I have 64GB RAM, Quad GPU & 12 cores available but OpenSCAD only seems to
use max 3GB RAM, No GPU and 10% CPU.

Can anyone point me in the direction of any performance optimizations?
I'll compile the code but need a couple pointers to get me started.

Openscad does not multi-thread, so will only use 1 core.

I suspect that it would be a big job to convert to multi-threading, just
because otherwise it would have been done long ago. Even the relatively
simple step of running a separate UI thread hasn't been done.

Frank

On 8/02/2018 10:50 AM, "Matt Stark" thastark@gmail.com wrote:

I have 64GB RAM, Quad GPU & 12 cores available but OpenSCAD only seems to
use max 3GB RAM, No GPU and 10% CPU.

Can anyone point me in the direction of any performance optimizations?
I'll compile the code but need a couple pointers to get me started.

OpenSCAD mailing list
Discuss@lists.openscad.org
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

OpenSCAD runs rendering in a separate thread from the UI thread.
Turning the rendering into a multi-threaded version is work in progress and
could need more eyes: https://github.com/openscad/openscad/pull/1980

Note: This will only split the application of individual (e.g. boolean)
operators out into separate threads, not make the operations themselves
multi-threaded. The rather non-obvious effect of this is that the maximum
speedup tends to be capped at ~35% since the last union operation will
still take half of the time and run single-threaded.

-Marius

Maybe this is more of a question for github, but what about making the
interpreter multi-threaded?  Immutable variables are supposed to make that
easier.

--
Sent from: http://forum.openscad.org/

Thanks all.

I get parallel processing would be akin to a rewrite.  I was interested in
understanding if there are any configurable bottlenecks. I'll run over to
github & see what's going on there.  Love the community & love the tool.
Really appreciate the feedback!

Cheers -

Matt

--
Sent from: http://forum.openscad.org/