Hi Fran.

A single patch can run 21 plug-ins at the same time.

That said, remember this is only a tiny little 333 euro device with a 1GHz processor. Just because it can run 21 plugins at the same time doesn't mean it will. Same goes for your Mediastation.

Btw: It is running Linux too. It's got a separate CPU for the OS and the 1GHz processor is devoted entirely to the VSTi's your running so at no point does it get sluggish.

James,

Many thanks for taking the time to create the video demo of the V-Machine, it's the best I've seen so far! Now that the software interface has matured a bit and the Mac port is almost finished (out of beta), I would definitely be willing to give it a whirl. Hopefully, they will upgrade the CPU soon, especially with the new smaller-form Muse Receptor product on its way.

Sharp, absolutely loved hearing the Irish accent, as well as the way you have your M3 module situated on the keyboard. Nicely done!

Michael

Fascinating little machine... great price too...

My main question would be, you keep mentioning 64 voices of polyphony. How does it balance processor load to guarantee this? It sounds like you are using primarily sampled playback, which doesn't stress a CPU that much. But presumably it can run things like the Arturia stuff (I love their Memorymoog, CS80 and Jupiter8 sims) but these are processor hogs. How does it deal with this?

BTW, is the reverb a separate plug in? How does it handle effects routings?

I've read a few reports of the V-Machine being a hair buggy, but that could have been when first released. How do you find the stability (how hard are you pushing that CPU)?

I still don't see how this addresses the issue of dynamic CPU load when things get fast and furious...

I don't think I completely agree with that, Bachus. If you look inside most closed arrangers, you will see that there are MULTIPLE chips in there. Yes, there's a CPU for running the software side of things, but it is likely to be an embedded custom CPU, not a commercial general purpose one, and then you will also find chips for the voice generation and effects stuff, harmonizer, etc.. Basically, you are taking a normal synth/ROMpler's electronics, and then adding the CPU to do the arranger stuff.

But an 'open' architecture arranger, the CPU does EVERYTHING. And hence, if the voice generation gets taxed, it brings down everything else too.

One of the things that has improved the MOST on newer closed keyboards is that the faster embedded chips for voice generation have got VERY, VERY low latency. And yet, can still GUARANTEE 128 voices, no matter WHAT you are doing. On my G1000 I used to be able to hear subtle flamming even on two voice layers of very percussive sounds (say a clav and a marimba) when big chords are rapidly played, but on my G70, I can layer up to five stereo sounds on one note, and not hear the HINT of flamming any more. Ten voices per note! Thirty voices playing each chord, and NO flamming at all! I have yet to hear a software system that good that didn't get flaky as you pushed the CPU.

Will it get better? Of course... Is it good enough for demanding live applications yet, if you want to use today's most CPU hungry VSTi's? I am not yet convinced...

Hi Diki.



As you know each plugin generates it's own amount of CPU usage. Some are more efficient than others so the 64 notes I'm quoting are what I set as the limitation my Playback Sampler called Sample Lord can produce because it's not the only plugin running in at the same time.

Basically I did tests and monitored the CPU usage and then I told the plugin that it is not allowed to produce more than 64 notes. Doing this will allow it to smartly cut notes if you try exceed 64 notes, and it eliminates the chance of you reaching your CPU upper limit.




Yes I'm running a reverb plugin at the same time as Sample Lord.

You have FX slots that allow you to run more than one Effect at the same time, and by using FX slots the plugin is placed in an area that can make use of “Send” values which give you control over how much of the VSTi Instrument gets sent to the Effect sitting in the FX Slot.

In addition to that you can also fully control the effect length, type or anything you like from the V-Machine interface. When you like what you hear you press two buttons to update the preset and save the changes.

Pretty straight forward setup actually.



The V-Machine was never buggy, it was always rock solid and reliable as dedicated hardware.

The problems you where reading about are to do with the program you install on your PC / MAC which allows you to load new VSTi's into the V-Machine. It's a program called VFX and yes that certainly has it's problems.

It's not all that bad though in it's current state, and you can workaround it's problems simply by not connecting it to the V-Machine at all and just using a Memory Stick to transfer the data instead.



If you have any specific questions just let me know. For the most part it's up to you to workout how much freedom you give a Plugin to use the processor by limiting the Poly Count.

It's only a tiny 1Ghz processor, but I do believe the upgrade kit will double this for in around 100 to 150 USD.

2Ghz inside this tiny box will be hilarious. Can't wait to take that for a spin.

Sounds great, James, especially the Mac port.

I am going to keep a VERY close eye out on this as it progresses.

Couple more questions... when you restrict polyphony, does it cut notes if you exceed the limit on a first note basis, or can you use more sophisticated rules (like lowest level, or decayed layer) like hardware samplers (like my K2500) can? Or does it simply not play any more notes until some get released?

And also, is the latency adjustable, and what figure for it are you usually using?

I appreciate the answers... this thing looks pretty good for the price (especially when the dual proc comes out)

Hi Diki



Sorry but I don't know full the answer to that.

I can't tell how it's managed but I can say it's done extremely well. There's certainly no problems with laying down new notes at any point when you do exceed 64 notes.

It is very transparent and I've never felt like I could hear anything cutting off.



Yes it's fully adjustable and what works for me with Sample lord is a latency of 4.8ms for my sounds. You can go down to 2.4 but that's overkill in my opinion. Better not to run the risk of the sounds breaking up and to go with the 4.8ms

[This message has been edited by Irishacts (edited 05-30-2009).]

Very nice little box! I still would prefer a unit that gives me the flexibility to upgrade the hardware as I please. The currently available computer technology certainly eliminates the issue of latency and flexibility.
Finally, I am very pleased to see that topics of this type are given more attention.

[This message has been edited by hitman (edited 05-30-2009).]

@ Diki

You seem to forget the sheer power of modernday quadcore CPU's..

If an instrument can play 1000 notes of polyphony (Thats true) when CPU pressure/Memmory bandwith becomes a problem, it can cut off some notes and your ears will not even notice that

Your fears date back to dates from before quad CPU's and 8 gigs of memmory...

After all its just a process and it doesn't make a difference wether i controll a process with hardware, or controll it with software.;.. as long as i controll it.

Trust me, Bachus... go and try the Arturia stuff, or Pianoteq3, or Omnisphere.... There's no way you can get those to give you hundreds of voices with any modern CPU (let alone at a low enough latency to feel like hardware).

Sample playback is stupidly simple for a CPU, but virtualization, and modeling is an altogether different thing. The thing is, most VSTi's designers look at the CPU as just THEIR plaything, and use as much of it as it takes to create the most complex sounds you can possibly imagine. Good for us as musicians. Bad for the CPU. Seems that it doesn't matter how powerful he gets, someone writes something for him to do that truly taxes it!

The thing I particularly notice is that latency makes a dramatic difference on how well you feel 'connected' to playing a sound. From playing around with my soundcard's settings, and comparing the immediacy of the sound and the impression you are playing an instrument, rather than triggering a sound, to the latest hardware, I honestly don't think that 5-6ms cuts it. That's 90's hardware numbers. I honestly wouldn't be surprised if my G70 tops out at 2-3ms (twice as fast). It certainly FEELS far snappier than any software instrument I've played.

I realize that GIGA sampler can run hundreds of voices in realtime from a pretty average CPU. But the minute you start asking it to virtualize old analog gear, or model acoustic instruments, polyphony goes out the window....