How to improve the display time for Preview?

I'm having a problem when previewing complex objects - any movement or rotation takes a very long time to complete (sometimes minutes) and it's getting on my nerves!

I have a fairly powerful (8-core) AMD CPU in my Linux Mint 18.1 64-bit machine, but a fairly ordinary graphics card. The system's CPU monitor shows that one CPU core is at 100% when redrawing is taking place, so it doesn't seem to be 'sharing the load'. Rendering also takes quite a while (also only using 1 processor core at 100%) although when complete the on-screen image can be moved and rotated very rapidly and without delay.

Is there anything I can do to minimise the redraw time in preview? Can I increase the number of concurrent threads in use, for example?

Note: I appreciate that preview is slow due to all the work it has to do - but I would like to see it run a bit faster, if changing some system parameters will help.

Edited 1 time(s). Last edit at 01/16/2017 11:03AM by David J.

And in the meantime... I'm making the model simpler!

It's quite likely that the nature of the task does not lend itself to concurrent processing. ISTM that each stage would need to be computed serially.

Rotating an F5 render is a lot faster than working with an F6 render, though the F5 rendering often has errors.

Sometimes I design the part in sections so that I can work on each section individually before rendering the entire part.

Dave

Concurrent processing would be nice... but, while I'm waiting for that (), I find that the Ctrl-numerical (e.g. Ctrl-0, Ctrl-9, etc.) keys help to get quick changes of viewpoint as they're almost instantaneous.

Note: for me, it's the F5 render that's agonisingly slow - the F6 render is quick and easy.

Edited 1 time(s). Last edit at 01/17/2017 10:16AM by David J.

My f5 is often quick, the f6 is boring slow.
I am considering designing a cap on the f6 key, to prevent it being pressed by accident :-(

My renders are so complex that canceling the f6 will often take as long as finishing it :-(

---

www.3daybreaker.blogspot.com

Orca V4.4 rebuild to Ramps with Mk8 and E3D, as well as a Rostock Delta Mini and an OLO in backorder :-)

Can we not request a design change ? To allow GPU processing?
So I can build a quad-quad gpu machine and render in there ?

---

www.3daybreaker.blogspot.com

Orca V4.4 rebuild to Ramps with Mk8 and E3D, as well as a Rostock Delta Mini and an OLO in backorder :-)

Quote
Replace
My f5 is often quick, the f6 is boring slow.
I am considering designing a cap on the f6 key, to prevent it being pressed by accident :-(

My renders are so complex that canceling the f6 will often take as long as finishing it :-(

Ahh - I think we're talking about different things - you're saying that F6 takes ages to complete. Same for me, but what I'm talking about is moving the image around once F5 and F6 completes. it's easy and quick to move the image around once the F6 render has finished, but the F5 render image is the one that causes grief for me.

As for requesting a design change - I did a bit of digging on the OpenSCAD git yesterday and it seems that this is a big and well-known problem looking for someone to fix it...

Just a thought, you could reduce the $fn function while you design and then re-instate it for the final render.
This works for me and make a big difference.

---

appjaws - Core XYUV Duet Ethernet Duex5
firmware 3.1.1 Web Interface 3.1.1
Ormerod 1-converted to laser engraver, Duet wifi
OpenSCAD version 2020.07
slic3r-1.3.0, Simplify3D 4.1.2, Cura-4.4.1

It only helps a little... still very slow even when working at $fn = 20.

I've noticed that F5 rendering doesn't deal with co-incident surfaces (especially involving subtractions) as well as F6.
With F5 you get "Z fighting" with shimmering surfaces which must be composed of a whole load of triangles which would add considerably to the rendering burden in the preview screen.

Perhaps that's what's going on?

I often design complex parts in sections to get over the rendering & rotation times. I make separate modules for each section and call just 1 or two modules to display while debugging the code.

Dave