Hi, I’ve been reading up on Ardour for the past week and am deciding to give it a shot now. So I’m still quite new to all this.
I have an idea which might prove interesting.
I recall reading in another pro audio forum (largely windows based users), where people compile their own plugins in C++ and the sort (forgot what it is called), that there was some kind of development of a certain range of plugins that uses modern graphic cards as DSP.
My suggestion is to enable Ardour to automatically tap on GPUs for its plugins. I mean, modern GPUs have come to a stage where it is so powerful, even ultra graphical games fail to fully utilize it all (like frame rates in excess of 100+ @ 1600x1200 on a triple SLi platform).
What’s more, it is far cheaper to have an SLi system (a pair of 8600GTS shouldn’t cost more than US$150) than say, a quad DSP UAD-2… granted you don’t get the Universal Audio Neve approved virtual vintage neve plugins and such but still, you’ll have plenty of horsepower nonetheless.
Also, nVidia announced the CUDA development platform allowing people to use C programming to utilize the sheer processing power for other complex computational tasks.
A new source even reported on a group of russian programmers utilizing CUDA to crack a 128-bit WEP key in 45minutes (or a timeframe that is equally short in magnitude)
I believe this is a very interesting avenue to look into. Sorry I can’t contribute more technically.
Thanks for reading.
Nothing like the smell of fear…
I think Ardour is not more widely adopted, because of ignorance, and a strange believe that nothing that can be acquired for free can be better than something you pay for. This is slowly changing, and everyone that sees me using Ardour is instantly impressed. They are especially impressed when they hear it can be downloaded for free, with optional donation. Then they seem slightly miffed, because they just paid a bucketload of money for some app that essentially does the same (or less)…
Though I like the idea of using “untapped” resources on my system, I don’t need any more at the moment. Linux for multimedia is much leaner than other Generalised OSes, and this leaves you with more CPU power to drive those plug-ins
I say: Let Paul and his team focus on what they do best, keep improving on Ardour (and of I may be so bold to ask - Jack )
I see, perhaps the dev team could start with some of the basic effects and processors like eq, compressors and reverbs.
EQ units that emulate the tonal characteristics of famous analogue ones seem to need a lot of dsp power.
Reverbs tend to be very dsp heavy too. Virtual Lexicon MPX1 anyone?
I think the whole idea is to set a standard in the way that future plugins access gpu processing power as an alternative DSP source and show the industry that it works and is effective at that.
Once people realize that this is a very cost effective alternative, Ardour could draw the attention of companies like Waves, IK Multimedia and the like and compel them to create LADSPA versions of some of their closed source plugins. Thus drawing even more attention from wide range of users like budding musicians with limited budgets, producers, home studio owners, project studios to seasoned Pro tools users or even cost conscious animation facilities that probably have a storeroom full of outdated server farms which could still provide a good deal of productivity for in-house voice over and foley recording.
the problem is that one hand this is indeed an exciting idea from a technological standpoint, but at the same time, it is precisely the lack of any ability to use existing plugins that is the single biggest roadblock to wider adoption of ardour today.
Don’t get me wrong, I’m not demanding that it happens right here right now.
Just that I’m very excited about this project and am glad to have found out about it.
I wish to test my install soon and try running a few sessions with it asap.
And the fact that it is open source alone gives it such vast potential to be much more than a “protools clone” for linux. Like how AMD started as a “pentium clone” company ended up being the biggest darned rival Intel ever challenged till today… I think it’s cool that Ardour could just be the next AMD in the DAW arena.
Hmm, sorry Paul, not sure where to start. The last time I noticed the development and its link, the site was dead.
But for what users will want to use, I realize that the analog channel strip emulators tend to be very intensive on pure CPU power, like the Waves SSL 4000 RTAS plug-in used in a Protools M-powered setup. Have 20 of these running and you’ll see a good jump in CPU usage.
I was just thinking if, somewhere along the development roadmap, there will be some kind of plan to source and integrate GPU power into Ardour’s functionality.
Like how UAD’s software allows programs like Protools to use the UAD dsp power when a UA plug in is used, or a TC Powercore with its own plugins… etc.
But instead of separate proprietary hardware with its own set of plugins, drivers and parameters, perhaps one can install an add-on driver into Ardour so that it can harness a GPU or two for usage with ANY plugin installed into the DAW host.
Sorry, I haven’t got to understand how Ardour fundamentally works, so what I’m curious about may be a huge undertaking if it were to be pursued. Thought I’d just ask anyway.
I simply think that this implementation may shake the whole dedicated-dsp-plugin market and really shift attention to this open source alternative.
Sure there’s Pro Tools HD and its own set of dsp farms, cheaper alternatives that works with logic and nuendo and what not, but if the open source community can have its own answer that can rival the power of a HD/dedicated DSP farm system at a very tiny fraction of the cost, I am sure this project will receive a lot of attention, funding and resources it needs to be the next big thing in the audio industry.
Name a single plugin that would run on this processor. Next name a single plugin that users would find compelling that would run on this processor. Next name the developer(s) that would change this situation.
Oh yes, and I forgot to mention the sheer bandwidth difference between most pci-e 1x slots which most DSP cards are on these days and pci-e 16x (mainboards with all-16x pci-e slots are becoming a norm these days too) that will prove to be very VERY useful for a massive session of lets say 64 tracks each with 3 or 4 plug-ins, 6 reverb aux ins and 3 delays (some guy on the Mix it Like a Pro video series uses this channel configuration… on a PT HD system, no less).
And this could even be beneficial for notebook users who have discreet graphic solutions. Discreet graphics hardware hardwired onto mainboard vs firewire bandwidth (the wildest notebook I’ve seen so far is the one by Toshiba with onboard tri-SLi featuring mobile variant of the nvidia 9800GT chip).
I believe that with this functionality, Ardour might have a far superior competitive edge with home-studio/project-studio owners… at least. i.e. more users = more chance of getting people who can contribute code to this great project perhaps?
Ardour already has the attention of companies such as Waves. You will note that their plugins are not available for use on Linux. There are no plans for this to alter at any time in the foreseeable future.
Really, this is a chicken-and-egg situation, and we do not have the manpower to do anything about this. I think you overestimate the impact of this idea - although UA have been successful with the UAD-1, they are hardly taking the industry by storm. Even if this uses h/w that some users have installed, it doesn’t use h/w that every user has installed.
I don’t mean to pour cold water on the idea, its just that for it to go anywhere, somebody has to put a huge pile of man-hours into the idea. And when I say “huge”, I mean “huge”.
Well, I understand that it’ll take a lot out of the dev team and that there are other things to focus on at hand.
Please by all means, don’t go out of your way.
I just wanted to share what I had in mind inspired from the GPU idea I noticed floating around www.kvraudio.com/forum (the forum I was talking about) and the potential of Nvidia’s CUDA sdk, that’s all.
Ardour is a GREAT app, btw. Just took a tour of its features and settings, I think I can get used to this.
When it comes to using softwares, I am not a reference as I prefer using hardware tools (not only in music). But ardour is one of the few softwares that I use as much as a real physical tool or instrument. It is fucking great to have and I wonder how I would do without. Sure there would be ways, but ardour is just what I need for my music projects. I wish the few other softs I use (and there aren’t many) would be as cool as ardour. If you need more power, upgrade your system. I don’t think however that you need a powerful GPU for a DAW but it may be just me.
I don’t think we’re on the limit of CPU-powered processing yet, although it’s wise to think towards the future. We should also be thinking of proper multicore/multicpu usage like jackdmp. How is jackdmp doing, anyway?
jackdmp has very little to do with this. the data flow through JACK is rarely parallelisable. stephane has an experimental version which subdivides blocks to create “artificial” parallelism, but at the cost of a bit more latency. there are plans for parallelism within ardour, but time is the critical resource here, not cleverness.
Well nowadays games are very graphical oriented and need a good graphic card so that the game runs smooth and with no errors. Really nice to know on the facts that these graphic card can be used an additional DSP power which is good.Thanks for the post, its just like a Bunny Invasion
asldnalkfsm Wow. sohbet This is a very useful and informative article.sohbet odaları You definitely not only know your stuff, you know how to illustrate your point well.chat