01:46 <InPhase> TylerTork: I get superb OpenSCAD performance by two main features, which are lots of RAM (for larger more complicated designs), where I think at least 32GB is pretty important for good scope (I have 64GB, but rarely need that much for OpenSCAD, using the rest mostly just for local LLMs). The other is good thermal regulation, which is important for laptops so that they don't throttle.

01:48 <InPhase> TylerTork: I think RAM bus speed probably matters a lot as well, just from my general understanding of how these sorts of computations typically operate, but I haven't explicitly benchmarked a comparison of that.

01:49 <InPhase> TylerTork: I suspect you'll find more impact from RAM bus speed than from cpu speed.

01:49 <InPhase> Particularly on larger more complicated stuff. There are limits to what you can cache with this sort of thing.

01:51 <InPhase> TylerTork: Like JordanBrown, I'm also using an integrated card. I think a fancy-pants 3D card will only help for preview of complicated designs, and preview is rapidly losing utility. We're even discussing purging that path in the future (sometime after the next release).

13:01 <TylerTork> InPhase: I want to make sure you're aware of the use cases I'm finding for preview so you can maybe accommodate those in some other way. I've been treating preview mode as the user's opportunity to evaluate whether the parameters are correct -- check parts for fit, see another object for scale, etc. Then when they render I align the parts as they

13:01 <TylerTork> should be for printing and remove any annotations, scale objects, anything that's not needed for printing.

13:02 <TylerTork> Thanks all for the hardware advice.

13:02 <gbruno> [github] pkriens opened pull request #6022 (Provide $this and automatic context for functions in objects) https://github.com/openscad/openscad/pull/6022

13:03 <gbruno> [github] pkriens converted_to_draft pull request #6022 (Provide $this and automatic context for functions in objects) https://github.com/openscad/openscad/pull/6022

13:05 <InPhase> TylerTork: The reason for purging preview would be because it's stopped being faster.

13:09 <gbruno> [github] pkriens edited pull request #6022 (Provide $this and automatic context for functions in objects) https://github.com/openscad/openscad/pull/6022

13:41 <TylerTork> InPhase: I get that -- I just think that besides the speed difference, there's some benefit to having a built-in feature to toggle between seeing things arranged for evaluation, versus arranged for printing, without having to modify a parameter and have unsaved changes, which raises the question whether that's the only unsaved change and thus is it

13:41 <TylerTork> safe to save.

15:19 <gbruno> [github] pkriens synchronize pull request #6022 (Provide $this and automatic context for functions in objects) https://github.com/openscad/openscad/pull/6022

16:14 <pca006132> for spec, you can try to check RAM overclock compatibility stuff

16:15 <pca006132> I think some RAM has built-in profiles for intel/AMD CPUs, but may not work under some motherboard combination, and when you get to 128GB RAM overclock may not work

16:16 <pca006132> you can probably get 10-20% better performance from this for manifold

16:16 <pca006132> (when the model is big, i.e. requires several GB of RAM to eval)

16:18 <pca006132> I wish I could have time to look into a GPU implementation of manifold, but it will not happen in a few years I guess

16:24 <pca006132> oh, and one point is that I don't use windows, so manifold is probably not well optimized for windows

16:25 <pca006132> and probably for openscad devs as well

16:25 <pca006132> so if you use linux, it may be faster

17:37 <nihil> everything is faster on linux

17:37 <nihil> except the crash and compromise rates

17:42 <sculptor> you shoot your foot faster on linux

17:49 <J25k89> windows: do you really want to shoot your foot?   linux:  you have successfully shoot your foot.

17:51 <JordanBrown> Linux doesn't have a GUI for foot-shooting.

18:23 <nihil> doesn't need a GUI for it, there are a million different ways to shoot your foot, and they are equally efficient