Digi CS, HD voice allocation vs DSP question?

[Rich Breen]

OK, I give up! I've perused the manual, but still can't seem to figure it out: In the "Playback Engine" dialog box, what's the difference between, for example, "64 voices / 2 DSP" and "64 voices / 4 DSP" (likewise, 32/1 vs 32/2, 96/2, 96/3), and why would you choose one over the other?

many thanks,
rich

[Ed Kish]

This is a good question that needs an answer.

Does it just hold the chip so it is available when you need it?

[Allan Speers]

I've tried asking FOUR seperate Digi techs, about this question and the rest of the playback setup.

Basically they just send me back excerpts or paraphrases from the manual, which explains nothing.

other, related questions:
Why does the manual suggests allocating more cpu resources to Protools when you are running a lot of NATIVE synths via direct-connect? Absurd, but there it is in print.

Here's another: Why does SSC have a selection for ram-allocation, when the app will only take what it needs based on the samples actually loaded? The setting seems superfluous at best. (and why does the manual state that SSC has a max of 64 voices, while the polyphony selector goes to 96 voices?

The list could go on for pages.....

Digi evidently feels that we don't need to know how to fully optimize our systems. It's been this way for nine years now.
=======================

BUT TO TRY TO ANSWER THE ORIGINAL QUESTION:

The one thing I can see is that using more chips lowers the amount of PCI overhead. try changing settings while running a large session with the "dsp resources" window open.

Obviously, this setting would be extra critical if you are running a video-capture card on the same bus. It may not matter at all without video, but who knows?

How much bandwidth is dangerous? Is 50% ok? 98% ? Are there other trade-offs, something like "lowering your pci bandwidth will allow you to run soft-synths with lower latency settings" ?

[ProdInfo]

Originally posted by Rich Breen:
OK, I give up! I've perused the manual, but still can't seem to figure it out: In the "Playback Engine" dialog box, what's the difference between, for example, "64 voices / 2 DSP" and "64 voices / 4 DSP" (likewise, 32/1 vs 32/2, 96/2, 96/3), and why would you choose one over the other?

many thanks,
rich

<font size="2" face="Verdana, Arial">Hi Rich,

We've passed your question on to one of the folks at Digi that might be able to better explain the reasoning behind this. It might take a few days to get the answer and post it (since it's a weekend), but we'll try to get the info for you as soon as we can.

Thanks very much!

[Rich Breen]

Originally posted by ProdInfo:
Hi Rich,

We've passed your question on to one of the folks at Digi that might be able to better explain the reasoning behind this. It might take a few days to get the answer and post it (since it's a weekend), but we'll try to get the info for you as soon as we can.

Thanks very much!

<font size="2" face="Verdana, Arial">Many thanks - looking forward to a definitive answer on this!

rich

[JimWest]

One reason you should use fewer voices for each DSP is if your PT rig is part of a Unity MediaNet network. From http://www.digidesign.com/support/do...s_MediaNet.pdf pg. 17, "Pro Tools|HD workstations have multiple playback modes at different sample rates. For optimum performance in a MediaNet workgroup, it is recommended that you use the least number of voices per DSP."

Jim West
Digi Engineering

[Allan Speers]

Jim,

thanks for the link. I suppose this is worth reading, but of course it doesn't really answer the question.

Those of us NOT using unity need to understand the pros and cons of this setting, so that we can properly optimize the rest of our systems.

the same goes for ALL of the hardware and software settings / choices.

This kind of information is more now than ever, due to the complexity of HD's settings. Also because so many more demands are being placed on the computer than in years past.I have started a new thread, with a list of specific questions on this topic.

http://duc. digidesign.com/cgi-bin/ubbcgi/ultimatebb.cgi?ubb=get_topic;f=16;t=015005

Hopefully that will make it easier to address the specific issues. PLEASE go to this thread and get us the answers we need!!!!

-Thanks.

[Digi Engineering]

Speer - you are correct.
The one thing I can see is that using more chips lowers the amount of PCI overhead. try changing settings while running a large session with the "dsp resources" window open.

With the problems of DAE -6042 errors, we decided to provide more options for "tuning" your PCI usage with HD. If you are not having -6042 problems, then it makes sense to use large voices-per-DSP (fewer DSPs used for the audio engine). But if you are running into -6042 "PCI bus too slow" errors with a video card, Unity, an expansion chassis, or some other PCI card, then using fewer voices per DSP can allow you to get just as many tracks at the expense of using more DSP for the audio engine.
-Toby Dunn, Digi SW

[Rich Breen]

Originally posted by Digi Engineering:
[QBWith the problems of DAE -6042 errors, we decided to provide more options for "tuning" your PCI usage with HD....
-Toby Dunn, Digi SW[/QB]

<font size="2" face="Verdana, Arial">Toby - That's great, makes perfect sense - many thanks for the explanation!

If you have a little more time to devote to this, it's not entirely clear to me why distributing voices over a larger number of DSP chips reduces the host processor's PCI traffic (if the chips used were on different process cards, I would think it would actually *increase* PCI traffic, and if the chips were on the same card, I don't see how it would make a difference.) - if you can, an explanation of the actual mechanics would be useful to me, if not - no problem!

Again, thanks for the useful information!
rich

[Allan Speers]

Thanks, Toby.

Are you saying, then, that if I am not getting -6042 errors, then there is NO OTHER benefit to lowering the pci bandwidth? no benefit to soft-synths or HTDM / RTAS?

-And PLEASE post on my other thread. I've laid out some of the basic questions, to make it easier to explain the various settings. So far their hasn't been a single response. [img]images/icons/frown.gif[/img]