At the January NAMM show, TC Electronic showed the TC PowerCore, a PCI card with sufficiently beefy DSP to run multiple copies of TC's most substantive reverbs. The company intends that this card be an add-on to native-based digital audio workstations to provide the kind of plug-in power that would choke even a dual processor CPU. TC wasn't alone: Kind of Loud (KOL) had its UAD-1 card, which runs KOL's beefiest reverb as well as Universal Audio's compressor plug-ins. Reports circulated at the show about other manufacturers either showing or contemplating similar cards.
What's this all about?
Hardware DSP cards for computer-based DAWs are not new. Digidesign built Pro Tools around its DSP Farm (now Mix Farm) cards; Lexicon made the beautiful but tragic NuVerb, which was essentially a model 300 on a NuBus card; and there have been other such systems for the Mac and PC. Accelerator cards are well established in other arenas, such as video and graphics.
Aside from the obvious fact that manufacturers can utilize more powerful DSP chips than in the past, what is different is that these new cards will not be part of the core system as Digi's Farm cards are, but powerful add-ons. NuVerb fits this description in that it was intended to be an add-on to Pro Tools, but the concept of plug-ins and the requisite effort on interfacing had yet to reach full flower at the time. The new cards will therefore need to integrate closely and seamlessly with the user's software environment. Fortunately, standards exist now that make that easier to achieve.
A key hurdle is that plug-in developers must be convinced it's worth the effort of coding for the card's hardware, which could be more of a marketing challenge than a technical one. If TC Electronic or KOL is alone in the plug-in accelerator market, even for a moderate period of time, they might be able to garner support from third-party plug-in developers. But if several competing cards run card-specific versions of the plug-ins, developers could be pressed to choose among them.
Even with only one card on the market, the numbers sold are likely to be only a fraction of the number of DAWs out there. The special programming effort required to support the cards, combined with relatively modest sales revenue to amortize the expense, suggests that plug-ins for such cards could cost more than native versions.
Bang for the buck is what will make these cards worthwhile to users. If a card costs a couple of thousand dollars but there are oodles of plug-ins that run on it and it will run lots of plug-ins simultaneously, users at many levels will be interested. Even if TC Works' PowerCore, for example, only runs TC plug-ins, the product could still be a good value for those who appreciate TC Works' high-quality software.
Certainly, post houses and other places that need to run lots of heavy plug-ins will fill machines with DSP plug-in cards that run the reverbs and effects they want. But personal-studio owners will look more closely at the cost of entry.
Another question is obsolescence. My NuVerb is a NuBus card that can never migrate to a PCI-based Mac, but it still sounds beautiful as a standalone processor housed in a Mac Centris 650. I'd get nothing for the Mac if I sold it, so I keep it as a NuVerb chassis. Another concern with the new cards is hardware compatibility: how nicely will everyone play together sharing the PCI bus and CPU resources?
DSP accelerator cards have always had their place in workstations, but the context is shifting. The fact that DSP cards persist indicates that manufacturers feel the cards still have intrinsic value that has not yet been displaced by something else. However, drawbacks have dogged the concept throughout its history. Needless to say, the complexities in the situation go beyond what I can summarize in a column this short, and there are too many variables to confidently predict the likelihood of success for these new products. It will be interesting to see how this hand plays out.