@cm said:

the X58 motherboards are not compatible with ECC memory modules (see above) and i can't confirm the latency issue for the 5520

christian

Yes I agree, X58 isn't compatible with ECC, but it doesn't have to. The memory controler isn't in the chipset but, in the Cpu. Ecc support is one of the few difference between the Xeon 35xx/55xx and the Core i7. So you can use Ecc memory on a standard X58 motherboard if you use a Xeon.

But.......... but the thing I didn't notice is that you can't use ECC registered sticks on a X58 motherboard (I don't know why), and non-registered ECC sticks are more expensive than non ECC.

@Trailerman said:

Spec is Supermicro X8DT3, 2xIntel Xeon W5580 (3.2GHz), 24GB Ram in 4GB modules (this way the memory bus can still run at 1333MHz because there's only 1 DIMM in each channel - this mobo has 12 memory slots), Vista 64bit Ultimate, mixture of 1TB SAS and SATA drives, 512MB 9500GT Graphics etc.

I have very bad results with the S5520, but everything is ok on a X8DAi, so I think you shoudn't have any problem with the X8DT3.

 Thanks Stephane - I feel better about the motherboard choice.

You're right that the memory situation is a little confusing, and the wild variations in price between buffered, unbuffered, ECC, non-ECC, registered etc. doesn't help.

Jules

IMO the most important specification for an intel 5520 based motherboard is that it offers a PCIe-x16 PEG slot for grafic card ...

for the memory i can only repeatedly recommend to strictly stick to the compatibility list of the motherboard and actually have a close look at the possible combinations of memory modules (dual rank, quad rank, buffered, unbuffered, ECC)

christian

Hi Christian

Can I ask why you feel an x16 slot is so important?  I would have thought it should only really be necessary if you're gaming.  IS there something I should be aware of here?

 I think the board my system builder is currently using has an x8 slot in an x16 housing, although I'm trying to confirm.  This is fairly critical if it's going to cause a problem.

Many thanks for your help.

Jules

just a practical issue - while the available range of grafic cards for x16 PEG slots is huge the assortment of models for PCIe-x1 is very limited.

i also noticed that not all grafic cards are working in x8 slots or vice versa not all x8 slots support grafic cards (we already have a nice collection of not working combinations resp. combinations working only in 16bit mode and/or VGA resolution here ....)

of course this is only important if you need to have a monitor or two attached ...

christian

Many thanks Christian - that's very helpful.

As soon as the bare bones of the system is in place, I'll ensure that the graphics slot and card are fully compatible and get the builder to test properly in 32bit mode.

If I understand you correctly, everything should be ok if there are no graphics card compatibility issues and it delivers solid performance at 16 and 32bits or are there other graphics specific issues I should be aware of?

Many thanks again for your help.

Jules

you wouldn't like to watch any current OS or applikation at 16 bit (coulor depth);-) MIR needs the grafic card to have proper support  for openGL and directX10, but that shouldn't be topic ourdays ...

christian

 Thanks Christian - all clear.

I was concerned that there was some technical reason why the motherboard needed an x16 slot; perhaps your tests were suggesting any system running this amount of processing and sample-streaming simultaneously suffered PCI bus overloads if graphics were not run over an x16 interface.  It sounds like it's more a general compatibility issue, which I'm sure we can cope with.

Sorry to have pushed you for more details.  It would just break my heart if this system had a fundamental issue which prevented it from running Mir.

All the best

Jules

you're welcome, i'm happy providing this information to avoid other users running into the same issue than me - we had to try 4 grafic cards on an X7 motherboard before receiving a useful result.

christian

@Dietz said:

Re: RAM

In a stress test I ran today, I had 50 indivdual VIs playing glitch-free at the same time in stereo, latency set to 1024 samples, CPU @ 80%, with a system comparable to the one you outlined above (12 GB RAM). I used 11.2 GB RAM for this setup, with Nuendo 4 for MIDI output on the same machine.

Taking into account MIR's abilitiy for dynamic processing (only those VI's that really play will use MIR's convolution core and thus tax the CPU), you will most likely run out of RAM before the CPU overloads. 😊

HTH,

If I get an Asus P6T, and want more than 12GB ram what shall I get? Are there any other kits available rather than the Kingston 12GB unECC kit of 3, that costs over $1000? thanks

Hi  Christian and the rest of the Vienna team.

First of all: Congratulations with getting MIR on the market - a big step for Vienna, but a giant leap for mankind -at least composers etc.[;)] 

DG raised an importent question about latency, which sort of got lost in the mail. Running with 1024 is sort of ok when rendering a project, but as DG pointed out: What about the recording phase - when playing, 1024 is simply too much delay for comfort.

Can you operate with a simple reverb setup during recording, allowing you to set latency to say 256 or better still: 128 ?

And then afterwards turn on all the bells and whistles.

Achievable latency is always going to reflect the specifications of your system and the complexity of the project you're running within Mir.

For example, I'm running Mir at 128ms hardware buffer and have no problems whatsoever with what I consider to be decent sized arrangements - say 50 VI instances containing fully scripted matrices, with up to 25-30 of them playing simultaneously.  If I had to do something much more involved I would increase the buffer size if I saw CPU load issues.  That said, I have a very fast quad-core Xeon system which was specified primarily for Mir.

In essence, you're absolutely right, you can always run at a lower buffer setting whilst recording indivdual parts, mapping out ideas and basic arrangements etc. and if you find further down the line that the full arrangement is stressing the system, increase the buffer size when you render the final mix. 

Jules

@hose said:

If I get an Asus P6T, and want more than 12GB ram what shall I get? Are there any other kits available rather than the Kingston 12GB unECC kit of 3, that costs over $1000? thanks

@Trailerman said:

Achievable latency is always going to reflect the specifications of your system and the complexity of the project you're running within Mir.

For example, I'm running Mir at 128ms hardware buffer and have no problems whatsoever with what I consider to be decent sized arrangements - say 50 VI instances containing fully scripted matrices, with up to 25-30 of them playing simultaneously.  If I had to do something much more involved I would increase the buffer size if I saw CPU load issues.  That said, I have a very fast quad-core Xeon system which was specified primarily for Mir.

In essence, you're absolutely right, you can always run at a lower buffer setting whilst recording indivdual parts, mapping out ideas and basic arrangements etc. and if you find further down the line that the full arrangement is stressing the system, increase the buffer size when you render the final mix. 

Jules

@hose said:

If I get an Asus P6T, and want more than 12GB ram what shall I get? Are there any other kits available rather than the Kingston 12GB unECC kit of 3, that costs over $1000? thanks

Hi, It's not the ECC that is the problem (the i7 can just ignore that) but rather the Registered and unRegistered RAM modules. At the moment the 3x4GB DDR3 unRegistered modules aren't really available, although you may see them advertised, they are just hugely expensive first run if you can find them at all.

This will change in the not too distant future but at the moment you do have to go with Registered dimms if you want 3x4GB kits and that means Xeon 5500 series.

Something I will get around to testing shortly is running a 5500 series xeon on a standard desktop i7 motherboard, which does work apparently, but I wonder, because the memory controller is on the chip if that means it will then take the Regsitered dimms, I doubt it but it is worth a try when I get a spare min.

Back to your question though, you will have to wait to use more then 12GB on the desktop i7 systems, probably until the end of the year.

Thanks for the info, and welcome to our little club....! Nice to see you here.[;)]

DG

txh Jules for responding.

I suppose that maybe a dual XEON 5580 can keep up with a full orchestra at 128, so with this system it is not a problem, but what about less powerfull systems like i7 platforms? - here a different approach is needed.  

What I imagine is that you could have all the instruments loaded in MIR without all the positioning/reverb turned on thus preserving power, but with some sort of basic reverb turned on, during the recording process.

The only thing that could be very frustrating is if you have to setup the MIR positions for every project, when rendering, so is it possible to have a template with all your complete orchestra including their positioning, directions etc. but with an on/off switch. On being - "yes compute all the MIR stuff" and off the "bypass the MIR reverb stuff"?

@DG said:

Jules, for me the point is how do you run at a lower buffer whilst programming? Do you switch something off? Can you switch something off? My sessions have around 120 instruments, often with 70 or so playing at the same time, so this is really important to me. Currently I just work without reverb and plugs whilst programming, and then raise the buffer for mixing.

DG

 Hi DG, Sapkiller

Obviously, if a system doesn't really have the grunt to run MIR, then it's going to be a struggle from the get-go.   You also can't (I don't think) turn off the reverb component of MIR altogether and work without it until you're ready to mix, although there are different rooms and some may be heavier or lighter on the CPU than others (again I'd have to check this).  The MIR engine and audio engine seem to be very tightly integrated, so I'm not sure how practical it would be do disable the MIR engine altogether, although I agree that perhaps having a 'dummy' room which reduced the MIR overhead as much as possible would be very useful.  

The number of instruments loaded in MIR doesn't really seem to have a big bearing on CPU overhead - it's how many are playing simultaneously that determines CPU load.   Therefore loading up your template shouldn't be a problem.   If the system could not cope with running the entire arrangement at the buffer size you wanted to work at, you would need to do your MIDI input work within more managable sections, ie. muting any less critical parts whilst you wrote. 

The alternative is to run everything at a higher buffer setting - I ran some tests at 512ms and couldn't get anywhere near the ceiling of the system you see in my sig (50 VI's running simultaneously, all loaded with perf-legato-all, perf-legato-speed or perf-universal matrices - ie. lots of complex scripts running - used about 55% CPU load).   I also found that working at 512ms was perfectly acceptable for writing; heck I've spent years programming beats and doing dance production at 512ms and above and never had a problem.

Taking that further, if I had a slower system, I personally would be happy working at 512ms when writing, muting the odd part and living with ocassional overloads and then increasing the buffer to say 1024 when I was ready to mix.   I guess the answer is to keep an eye on the performance people are seeing from different setups and try and find a performance sweet-spot that fits the budget.

I guess the bottom line is that if you feel you need to work at very low buffer settings (256ms and below), use very large arrangements, and can't bear the thought of muting a few parts whilst you write, then you will need a very powerful system to work within MIR, or you would need to work in VE and then transpose your arrangement into MIR when it's complete. I'd say this is a fairly extreme scenario though.

I hope this helps a little.

Jules

@Sapkiller said:

but what about less powerfull systems like i7 platforms?

such systems can run reasonable arrangements (like the zarathustra in the video) very well - i'd say the higher the demand, the more powerful the tools need to be.

but if i'm not wrong there actually _is_ a bypass button ... christian