PTHD Processing Allocations

[NathS101]

hey guys I'm running PTHD v8 with an HD2 system on a 2006 MacPro (1,1; 2x 2.66GHz dual core intel xeon; 14Gb ram) and keep running out of processing power on high sample rate &/or high track count projects.
The best solution I can come up with is to buy another Accel card instead of buying a new MacPro or upgrading to HDX...
What else would you suggest?

Also I keep keep thinking there must be a better way to structure my track count/buss & hw insert mixing delays/DSP/voice/TDM/CPU allocations...
What do you guys normally do??

Cheers,
steeley

[cLoudBase]

I'm imagining you're running out of voices ?

I tend to run at 48k to get the full 192 voice availability for complex sessions.
44.1/48 = 192
88.2/96 = 96
176.4/192 = 48
Using RTAS plugins will also help with TDM resource allocation to maximise voice availablity.
Another Accel card won't add any more voices, but will give TDM plugs more resources.
You could get 256 voices @ 48k or 128 voices @96k if you upgraded to a HD Native or HDX card using 10HD or 11HD.

Cheers
Matt

[Sammas]

Perhaps stating the obvious - but incorrect use of RTAS plugins will chew through voices much quicker as well.

- RTAS on AUX tracks will chew through 4 additional voices.
- Using RTAS after a TDM plugin on any channel will also chew through additional voices.

[cLoudBase]

Here's a couple of links to help explain voice count issues & best methods of RTAS & TDM use.

http://protoolsformedia.blogspot.com.au/2009/08/where-have-my-voices-gone-on-tdm-pro.html
http://duc.avid.com/showpost.php?p=1813823&postcount=2

[NathS101]

Nath, It wasn't obvious as I have never heard that...!
Matt, I'll check out those links.
Thanks guys.

Anyone else have some words of wisdom?
cheers

[Sammas]

A voice is essentially a "lane" to or from the native cpu & the TDM DSP architecture.

Imagine the signal chain to look something like: harddrive -> native CPU -> TDM DSP -> TDM mix engine including sends & returns -> audio I/O.

A mono track requires a "voice" every time it commutes between the TDM DSP system and the native system. So in the event that you place an RTAS plugin after a TDM plugin - it uses 3 voices. One to send the audio from the harddrive to the TDM system (where the TDM plugin is), another voice to send the audio back to the native system for the RTAS plugin processing, and the third to then send the audio back to the TDM mix engine. If its a stereo track, it doubles!

If you pile a track full of just RTAS plugins, then add a TDM plugin afterwards, it will only use 1 voice as it only makes the 1 trip to the TDM DSP system.

In one of the menus for setting up your system allocations, you can change the way your mix engine works as well. IE: Despite having a HD2 system, you can actually set the system to have just 48voices on 2 DSP chips, etc.

It can be fiddly business on HD2 accel! It gets quite a bit easier if you load the system with a core card & 2 HD accel cards.