Just added a second Pulsar II to an existing Pulsar II with 2 STDM cables. Works fine (after a bit of hair pulling and getting them both to fit in the case) but...
With just one Pulsar II (6 DSPs), a trivial project with one wave source and one ADAT destination takes roughly 6% on 5 of the DSPs and maybe 10% on the 6th. With 2 Pulsar IIs (12 DSPs), 9 of them start off at 20-25% with the same trivial project.
I know there's some overhead involved in managing the STDM bus, but this means that (roughly) doubling the number of DSP chips has actually only achived a 30% increase in DSP power.
Not that it's a problem for what I'm doing, but it doesn't seem quite right...
Second Pulsar II - is this right?
Must be some indication problem... first time I hear of it though.
more has been done with less
https://soundcloud.com/at0m-studio
https://soundcloud.com/at0m-studio
-
ScofieldKid
- Posts: 307
- Joined: Thu Nov 13, 2003 4:00 pm
- Location: Oregon
- Contact:
Like I said, this is with one wave source and one ADAT destination - nothing else at all. I'd say that's a pretty trivial project 
Unfortunately, I can't share an IRQ between the Pulsars with this board as there's a capacitor under one of the slots that does share IRQs and it hits I/O part of the Pulsar - so the cards are not sharing IRQs. No nasty little problems though that you might run into like glitches or crackles in the audio.
Suppose I can try with just one STDM cable - you never know!
Unfortunately, I can't share an IRQ between the Pulsars with this board as there's a capacitor under one of the slots that does share IRQs and it hits I/O part of the Pulsar - so the cards are not sharing IRQs. No nasty little problems though that you might run into like glitches or crackles in the audio.
Suppose I can try with just one STDM cable - you never know!
Removing one STDM cable makes a *huge* difference to DSP load, so thanks for the suggestion (although my manual does show 2 * Pulsar II's connected with 2 cables in the diagram): I get approx one DSP back completely from doing this (spread across several chips of course).
Even so, with a blank project (ie. nothing loaded at all), there's still one DSP running at 24% and one at 21% on each board, rather than idling at 6-8% with just one board present.
Even so, with a blank project (ie. nothing loaded at all), there's still one DSP running at 24% and one at 21% on each board, rather than idling at 6-8% with just one board present.
