Running out of DSP power? System limitations?

[MichelJ]

Hi Farhoof, thank you a million for helping me. This is very interesting. I am a composer that is thrown into a production project and I have to learn by trial and error.

After I set the playback engine to max voices I opened my session again and saw that one of two revibes got inactive (not enough DSP) BUT I could duplicate tracks again and add aux tracks. Then I removed some plug-ins from the plug-ins folder that I never use and I could reactivated the inactive revibe. However I still could not add more plug-ins to tracks.

My busses are all stereo and there is no 5.1. Also there was no stereo dithered mixer.dpm, just stereo mixer.dpm.

I am only using TDM plug-ins.


These are my settings:

My playback engine:
H/W Buffer size 8192 sample
Host Processors 3
CPU Usage Limit 70%
Number of voices 96 (6DSPs)
Delay Compensation None

DAE Playback Buffer
Size 4, Cache size Large

Plug-In Streaming Buffer
Size Level 8

System Usage:
PCI 34%
CPU(RTAS) 1%
CPU(Elastic) 0%
Disk 0%

TDM
Voices Allocated 39/96
Time Slots Used 50/256

HD Core PCIe #1
HD Engine 100%
HD Engine 100%
HD Engine 100%
Revibe 100%
Revibe 100%
Waveshell DAE 98%
Waveshall DAE 93%
StereoMixer24 91%
StereoMixer24 96%

HD Accel PCIe #1

HD Engine 100%
HD Engine 100%
HD Engine 100%
Waveshell DAE 100%
Waveshell DAE 98%
Waveshell DAE 98%
Waveshell DAE 99%
Plug-In Sharing 97%
Waveshell DAE 97%

Activity Monitor Protools CPU 12.4, Threads 42, Real Memory 1.07GB, Virtual Memory 2.57 GB, Kind Intel.

I never saw it run out of CPU power so I guess I am fine there

[Farhoof]

It seems currently you only need 39 voices so you can lower the max voices to about 48 or 64 and free up a few extra dsp slots.
If this is not enough you can convert some track plugins to rtas. Keep an eye on the system usage window (CPU RTAS) to see if your cpu can handle the load.

If you don't need live input monitoring though a track, I'd convert all the track plugins to rtas. Keep the busses and revibe on TDM.

[suicune]

I would also suggest you disable DSP cache as it can cause DSPs to get allocated unnecessarily when switching sessions.

FWIW we're also still running a 1,1 albeit with HD3, and I don't really struggle for mixing power in terms of DSP or CPU as long as I slave my most intensive VIs out with VEPro. You should maybe try pick up another accel card if you can, they're dirt cheap these days.

[MichelJ]

You are not Batman's neighbour, you come straight from heaven pal.

Its working! Basically I just change my inserts from TDM to RTAS right? Does this do anything to sound quality?

[suicune]

Its working! Basically I just change my inserts from TDM to RTAS right? Does this do anything to sound quality?

No, it's just using the computer's CPU instead of the DSP card. Only difference is you can't track through RTAS plugins without noticeable latency unless you drop your HW buffer really low.

[MichelJ]

Thank you a lot for your good help. So TDM for tracking RTAS for mixing?

[MichelJ]

I changed most of my plug-ins to RTAS and now there is latency when I play back my session and it sometimes stops playing. Do I compensate that with the time adjuster plug-in or is there something I have to change in my playback engine?

[MichelJ]

Oh man!! Stupid me! There was no latency, I copied a track a bit to far on the grid. Sorry.