DAW Topology:
Native Versus DSP-Based Processing

Go To Page

1 2
Go To Page

Help me, Rhonda! (Lady driver.)

As they say in wacky world of drag racing, “Nothing beats cubes*.” To date, short of jet engines, nobody has come up with anything to match nitro-burning, supercharged big-block V-8s, coupled with brute-force engineering designs and some of the steeliest nerves this side of NASA.

These top-fuel leviathans can turn quarter-miles in the mid-four second range with terminal speeds of 330+ mph. It’s an awesome sight, and one to thrill the blood. (Hey, and you thought audio journalists were all wimpy nerds!)

As with 5,000-HP slingshots, so with digital audio workstations: we can never have too much processing horsepower. Trouble is, unlike members of the National Hot Rod Association, our community is limited in the number of “big-block” processors at its disposal. Which leaves us with but a few system topologies for fabricating DAWs capable of accommodating multiple tracks of full-on audio processing, mixing and editing.

Which is hardly surprising, when you think about it. The personal computer industry is barely 20 years old, with the majority of advances being made in the past decade. Without the business community placing demands on PC manufacturers for more processing speed, storage capacity and data bandwidth, the pro-audio industry would still, metaphorically, be rubbing two sticks together to make fire. (Face facts: our purchasing demands are miniscule compared to those of Corporate America.)

Of course, two decades ago we could have ordered a multi-CPU system from the likes of Cray Research, lit the blue fuse and stood well clear. Big bang, bigger bucks. Yes, there were systems capable of performing several trillion floating-point instructions per second – and which found application in the Scientific Establishment and for modeling global economies, plus designing weapons of mass destruction. Of course, you needed access to the GNP of a country like Liberia to buy one.

These days, coming back to earth, the gap between what we need to run a multitrack recording session and what’s actually affordable is narrowing dramatically.

“It was twenty years ago today….” Well, August anyway. Original IBM PC, 1981.

Just a few years ago, the first Pentium-based PCs and PowerPC-focused Macs appeared on the market and were immediately snatched up for serious consideration by DAW manufacturers. (Okay, prior to that we had systems based around Intel 386s and Motorola 68000 Series micros, but they needed a lot of assistance to provide more that a few tracks of record/replay bandwidth, and modest real-time processing.)

While the base computer took care of full-color graphics display and basic disc access, custom-developed plug-in card were necessary for some critical functions. Such as real-time mixing and DSP.

Why the bolt-ons? Simply because, until recently, micros utilized in popular PCs lacked to computational power to process multiple tracks of high-precision data. Nor were their bus architectures optimized to handle such functions. Word processing and accounting tasks – even page layout and image manipulation – do not require the computational horsepower we need to handle multitrack streams of 16/20/24-bit digital audio.

Email this story to a friend.

Next Page