I’ve taken this topic somewhere else to prevent going too much offtopic.
They have and they are called “receptors”
Maybe some general effects that could be implemented and some cards have these DSP modules, but not every host can handle that. But an audiocard could never be designed to handle arbitrary code which generates audio like VSTI synths do. A VSTI is an application itself and requires a real CPU and not some chipset with a few embedded rendering options.
A receptor is currently the only resource effective way to go.
Actually, Receptor is not a DSP chip, nor a card. It’s a whole PC running a modified Linux distribution. Check Receptor’s website. Especially the FAQs and Configurations sections. It should clear things up for you.
So, no, there is no VST processing chip and there probably won’t be. VST effects and instruments are too high-level for that.
Ambi’s got it right about Receptor - seperate box.
MY argument is that you can choose to have the CPU or a dedicated chip to run 3D and Physics, why not a dedicated chip for audio plug processing? And, no I do not mean on-board DSP bundle like some old soundcards have - I mean a seperate board and chip that apps know how to ship audio processes to. ASIO by itself just doesn’t cut it. Nor would the addition of Core Audio. A seperate chip would finally close the gap.
Anyway, we would have heard about such a thing by now if it did exist. And none of you guys are hardware engineers, so it remains a fantasy.
I once read somewhere, something about people looking into using their graphics accelerating cards to dsp audio, run vsts on. Was some in development opensource thing, but I can’t remember where I read it.
Commodore had this nature to design dedicated chips for every process besides the CPU itself (Amiga:Agnus for the Graphics, Paula for audio). That’s why I liked Commodore in the first place, they not only did good design but also cleverly thought things through and understood that each process should handle it’s things it can do best including processing power.
IBM and all the clone makers never figured that out specifically, the CPU was the core and only a seperate FPU had to do the hard work for tremendous and complex computations but that was generally about it.
Well, as Jonas has pointed out, the design philosophy will change in the next few years. Users are sick of putting up with crap and are willing to pay to get performance results. I think things can only get better, especially if we are patient - that’s something I have to practice at!
Escii are you saying if you had Kore then you wouldn’t have to worry about standards differences? Still need the hardware to juice it.
A VST processor card would contain x86 CPUs, since that’s where VSTs are written in. That’s why they don’t exist, just buy a new CPU (or 2). NI Kore doesn’t accelerate anything, it’s not intended to.
The x86 architecture isn’t that great for DSP intensive tasks. But the future all depends on the plans of Cubase’s VST architecture and large software houses like NI, which still command the standard how nice opensource ideas may be. Loading processing off to the videocard is one option, but this only works with the PCI-e architecture due to bandwith. However, with the introduction of large scale vector based accelerated CPUs like the Cell processor (bet on it Intel is developing something similar), I think the big guns will go that way.
Here’s a working Delay and a Chorus GPU plugin.
I have not tried the latest versions. But it did not work quite well in renoise. They worked in energyXT though.