Not enough DSP memory
This has got nothing to do with your computer. There's small amounts of memory on your DSP card. It's used for ie. Optimaster's look ahead function and other stuff. I think also short delay effects use this DSP memory. It's way faster, but to expensive to put loads on a board. Also some of the program/devices will be stored in that RAM for processing I think.
So you could try to disable or shorten look-ahead function, load Delay L(ong) modules instead of the short versions etc. to free up that onboard RAM.
So you could try to disable or shorten look-ahead function, load Delay L(ong) modules instead of the short versions etc. to free up that onboard RAM.
His problem is not PCI overflow, but shortage on onboard DSP memory. I encounter it too once in a while and it is a limit to your system, just like DSP capacity, system RAM, etc.
It's an annoying error, but can usually be solved by reloading DSP or lowering used DSP memory. Freeing up DSP memory can be done by turning of look-ahead functions, using delays of long type,... More examples are always welcome.
It's an annoying error, but can usually be solved by reloading DSP or lowering used DSP memory. Freeing up DSP memory can be done by turning of look-ahead functions, using delays of long type,... More examples are always welcome.
On 2003-01-24 08:49, Immanuel wrote:
I believe, that the scope boards have 32MB memory on them.
This 32mb RAM is not the same as on-chip DSP RAM. In fact, all the knowledgable Pulsar people I've spoken to so far either do not know what this RAM is for, or do not think it is used at all.
Anyone know for sure?
peace
We've done some tests, during a previous quest for the 32MB Scope RAM, but PCI Geiger didn't show any difference on PCI loads. So back then, we concluded it was not used... I'd be glad to hear I'm wrong.
more has been done with less
https://soundcloud.com/at0m-studio
https://soundcloud.com/at0m-studio
the DSP ram is used for the task running on the chip, so a different distribution of tasks might indeed have an influence - but I wouldn't hold my breath on it.
Like at0mic wrote above, 8 different masterverbs are far from practical use, but of course you're free to design whatever sound you like
cheers, Tom
Like at0mic wrote above, 8 different masterverbs are far from practical use, but of course you're free to design whatever sound you like

cheers, Tom
-
- Posts: 777
- Joined: Sat May 25, 2002 4:00 pm
- Location: The Great White North
- Contact:
Has anyone uncovered more info on DSP memory?
I've got a huge project. No look-aheads. Only one delay effect. No reverbs. 1 sampler and 1 EDS16. Mostly it's mixers and lots of audio interconnections.
I get the 'Not enough DSP memory' message when making a MIDI connection. (Midi Merger to a mixer, for automation.) It doesn't seem to matter what I attach though, every possible connection seems to be a straw that breaks the camel's back.
Is it possible that all interactions between DSPs go into an on-chip memory buffer? It would make sense from a programming standpoint...
Am I likely to improve things by increasing # of DSPs? I.e. upgrading again? Spoimala, is this the approach you ended up taking?
Any other suggestions?
My project is obscenely large. I ran out of DSP so ordered another card. Ran out of bus width so ordered another S/TDM cable. (Thanks Shawn and Ali! Ali helped me out on this one during a statutory holiday. Infinite Vortex sent me a free cable. Cool.
)
Now I'm out of DSP memory but I've already removed as much functionality as I can afford to.
Has there been any new thinking on this problem since the last post (over a year ago)?
Thanks and cheers,
Johann
I've got a huge project. No look-aheads. Only one delay effect. No reverbs. 1 sampler and 1 EDS16. Mostly it's mixers and lots of audio interconnections.
I get the 'Not enough DSP memory' message when making a MIDI connection. (Midi Merger to a mixer, for automation.) It doesn't seem to matter what I attach though, every possible connection seems to be a straw that breaks the camel's back.
Is it possible that all interactions between DSPs go into an on-chip memory buffer? It would make sense from a programming standpoint...
Am I likely to improve things by increasing # of DSPs? I.e. upgrading again? Spoimala, is this the approach you ended up taking?
Any other suggestions?
My project is obscenely large. I ran out of DSP so ordered another card. Ran out of bus width so ordered another S/TDM cable. (Thanks Shawn and Ali! Ali helped me out on this one during a statutory holiday. Infinite Vortex sent me a free cable. Cool.

Now I'm out of DSP memory but I've already removed as much functionality as I can afford to.
Has there been any new thinking on this problem since the last post (over a year ago)?
Thanks and cheers,
Johann
My current project's been recycled many times, and now I can't make any connections in the Project Window anymore, as if the second click never takes place. I can still make the connections from the Livebar, luckily. I'm afraid this workaround won't do when you get that error. I'm not getting any of the DSP memory problems. Adding a card could spread the load, and you could make some more connections. Have you tried reloading/reorganising the DSP by cycling samplerate to slave and back to master?
-
- Posts: 777
- Joined: Sat May 25, 2002 4:00 pm
- Location: The Great White North
- Contact:
-
- Posts: 777
- Joined: Sat May 25, 2002 4:00 pm
- Location: The Great White North
- Contact: