In SketchUp Help center under the headline “SketchUp Hardware and Software Requirements” we can read:
“Note: SketchUp will run on multiple-processor machines; however, SketchUp will only use one processor. SketchUp doesn’t support hyper-threading or multi-threading at this time”
This puts a finger to the key issue of a CAD system - it’s capacity of number crunching.
Today it seems as Autodesk’s Revit, Maya etc has enormous capacity of handling big heavy geometry files, while SkUp isn’t there (yet?).
As I have understood it, there may be three major methods to enable this in SkUp:
Multi processor use (as mentioned above)
Low level (very expensive) programming of the engine, making super fast code
Suitable geometry system (how geometry is saved in the CAD database)
In SkUp there is a pretty good work-around for big architectural projects, structuring them into sub-projects and sub-parts of geometry using the combination of hidden geometry parallel to layers. But eventually it still has a big impact of the total system usability (render time with heavy vegetation, complicated furry animals etc).
Therefore it would be extremely interesting to hear what the SkUp development team has to say in these matters. Can we expect a “super”-SkUp in the future? My impression so far is that since Trimble took over - any big things are possible for SkUp in the future! (Ok, I am aware that Ford doesn’t show their next model too soon so their competitors can adapt, but they also show their concept cars - to scare them. So maybe we still can get a glimp of the SkUp future here…)
Modeling, i.e. creating geometry based/depending on another geometry is a serial process by nature and therefore obviously cannot be split in several threads as e.g. rendering an image by multiple bandings which can be simply assembled to the complete image at the end of the process.
[quote=“cstrom, post:1, topic:35376, full:true”]This puts a finger to the key issue of a CAD system - it’s capacity of number crunching.
Today it seems as Autodesk’s Revit, Maya etc has enormous capacity of handling big heavy geometry files, while SkUp isn’t there (yet?).[/quote]
SU is a polygon modeler based on lines and faces defined by facettes whereas NURBS modelers in the CAD area as e.g. Revit/Inventor etc. are based on smooth cuves and surfaces defined by (much less) curves.
surely not because the Trimble SU division never discusses future devlopment in an public user forum.
None of the applications you mention, unfortunately, has multiprocessor support for modelling (you will discover this if you read the “small print”). They can split tasks like rendering photorealistic images into multiple threads, but you build the model using one processor thread, just like SketchUp. This is a nut that the best brains in the branch have been trying to break for more than 20 years already, ever since the first Pentium processor was released. It might be impossible.
So in this respect Revit that is my favourite BIM application has no edge over SketchUp.
How well the code has been optimized in an application is probably something not very easily quantifiable, especially with so very different applications that you are comparing.
I know the developers of PowerCADD over the years have stated that it doesn’t use more than one processor. The bottom line when buying a computer, then, is more processors doesn’t help, but more clock speed, graphics speed, etc. does, as I understand it.
For SketchUp modelling, this is how it is, but the instant you add a rendering plugin or rendering application the “surplus” processor cores become useful. Renderers can split their tasks between multiple simultaneous threads, and the rendering time needed will decrease almost in proportion to the number of threads.
I may be way off the mark here, but it seems to me that one benefit of using one thread is you can leave one instance of SU chugging away on complex geometry and still work happily in another. Even if the first one crashes your other one is unaffected.
So I imagine, perhaps, if it was possible to use multi threading you may gain speed but lose some versatility.
What I have read the upper limit for CPU speed is about 3.5GHz. That is reason for the big push for the multi processors and the thermal cooling of small units. You can do what you want but takes $$$$$$$$ plus when you make a coding change its impact can propagate thru the whole processing chain.
I have the i7-6700K CPU in my new Dell Precision T3620
Beside the fast clock speed, it’s Hyper-Threading feature delivers two processing threads per physical core.
Thus, eight parallel rendering threads from the quad core CPU.
It makes Kerkythea run like a racehorse!