as well as many plug-ins. Plenty of CPU-clogging stuff going on here!

The 5.1-enabled D Room at the Conservatory of Recording Arts and Sciences in Arizona was our host space. Our room was set up to facilitate having our five musicians and engineer all in the same space, much the same as Peter Gabriel and producer Bill Botrell have done. With no one separated by walls, everyone became an integral part of the process. At the recording helm was Logic Pro über-man Robert Brock, helped by two able assistants from the Conservatory.

IN THIS CORNER
Apple accommodated us by sending a 17-inch MacBook Pro running a Dual-Core 2.16GHz engine. Our OS was 10.4.7, and Apple's Logic Pro was Version 7.2.1, the first version able to run on the Intel-based Macs. Upgrade features include 32-channel support for multichannel Audio Units instruments, integrated support for the new Apogee Ensemble interface (we tried to get it for this review, but it wasn't ready), Digidesign Pro Tools HD7 DAE support, Serato Pitch 'n Time support, improved ReWire support and Euphonix EuCon support, along with improvements for other control surfaces.

The MacBook Pro we tested weighed just under 7 pounds and featured a FireWire 400 port, a FireWire 800 port, three USB 2 ports and 2 MB of L2 cache that could be shared between both cores as needed. The front side bus runs at 667 MHz, and it's the first laptop to offer PCI Express, allowing data to travel in parallel across links and lanes, making for speedy throughput. Suffice it to say, this unit is top of the line.

The MacBook was maxed at 2 GB of RAM, and carried the 7,200 rpm SATA drive option, essential for our application. The 2 GBs of RAM let us run our sample libraries more efficiently. The libraries we used load the leading edge of all their samples in RAM, so the more we can get, the better. The first item on our Apple wish list is that we'd like to have had more RAM available — 2 GB is enough, but just barely enough.

A definite snag for this test was that we were ahead of the curve with regard to third-party plug-in support for the Intel Mac, which is low at this time. Even the libraries that offer support are running beta versions. For this reason, we were limited in our plug-in selection, but got to use some great collections (generously provided by ILIO and Prominy) during mixdown on a non-Intel — based system. The nice thing about Logic is that it comes loaded with some great stock plug-in options, including the Space Designer convolution reverb, which is a real DSP hog — perfect for our test.

STORAGE
Because the session involved both triggering and playback of MIDI samples, we needed a lot of storage. Glyph provided two PortaGig 7,200 rpm 100GB drives that we configured in a RAID stripe, using the stock RAID software in the Mac OS X. The PortaGigs can be bus-powered, but Peter Glanville of Glyph suggested we run the drives using their AC power supplies, as some computers don't supply enough bus power to sustain the 7,200 rpm drive (even if the computer itself is using AC power). These two drives held the fxpansion BFD library, which was nearly 200 GB and featured the new Jazz & Funk Expansion Pack, among others.

Other storage duties were ably filled by two LaCie Big Disk Extreme 500GB, 7,200 rpm drives. Each LaCie enclosure holds two drives configured in a RAID stripe and transfers data over a single FireWire 800 or 400 bus. These drives held the whopping Vienna Symphonic Library VSL Cube, featuring more than 800,000 samples and 550 GB of sample data recorded at 44.1kHz/24-bit. In addition, we had the latest version of Stylus RMX with four new S.A.G.E. Expanders, Synthogy's Ivory Pianos and Prominy's LPC Distortion and Clean Guitar. Adding the internal 100GB, 7,200 rpm drive on our laptop gave us 1.3 TB of drive capacity, and we used nearly every bit.

OUTSIDE THE LAPTOP
I/O was provided by the very portable MOTU UltraLite, a FireWire powerhouse offering two mic/instrument inputs with phantom-powered preamps with front-panel gain and three-way pad. I/O comprised six balanced line-level ¼-inch (TRS) inputs and 10 analog TRS outputs, stereo S/PDIF I/O, assignable stereo headphone output and one MIDI I/O. Just like our drives, we AC-powered the UltraLite to make it easier on the laptop.

One self-induced snag encountered with the UltraLite was when we tried to hook up our drives to the same FireWire port we used for our audio — never a good practice because it tends to clog the pipeline, but we tried it nonetheless. The UltraLite has a secondary FireWire port to connect to other devices, but when we hooked up our storage to that port, it would bring the session to a halt: Logic would stop outputting audio. Once the drive was removed from the port, the problem ceased.

Although for the sake of convenience we used the monitor section in the room's Focusrite Control 24 to get to our M&K surround monitor system, the UltraLite could have handled the task nicely. It lets you configure the volume control on the front of the box to control two, six or eight channels of audio (stereo, 5.1 or 7.1). This would allow you to have a handy master volume control for the whole mix, no matter where you were set up and what your configuration.

Other support gear included a Mackie C4 controller offering 32 rotary V-pot encoders with integrated pushbuttons; four 55×2-character backlit LCDs; and a Mackie Control Universal with 100mm motorized touch-faders, multifunction V-pots, automation controls, meter and timecode display. We had to break a lot of things out on hubs; for this, we went old-school with the Emagic MT4 bus-powered 2-in, 4-out MIDI interface, providing 32 input and 64 output MIDI channels, and a MIDIman Sport 2×2 hub. Bringing up the rear was the Compaq USB 2/FireWire 400 hub.

The only mic in the room was Korby Audio's Convertible microphone. This incredible transducer comes with four hot-swappable capsules modeled after a Neumann U47 and U67, 251 and AKG C 12. The vocalist/guitarist sang into the mic fitted with the 47 capsule, which ran directly into the MOTU's mic input. In total, we recorded live bass and two guitars and vocals — just four tracks — while our drummer and keyboard player triggered BFD samples and a modeled Rhodes via a Roland pad kit and Yamaha Motif ES6, respectively.

One of the guitars and the bass were taken directly into the UltraLite with Logic-resident plug-ins used for effects and processing. Due to the fact that the UltraLite has only two mic/instrument inputs, the other guitar was run through a PreSonus Eureka and then into the line input of the UltraLite. One slick thing we were able to do was use Logic's channel strip presets to get a quick setup for the instruments. Guitarist Leon Santiago's instrument was quickly dressed up with a Michael Landau preset that put a nice, slightly crunched stereo chorus across his guitar.

The drummer triggered the BFD samples while the keyboard player ran a resident Logic modeled Rhodes piano, which was replaced later during mixdown with one of the Ivory Pianos. For the groove, we ran a GarageBand loop. For reverb, we had two instances of Apple's Space Designer convolution reverb. Logic lets you jump between presets in real time; however, when we switched the channel strip preset to a clean sound, it reset the input source, making us go through channel input setup each time we switched the preset. Having to do this chore negated the feature's potential cool factor. It would be nice to be able to separate the input setting from the preset.

TESTING, 1, 2, 3
The first run at a take was at 44.1kHz/24-bit with the buffer size set to 256 samples. While the players didn't complain about latency, when we asked them, they said it was discernable and might have been a groove-killer if someone was picky. It didn't spoil the vibe of this particular band, though, with everyone locking pretty well into the track. The drummer's proximity to the monitors, however, kept him from feeling the mojo and locking to the track, but that was quickly remedied with a headphone feed sent to him from the front panel of the UltraLite. This output can be separately addressed using the software interface provided with the box.

This take went off without a hitch with the Logic system performance display showing one of the CPU meters (there's a meter for each core) at about 50 percent, while the Apple system performance meter showed that the whole system was running at about 50-percent load. While tracking, the twin meters showed that the computer seemed to run everything on a single core (meter); however, on playback, both meters were dancing and showing about 25-percent load each.

UH, OH!
A potentially fatal problem arose after a quick break. When we came back into the room, the MacBook Pro had crashed. A reboot resulted in the computer repeating the reboot sample over and over. After another attempt at rebooting by holding down the power button, the computer came up but the screen on the laptop would not. The pixels could vaguely be seen on the screen, but the backlight would not turn on. We eventually set the external monitor to display the main screen on a wall-mounted screen by shining a flashlight through the back of the monitor through the Apple logo and calling up the display screen to reset the external monitor to show the main screen. That logo isn't just for show after all; try to do that on a PC!

After our brief encounter with the infinite, we decided to kick it up a notch. The next test involved lowering the buffer sample size down to 128 to see if we could reduce latency. The latency was immediately cut in half, and the players were much happier. Latency was less of a problem for the players triggering MIDI as they only experience half the delay of those recording direct. This is because, with an audio input, there is latency going in and going out of the system, whereas with MIDI triggering, the latency is only on output when the sample is converted. We hit the red button and the song started recording just fine, but about two minutes later, Logic stopped recording and we got the Apple rainbow (i.e., the spinning wheel of death). The processor just couldn't keep up with the recording and sample rendering. This also happened on a second attempt.

We then started another take back at 256 samples and upped the sample rate to 88.2kHz, 24-bit, which, because latency is based on samples, is virtually the same latency as our second failed test. At this setting, the band got a few bars into the song and playback became distorted, and the sequence actually slowed in tempo. The buffer was then reset to 512 samples at 88.2 with the same results. The problem seemed to be the MIDI feeds and the system having a problem rendering the sounds at the advanced sample rates. The next solution was to try to up the buffer to 1,024 samples, which was better, but there were still unacceptable problems with distorted audio. For one last try at pushing the envelope, we recorded at 48kHz, 24-bit at the 256-sample buffer size. This worked fine, with just a few digital pops happening intermittently. The bottom line was that our first attempt at the 256-sample buffer size at 44.1kHz, 24-bit was the most doable solution.

THE LOGIC ANGLE
With any test where you're pushing a system to its limits, you're going to come up with problems and gripes, and while using Logic, we ran across some minor flaws and issues.

On the minor side, there were some graphical glitches. For instance, when you click on the surround panner, it is supposed to open a new, larger window that mimics the same panner on the channel, but this caused the channel panner and overview to become very glitchy with the meter and panner showing up, and then disappearing. Luckily, the problem went away with a simple reboot. A more serious bug was discovered when we used the bracket keys to toggle through presets on the Apple bass-modeling plug-in, which sent a loud spike through our speakers. This sometimes also happened when we changed the entire channel preset. Another minor annoyance was that when you are looking at the Track Mixer view, you can't move channels; you have to do it from the Arrange window, and when the Record button was on, it would eliminate the metering on the bass track.

Another interesting situation occurred when we were setting the level for the vocal mic in the studio. Of course, with surround speakers in the room, this setup presents a potentially dangerous feedback situation, but with precise mic placement, we were able to get enough isolation. The problem came when we put the track into record-enable. Logic sets different levels for the track when record is enabled and when it is not. This means that if you're not watching it, you can instantly have the input at unity gain, causing some nasty feedback.

There was one annoying feature that kept us on our toes. Except in a few nonobvious exceptions, in Logic Pro you cannot select more than one track at a time, and the only way to arm a MIDI track when stopped is to highlight it. The catch is that when you press Play with a MIDI track selected, the other audio tracks, even though they were record-enabled, will not go into record. Our workaround was to select an audio track, hit Play to go into record, and then go up to the MIDI track and select the MIDI track. Apple needs to make Logic more intuitive to the way engineers optimize workflow.

The last Logic gripe is a MIDI issue. Because Logic will not allow separate MIDI tracks for each instrument, you can't record to two MIDI tracks at the same time and address the outputs separately. This was a big problem for us because when we went to tweak the drums or keyboard MIDI feeds on the combo track, it would reset the outputs, resulting in drums being triggered from the keyboard and vice versa. In a nutshell, Logic should treat MIDI I/O assignment, track selection and record enabling the same way that audio tracks are handled.

BOTTOM LINE?
There is really no bad news here. Despite the screen problems and the fact that we hit the processor's ceiling, the performance of the MacBook Pro laptop was impressive. It's important not to lose perspective — this setup was formidable, with lots of DSP-intensive sample rendering going on, as well as audio recording using many plug-ins, including convolution reverb. After all, this is a 7-pound laptop!

What's even more exciting is where this all can go as the next-gen chips arrive and more manufacturers jump on the bandwagon. For instance, a week before our test, Intel released its new Core 2 Extreme processors, which are blazing-fast and run much cooler than past efforts; specs reveal that these chips run at 2.93GHz clock speed, 1,066MHz bus speed and offer 4 MB of L2 cache. Could we see these in next year's laptops? (For more info on these chips, check out: http://techreport.com/reviews/2006q3/core2/.) In addition, at MacWorld, Apogee released a new high-performance PCI-Express driver for its Symphony 32-channel PCI-Express card that boasts latency in the 2ms range. Although the laptop doesn't carry a traditional PCI-Express slot like a desktop, it does have an ExpressCard/34 slot that would allow using Symphony with an expansion chassis. This is exciting stuff and bodes well for portable and native audio.

All in all, this was a fun test, rendering some great sounds. Be sure to check out the mix in progress at mixonline.com. Thanks to Apple, Robert Brock and the musicians and the Conservatory of Recording Arts for providing the space.


Kevin Becka is Mix's technical editor.

SOFTWARE INSTRUMENTS
BFD Drums, featuring the Jazz & Funk Expansion Pack (www.fxpansion.com)

Vienna Symphonic Library Symphonic Cube (www.ilio.com)

LPC Electric Distortion and Clean Guitar (www.prominy.com)

Synthogy Ivory (www.ilio.com)

Stylus RMX with four S.A.G.E. Expanders including Noizbox, Stark Raving Beats, Ethno Techno and Skippy's Big Bad Beats (www.ilio.com)

THE COMPUTER AND DAW

  • This diagram illustrates the configuration of the equipment used for this article's tests.

    17-inch MacBook Pro using the Intel Core Duo processor running at 2.16GHz clock speed with a 7,200 rpm internal 100GB drive

  • Logic 7.2.1 upgrade: Intel-capable, 32-channel support for multichannel Audio Units instruments, plus support for Digidesign Pro Tools HD7 DAE, Serato Pitch 'n Time and more

OTHER HARDWARE

  • MOTU UltraLite bus-powered 10-in, 14-out, FireWire audio interface
  • Mackie C4 controller
  • Mackie Control Universal
  • Emagic MT4 bus-powered 2-in, 4-out MIDI interface
  • Korby Audio Convertible microphone using the U47 capsule
  • Compaq USB/FireWire hub
  • MIDIman Sport 2×2 hub

THE MUSICIANS AND THEIR INSTRUMENTS
Joe Morris (www.joemorris.com) triggered the BFD drums with a Roland pad kit using the Roland TD-6 percussion sound module as a MIDI translation device.

Bassist Felix Sainz was playing a bass with a G&L body, Fender neck and EMG pickups.

Leon Santiago played a custom guitar with a Moses neck, Hipshot Trilogy Bridge and Seymour Duncan active pickups.

Tony Kinchion played a Fender custom-shop Strat with DiMarzio pickups.

Ted Belledin played through a Yamaha Motif ES6, which was used to control various Logic instruments.