Memory Leaks?

So I’ve been running a patch as silent for hours last night and out of memory in the morning.

Lots of things not connected. But a top on linux with memory on the “top” has gone up 2% in the last hour or so. It’s really just not making sound, just sitting there rendering silence. I’ll let you know if it OOMs again. Latest Rack Linux 2.6.6. Modules updated.

Cheers

EDIT: I did add in two instances of VCV Recorder, but it OOMed before that, so. Got to record and get samples in …

EDIT2: Oh yes, and the fancy side EQ on channels 9-16, and 8 spare EQ channels using MELD/UNMELD has got to be a nice idea for EqMaster maxing.

EDIT3: Test https://drive.google.com/file/d/1LrG4nJoeBJRjsBsKPIbfk9yoXIaeZ22h/view?usp=drive_link patch file.

She’s a startin’ to swap cap’ain!

1 Like

Is there a sample in the GrooveBox? Looks empty.

Sure it’s empty, but that shouldn’t stop it from not allocating memory.

Latest news is it’s at 100% more memory than when first loaded. looks like I’ll be doing a divide the modules into two partial patches later, any rinse, repeat.

It’s already swapping out the mouse motion routines.

So test A:

Removed all drums and groovebox up to the EQ and MELD/UNMELD, removed all disconnected ones on top row but not the notes. Memory of rack process 4.6%.

EDIT: looks like a rock solid 4.6%, I’ll give it an hour.

1 Like

GENERAL MAKING DEBIAN KDE NIPPY.

On Chrome I did by chance find a Settings > Performance to close background tasks. Unrelated in a way, but does clean out a bit of memory for patches. I found a GUI refresh rate of 10 fps bearable for the processing savings.

Also lowered my zram allocation and set it to lz4 algorithm. Also not related but just preparation for audio tooling.

Found the KDE microphone mute OSD bug (been found 4 times before) made a report. It might be an attempt to get device status before the scheduling of a device state change has a chance to complete. It’s kind of pointless for an OSD in my opinion.

Test B:

top row right of QNT, and groovebox removed. 4.9% rack process memory. Running …

EDIT: Exciting times … still a solid 4.9% … listening to a bit of Plastic Boy - Silver Bath and other classics whiling away.

Test C:

Just removal of the groovebox. Process memory at 5.3%. …

OK, without Groovebox, it’s been stable on memory. So I might browse the source later. Still on a crazy number of split/merges with M/S to get aux return EQ mid/sides on Jr, But yes it seems to work.

So no groovebox today. It’s almost as though @cubistguitar has psychic musical skills.

1 Like

Valley Interzone filter throws NaN. Another one for the “hide” list. Just going through some sounding options.

EDIT: went with surge XT wavetable to 4ms decay to surge XT multi-modal (spare R channel-ish).

Mangled the group bus pass through to compress some drums too (minus cymbals, hats etc).

Maybe Interzone didn’t like poly V/Oct and mono trig?