Speeding Up Manouvering In 3D

Walter Vardanega

I find Bricscad BIM can be a bit slow when maneuvering about/within a model.
I think it has to do with the Visual Style setting. I usually use the Modeling visual style.
What settings do other user's use?

CADMebel

Yes, it is slow - not just BIM, but BricsCAD overall is slow when the file gets a little bit heavier. All the dwg based clones suffer from the same problem whit not so large 3D files! Make 2000 spheres and try to manipulate them with the manipulator ...

Michael Mayer

Vectorworks feels more fluid with the same DWG Imports.
I have a current, not so bad, RTX 2070 with 8 GB VRAM.
I assigned my 3D Apps to not save energy in Windows but best
GPU performance. Nevertheless, I have never seen the NVidia using
more than 15% GPU or more than 20% VRAM in my 3D Apps.

But there are a few things that CAD Views don't like :

Many Meshes or separate 3D Polygons.
(Once imported Solids with some Mesh Tree instances, unusable.
When I switch Trees off everything was fine again)

Revit imports which create Block from each Family Member,
mostly used for a single Instance in the File only.
This adds extra redundant complexity.
When I explode all Blocks, it works much better,
although there are basically more Elements for the GPU to handle.

Or exploded Hatches that result in millions of lines would be a
nightmare.

CAD Views like Blocks for any repetitive Element in File over just
simple copying or arrays.

Also there are some some deeper OpenGL View Setups in Settings.
I think it is worth to ask the support what setting would be best
for your specific hardware.

And CAD require a proper GPU for complex files. Especially they need
enough VRAM that the whole Model fits in. If it is too less it will get
really sloooooww.
Discrete GPUs (from AMD or NVidia) are much faster than
Intel Onboard Graphics, which even share slower RAM with the CPU.
Mobile GPUs in Laptops are much weaker than their Desktop counterparts.

Hans De Backer

Michael,

help!
The last thing you should do to improve performance is explode blocks.

It is the other way around: using blocks allows to benefit from using shared resources for repeated geometry, instead of wasting resources ( = RAM and CPU cycles) by exploding existing blocks.
Of course, if these blocks are merely containers for grouping, without any repetition, there will be no benefit, but from the moment there are 2 inserts you are making considerable benefit.

While we're at it, let's get another widespread misunderstanding out of the way.
For dwg-editor-applications, the bottleneck for the majority of operations is the CPU, not the GPU, it plays a really minor role. Any decent mid-class GPU will do fine, it is the CPU that matters, and more specifically, the Single Thread performance of the CPU, not the Multi-threaded performance: due to the way dwg data is organized (heavily inter-dependent) it is hostile by nature to parallel processing. Regardless how much effort we have already spent in this area, and we really did, there are only a few operations on dwg drawings that (can) actually benefit from having multiple cores available.

Concerning manipulating 2000 spheres: thanks for pointing out that the manipulator-implementation has room for further improvement.
The reason it takes a while till the manipulation can be started is that first temporary clones are created for all spheres, and only then, the manipulation can begin. Once started, the performance is ok.
The legacy MOVE and ROTATE commands have been optimized to deal with such operations on large datasets, they start moving/rotating rightaway, displaying only a partial preview of the dragged entities. We'll investigate if this approach can be migrated to the manipulator.

Michael Mayer

@Hans De Backer said:
Michael,

help!
The last thing you should do to improve performance is explode blocks.

Hi Hans,
that is exactly what said*

*or at least meant and I tried to write.

" created Block from each Family Member,
mostly used for a single Instance in the File only.
This adds extra redundant complexity.
+
When I explode all Blocks, it works much better,
although there are basically more Elements for the GPU to handle.
+
CAD Views like Blocks for any repetitive Element in File over just
simple copying or arrays."

Beside my dis/ability to explain or not,
I agree my current Project was a bad example.
Not everyone will get Revit exports done sorting by Families ....

Yes,
I am aware of problems to parallelize CAD stuff for CPUs.
Therefore I am more than happy to see the increasing number of
tasks available in Settings for Multithreading in Bricscad.
It is nice to see my CPU at 100% instead of 7 % when loading Files.

And as single core power didn't get noticebly faster over the last
decade I am a bit scared how we can work with higher BIM LODs
in the future ....