Hi Leif, 

I guess the reason why there are not many entries about MIR is that everyone seems to focus on VE Pro at the moment.

I'm myself a MIR-User and pleasantly surprised of its performance needs. On my Core i7 920 (still not OC'd) with 8 GB RAM I haven't encountered any kinds of problems running MIR with the full "Orchestral Strings 1+2" and the "Special Edition" Packages. You can expect around 40%-60% CPU usage on such a machine while playing a complex orchestral arrangement. 

Indeed most of the time I'm using the Mozart Saal, Grosser Saal and the new stage, so it might be different if someone's using the Foyer (imho most of the users won't need it anyway although under some rare circumstances it's a nice add). The only bottleneck is RAM; which honestly I don't really understand. For me it seems that MIR is compiling an IR-set every time you are moving an instrument on the stage. Why not compiling it to a 4 channel stereo-IR with the files having stored on the HD (instead of putting them into RAM). Ok, the stage positioning might take longer (at least not in realtime) but it would be a nice way of saving RAM. As far as I can tell, that's what MIR does though for a strange reason it allocates the RAM in advance (maybe the memory management could be improved). But I'm everything but a great programmer and I'm sure that there is a very good reason why the VSL-team decided for this solution. Maybe they will provide some more information on this. 

Though, there is one problem I have with MIR. It seems that it doesn't like my IO-Interface which is the well known original MOTU 828 (mk1). When I try starting the audio engine of MIR, a message box comes up which informs me that the audio device isn't available. Luckily I found a workaround which is to change to a different driver (ASIO4All), starting the MIR-engine and after that switching back to the original MOTU ASIO driver. This doesn't happen with any other ASIO-driven application - so it must be MIR-related. Still it's a bit annoying because it means that basically I have to restart the engine every time I load a new room.

But if we're talking in terms of quality, MIR really does it's job! Especially the stage-positioning and microphone-types are great and sound quite realistic - despite saving you lots of time as well. My favorite room is the Mozart Saal, definitely you should try it out.

Still there is one other thing I'm a bit annoyed of: MIR still doesn't support 3rd party VSTis although VSL promised to introduce this within the next free update (, which will happen soon, hopefully). But right now probably they're too busy fixing the issues with VE Pro (supporting that large array of VSTis available out there sounds like a tough task; so I don't blame them). Talking of this, it would be very nice to include the Audio/MIDI-over-Ethernet solution of VE-Pro in MIR - especially for all of those customers who already have bought VE Pro (again I'm talking of myself :-). 

Hope that helps. 

Best wishes, 

Hans-Peter.

 WoW great news Herb. I love MIR

Shakuman.

I can get the demo. However, I will need to purchase an audio card and video card to test! I have a supermicro superserver with 16 Gbts RAM and 2 x dual-core 2.6 GHz Xeon CPUs.

I was just wondering if anybody has tried MIR with such a set-up and how well it worked considering my CPUs are not i7s.

Thanks

Leif

Hello Dietz,

Thanks for your answer! Just to make sure: I'm very much aware that with real-time convolution it's not very smart to make use of any streaming technique. The ready-made/compiled IRs must be stored in RAM, but the source files could be easily located on the HD. Although I completely understand that MIR is more than "just" a reverbation unit which is based on convolution, the calculation of needed RAM doesn't really work out - for my understanding. Maybe you could shed some light on it and correct me if I'm wrong.

When MIR "compiles" / adjusts the enormous amount of IRs (which are stored on the HD) to a 4 channel IR (let's assume we are in stereo mode) and puts this 4-channel file into RAM, it might use around 10 to max. 20 MB. For each instrument this must be calculated, let's say we take the value 15MB - means on a setup with 30 instruments this sums up to 450 MB. Is it that MIR uses a different IR-format than other convolution systems, so that it would need more RAM and a value like 15 MB might be too low? Maybe that's related to the special multiphonics format MIR is using (e.g. in order to provide the microphone setup)?

If that's the case then it's still hard for me to understand why MIR is allocating a minimum of 1 GB of RAM already when the engine gets started. This would mean that it allocates the memory in advance? Doesn't it? I'm asking this because when adding instruments on a stage I can't see memory-usage growing for the MIR-processes (except when I load the samples of an instrument; but still its memory usage grows only as it would be the case with Vienna Ensemble).

Regarding the annoying MOTU 828 mk1 problem: Indeed I'm using the most up to date ASIO drivers (well; at least those which are provided officially). IMHO MOTU designs/builds very nice hardware and software for Mac whereas in a Windows environment their stuff is - put simply - crap (at least that's my impression).

Still the problem must be related to MIR, because no other Host seems to have a problem with it (e.g. Vienna Instruments. I would be so glad if we could find a solution for it.

Best,

Hans-Peter.

 you might want to read this post to get the picture (relation between voices, latency, instruments)

astonishingly the ProFire Lightbridge worked well (VISTA + W7 64bit)

depending on what supermicro server you have you should either choose a PCIe x16 grafic card (ATI 1650 or better) or if you're limited to PCI a model which supports directX 10 and openGL 2.1 with 512 MB memory or better (ATI MV 2260 or similar)

adding a firewire card try to get some with a TI chipset
another hint: run the DPC Latency Checker to find any bottlenecks on your PCI busses

hth, christian

Does it matter which version of Windows 7 (64) I purchase - will the basic version be any less powerful for MIR than the premium?

Leif

the biggest difference of the different W7 versions is that you can handle "just" 16GB of RAM in the Home Edition...

Bernd 

Welcome Kambiz, and thanks for your interest in Vienna MIR!

It's exactly as DG pointed out in his message above: MIR 1.0 is for Vienna Instruments only, MIR 1.1 will be open for 3rd-party VSTi's. It will be released in a few weeks (we are in the beta-testing phase right now).

If you want to get a first impression _now_ without "wasting" a demo-licence, you could download the MIR manual from your User Area. It's available for free.

BTW: Some new "official" audio demos are mixed already, with styles ranging from pure classical over Avantgarde to Jazz.  They will be announced soon.

HTH,

Like stated several times before - we still have the plan to release MIR on the Mac someday, but were some hurdles we couldn't overcome during the last few months. Right now, we are concentrating on version 1.1, so there will be no OSX-release ready in the near future, sorry to say so.