realtime mixer - how do digital mixers do ir

pretty much sums it up. especially with the mackie dl1608.

the thing is tiny. easily fits in a small back pack with iPad. has 16 ins 6 monitor outs plus stereo out and headphone outs.

with gate comp on every channel. graphic on all outputs and 2 buses for high quality fx.

all with low latency.

yet try to do that with a pc and you could well be struggling.

get into higher end desks that have 2 and 3 times the channel count with more plugins and more outputs. plus multiracial recording via FireWire.

what’s the secretary behind there hardware. what is doing all the processing.

Dedicated hardware. A PC is a general purpose computer, so it has to be able to process any kind of data and on top of the hardware you run a mish mash of off-the-shelf hardware, a kernel, an operating system, a desktop environment , an audio stack, a DAW, then any number of plugins. Compare this to a dedicated digital mixer where it is built from the bottom up specifically for the purpose of routing audio, it is far easier to get a low latency.

Um to do that with a computer you may not be struggling, it all depends.

For instance the iLive, which is a 64 channel console with up to 24 (IIRC) busses, gate, compressor, limiter/de-esser, parametric eq, HPF, and a few other toys on each input, and GEQ, PEQ, Compressor, Limiter on each Buss, EQ on each FX return, etc. It also runs a computer OS called cOS for processing (And Linux on the surface for control). In general many consoles these days are actually closer to general purpose computers than the mixers of old.

Many mixers still do some sort of DSP chip internally however specialized for these purposes, so don’t take that to mean it is just a general computer. But I can likely easily run a similar setup with appropriately coded DSP on my 6 core machine so long as the coding is done correctly.

    Seablade

@christophski

Your answer is much more appropriate with the first generations of digital mixers, but now more and more of them are on far less specialized hardware.

  Seablade

thanks for the answers.

Aftert looking on studio spares last night i discovered that you can buy DSP plugin cards. so that kind of asnwers my questions too. Howver these DSP cards are specific to certain plugins like UA plugins, and theres another for SSL.

they are expensive though.

And i dont know if they run on linux.

Yeah, i noticed that about the I live mixer that it runs linux on top of the underlying OS thought that was pretty neet. Itrs the first time ive seen linux in a commercial mixer. One of the reasons i like it so much.

The i live is pretty clever, as far as i can tell most of the processing is done in the stagebox and the desk is more or less a controller. Though it does have some outputs on the desk, but ive been told the outputs on the stage box are better. Great thing about the i-live is if the desk crashes (and i have had it crash on me once or twice) reset the desk and the show keeps going . The stage box stays in its current state allowing the band to keep playing until the desk reboots.

This comes to another point which im struggling with as im upgrading my computer at somepoint.

AMD have some very fairly priced cpus but from what ive read that most software cant take advantage of all that power. In benchmarks of video rendering with other background tests some of the 6 and 8 core cpus could only get to around 50-60 cpu load meaning all that extra power isnt being used.

Is it going to get better with more recent kernels, or has it pretty much been that software developers are being forced to intel.

@veda_sticks

It has more to do with how the software is programmed, not the kernel. Pretty much any kernel can utilize as many cores as you have these days, but the software may not be written in a way to take advantage of it. IE. Ardour2 could only utilize a single core for the heavy DSP lifting, A3 can utilize as many cores as you have and is much more efficient with multi-core machines in this manner.

I am impressed you have crashed the iLive surface, I have two of them here, may get a third and a could of GLD consoles. I can count on one hand the number of times I have had a lock up or hard crash on the surface. But you are correct in that most of the processing happens in the stagebox for the iLive, contrast to the GLD where it happens in the surface. These more distributed approaches are not uncommon these days, though it does still vary from console to console.

And no the UAD or SSL or pretty much any other DSP for running plugins does not run on Linux at this time, or in the foreseeable future.

         Seablade

@seablade:

And no the UAD or SSL or pretty much any other DSP for running plugins does not run on Linux at this time, or in the foreseeable future.

I think most of the SSL plugins are also now moving to ‘native’ versions anyway - (and as someone who was involved with the development of the SSL hardware and plugins, I can tell you it’s not always clear-cut that using external DSP automatically provides an advantage - a lot of development effort went into making plugins run within some limitations of the external DSP - so perhaps as a result you get ‘more efficient’ and / or ‘better’ code, but it’s a side-effect of having to cope with other limitations - it could be argued that with ever increasing native CPU power, the principal advantage of that kind of external DSP is one of copy protection more than anything else… but that’s a different subject entirely).
One thing you can do with dedicated DSP is optimize very specifically for that hardware, because you know exactly what that hardware is (whereas on a PC it’s much more generic obviously). This is also something which some other hardware manufacturers - for example Apple - benefit from, in increased system stability, especially for audio - because they control all aspects of their hardware / software platform.

@linuxdsp

Completely agree on all counts of course:) I do find that these days it is more about copy protection than anything else for the dedicated DSP platforms, not that it doesn’t get you a benefit but the question is how much you benefit you really get from a DSP in a 6+ core machine in modern times.

And of course you see the same thing in other ways when dealing with the modern crop of digital mixers, as I mentioned they are moving more towards a general purpose platform, but you are completely correct, they are a known platform no matter what and optimized for that platform, and thoroughly tested on that platform.

       Seablade
..not that it doesn't get you a benefit but the question is how much you benefit you really get from a DSP in a 6+ core machine in modern times..
Yes, of course, there should always be a benefit from moving some of the processing to an external DSP in terms of load on the host CPU - but naturally that has to be weighed against things such as how many instances of a plugin the external DSP(s) can run etc, and whether that is acceptable / economical if the hit on CPU usage is negligable for a particular plugin natively.

*negligible… Still can’t edit my posts on this forum… :slight_smile:

I do plan on upgrading my current old athlon 64 dual core 2ghz system to something abit more modern like fx 6 or 8 core. but i keep holding back as i keep reading that AMDS chips perform poorly compared to intels. specifically that AMD’s cpus never get fully utilised in benchmarks. amd cpus only seem to max out at 70% cpu load where as intel can be maxed to 90%.

the benchmark scores also reflect this with AMD scoring way before intels equivilent.

I havnt seen any linux benchmarks though, so maybe this is a windows issue. The reason being stated though was software wasnt optimised properly for AMD’s architecture or something like that.

Im hoping that when i make the move things improve, as ive been an AMD user for probably over 10 years. and there prices are very competive.

AMD’s do not perform poorly. The basic rule is: if you want the highest absolute performance, you want/need Intel. If you want the best performance/cost ratio, you want/need AMD.

I develop on an AMD 6 core system. It is blindingly fast and was very cheap. I could upgrade it and get about a 40% speed up for about $120. The equivalent Intel system is a lot more money.

Are you able to max out all cores?

Think you have pretty much made me make a decision. Ill Stick with AMD as cash is an issue. For about £300 i could build an amd 6 core 3.5ghz system with 8 gig of 1600mhz ram and a decent enough powersupply. then later on stick in an SSD drive.

I already have a PCI graphics card that i can upgrade later aswell though not sure if that would actually give any performance incease.

Toying with the idea of using it live as an FX rack in venues that have poor anologue units

Building Ardour maxes out all cores on all systems :slight_smile: To be honest, I don’t pay that close attention to such things.

I’m running it on a 6 core AMD CPU.
It works very nicely but I’m not convinced it uses all the cores all the times it could; for example exporting a mix shows quite low CPU usage per core, as if there was only one thread running. (Yes, I do have preferences set for “use all cores but one”.)
In contrast, rendering a video with KDENlive on the same machine pushes most cores to nearly 100% and really gets the cooling fans turning!

i have noticed that ardour doesnt seem to export as fast as it has done in the past, though its a bit random. some projects will export using only 1 core while others will use both cores.

The last project i exported ardour was reporting only 76% DSP utilisation during expert when normally its at 100%

DSP utilization? Do you mean the JACK DSP load? Or the CPU utilization level ?

jack DSP load, i normally see the DSP load at 100% during export, but on some occassions it sits around 75-76%

It could just be some setting that different in my projects.

In saying that, its certainly not slow. It is definatly faster than realtime .