HD3 not using resources efficiently. Advice??

[albee1952]

Got a simple session with around 36 audio tracks and 4 stereo AUX tracks.
All plugins are TDM(not a single RTAS plugin is in use)
Around 16 tracks are using sends to those AUX tracks for drum bus compression, room reverb, vocal reverb and delay(all plugins are stock plugins)
I keep getting playback stops with the error in screen shot and you can see the system resources window. It seems that things should be shared over the 3 cards, but it seems only the core card is actually handling anything and everything. Do I need to do something to force the system to spread the load over the 3 cards?
Session is on a separate 7200 rpm internal drive.
Session is 24 bit/48K
HD3 PCI cards in a Digi-branded expansion chassis
A single 192 IO is connected and all 16 analog inputs are working fine.

more system info:
G5 dual 2 GHz with 1.5GB RAM Leopard 10.5.8 and PT 8HD with license on gen 1 iLok.
500GB SSD system drive
160GB 7200 rpm drive for session(blank drive except for this session)
single 27" monitor @1920x1080(also tested with 24" monitor at several resolutions).
NOT connected to the internet for testing.
Tried copying the session to an external SCSI drive(15K rpm) but it was much worse)
HD3 PCI cards in Digidesign chassis(cards and 192 pass digitest).
Very new to Macs and HD systems(I have repaired disk permissions)
I am the admin and only account on this machine.

[Darryl Ramm]

Digitest OK?

[LDS]

From memory, you have a bunch of voice options available in a dropbox menu in the Playback Engine dialogue box. It looks like you have it set to only 48 voices, and have maxed them all out in your session. HD3 Accel should be able to run up to 192 voices, so I would try tinkering with that setting first.

[albee1952]

Originally Posted by LDS

From memory, you have a bunch of voice options available in a dropbox menu in the Playback Engine dialogue box. It looks like you have it set to only 48 voices, and have maxed them all out in your session. HD3 Accel should be able to run up to 192 voices, so I would try tinkering with that setting first.

I'll look for that and report back(fingers crossed!). Digitest did pass on everything.

[albee1952]

LDS gets the virtual SM58 to drop But, things are still not quite as I think they should be, so I must ask more of my teachers:
1-I tried setting to 192 voices and recording still stops quickly(with the message about reducing RTAS plugins)
2-I tried 128 voices and recording went a little further, but still not more than a minute
3-I am confused about the playback buffer setting on HD. Should I put this to 1024(or higher), and does this setting cause audible latency?(a quick test with a mic and headphones suggests that it does not) If not, what would be the need for a lower buffer setting?(at 1024, recording goes off without a hitch)
4-the system usage window shows red on the 3 HD Engines; is this merely showing that all the resources are "available" or that they are all used?
5-should 4 Dverbs(2 are mono) really burn up 100% of the core card? And a single delay plugin burn up 99% of the second card?(new screen shot)

[Darryl Ramm]

Originally Posted by albee1952

3-I am confused about the playback buffer setting on HD. Should I put this to 1024(or higher), and does this setting cause audible latency?(a quick test with a mic and headphones suggests that it does not) If not, what would be the need for a lower buffer setting?(at 1024, recording goes off without a hitch)

If the audio is staying fully TDM there is no IO Buffer use and no effect on latency.

IO buffer settings get used if you are going from the TDM to native world and back again. e.g. if you insert an RTAS plugin. No need to worry about it unless you are using RTAS plugins.

[LDS]

Originally Posted by albee1952

LDS gets the virtual SM58 to drop But, things are still not quite as I think they should be, so I must ask more of my teachers:
1-I tried setting to 192 voices and recording still stops quickly(with the message about reducing RTAS plugins)
2-I tried 128 voices and recording went a little further, but still not more than a minute
3-I am confused about the playback buffer setting on HD. Should I put this to 1024(or higher), and does this setting cause audible latency?(a quick test with a mic and headphones suggests that it does not) If not, what would be the need for a lower buffer setting?(at 1024, recording goes off without a hitch)
4-the system usage window shows red on the 3 HD Engines; is this merely showing that all the resources are "available" or that they are all used?
5-should 4 Dverbs(2 are mono) really burn up 100% of the core card? And a single delay plugin burn up 99% of the second card?(new screen shot)

Like Darryl mentioned, the native sample buffer really only matters if you are doing stuff with native plugins. If you are just using TDM plugins, you can set it to 1024 with no issues.

Red just means 100% of a DSP chip has been allocated to tasks. I'm probably over-explaining things, but DSP isn't dynamically allocated like native CPU power is. If you choose the full EQ III TDM plugin but only use the high pass filter, Pro Tools will still allocate all of the DSP to run all of the other bands and and low pass filter just the same. For this reason, a lot of Pro Tools DSP plugins have a variety of derivatives. Like a 1-band EQ-III plugin, or 3, 5 and 5 + filter versions of the Sonnox EQ. If you only need 1 band, just choose 1 band as it will save DSP.

Generally time-based effects like reverbs will use a lot more DSP than frequency and dynamics based effects. The red HD Engine bars reflect the DSP assigned to your chosen voice count. By using that dropdown menu in the Playback Engine dialogue, you can tailor your DSP usage a bit towards your session requirements. Lots of voices, set the voice count high - but it means you will have less DSP for plugins. Less voices, set your voice count low - it will mean you can't run a lot of tracks but you will have lots of DSP for plugins.

There were some great plugins for TDM systems. Try and get your mitts on Waves TDM bundles, Audioease reverbs, Softube Tubetech plugs, etc.