Hi Tene
There are two big differences between DAW's like Cubase and EigenD. The first is simply one of data rates - MIDI has a really low data rate which means that the software doesn't have to do that much when it receives a note - it only has a few bytes of information to interpret, and those are right at the front of the note. In Eigenharp land a note is a really large amount of continuously changing data (around 10K a second per key) - this is what makes it work so well. Even when it is down translated into MIDI for AU's it still sends a LOT more data to the AU then they normally receive, and in our native sounds we don't throw the data away so it all ends up being used. This tends to make everything use a load more CPU and has a major effect on polyphony. It's a tradeoff between expressivity and number of notes. We favour expressivity fairly aggressively, as you have probably noticed.
The second big difference is the sheer modularity of our system. It's very fine grained (there are over 600 individual Agents running in Alpha 3 with around 12,000 discrete connections between them) . At the moment you pay a price for this in performance (for a variety of reasons such as signal routing overhead and cache locality) but don't really get that much benefit as configuring the system is rather hard. Fixed routing architecture systems like DAW's don't have this problem as they can optimise one data path. The price you pay for that is a huge inflexibility that I personally find very frustrating (having grown up in the world of patchbays and jack plugs where almost any routing you can imagine was possible). DAW's are effectively a representation, a fossilisation even, of the routing architecture of the 80's/90's mega consoles like SSL's, but signal routing was always a changing, creative endeavour in studios before then, to great effect. Sorry, rant over!
You will find this a much better tradeoff when the Workbench is done (ah, the mystical Workbench I hear you say - it's coming along but will still be a little while I'm afraid!). The fine grained nature of the system (which if you've ever used Max/MSP you will have some idea of) is very very cool though, and we believe worth paying some price in performance for.
Having said this we are very aware that we'd like to increase available polyphony. It's not been a big issue up to now but some of our players are now starting to stretch the envelope and are running into CPU performance limits. There are a number of things that we can do to improve matters - code optimisation to make basic processing faster, adding the ability to 'freeze' recordings as audio rather than played notes and allowing for a lot of decimation of data when you don't care about expression. Of these the only one that is not a compromise is number one, and thats where we'll be working next. 'Freezing' takes has the downside that you can't change their key or tempo later (which I consider to be a major minus) and reducing data rates seems to be a bit pointless outside of banging drums. I think we'll end up doing both of the latter ones in time anyway though as they do promise to be handy, especially with some of the very CPU intensive sounds.
John
UK
