RTX 4050 graphics and Ryzen 7 vs GTX1050 and Ryzen 5 performance

[Jmstana51]

I have been using Fushion 360 on an Azus laptop with Ryzen 5, GTX 1050 graphics card and 8 Gig of memory.

I have been recently working with large STL files from a 3D scan (69K vertices).  I convert the STL to a body and then have to perform some cuts and combining with other bodies I create in Fushion.  I found the performance sluggish and sometimes crashes when I try to edit the mesh body.

I just acquired a Lenova laptop with Ryzen 7, RTX 4050 graphics card and 32gig of memory.  So far, I don't see a significant difference in performance when manipulating the same file.  Doing a cut operation of the mesh body file was faster but still threatened to stall Fushion 360 (System not responding message wait or close message).  Should I send the laptop back?  I thought the processor, graphics card and or RAM would help the problem.

[temple_dawson]

Fusion, like other CAD software (from my experience), likes faster single-threaded processors for modeling than processors with a high core/thread count (unless you're doing rendering or simulations). As you haven't specified the model of the CPU present in the laptop, it is difficult to tell what kind of performance boost you could actually expect.

Have you tested the two laptops side by side? Time how long it takes to load Fusion, open the file(s) that you typically work on, and run certain commands that you know to be sluggish on the older laptop. You may not see it, but if it takes 8 seconds on average to process a command on the older laptop and 6 seconds on the newer then the newer laptop has a 33% edge. 

You likely have seen a performance upgrade, it's just where and how you measure the improved performance.

[TrippyLighting]

Converting a .stl file directly into a solid body is a B.A.D idea!

That is particularly the case when you want to perform a solid modeling operation on the imported geometry.

69k is small for  triangulated mesh 😉

A solid model with 69k faces is huge!

Solid modeling operations are mostly single core, so more cores does not help. The graphics card makes no difference whatsoever!

[Jmstana51]

The newer laptop has AMD Ryzen 7 7840HS 3.80 Ghz and 32 Gig of Ram

The older laptop has AMD Ryzen 5 3500U 2.1 Ghz and 8 gig of ram, although the system says only about 6 is usable.

I am trying to modify an STL file from a 3D scan.  Most of the features are too complicated to model, hence the need for the 3D scan.  I just need to clean up a few defects and replace one cylinder with a more robust one.  I had been experiencing freezes and slow performance.  I can understand the processor speed being a limit, but I thought the memory would prevent the freezes.  Trying to compare the two on the same model right now.  The older one froze and the newer one didn't freeze but not sure it ever got finished either.  Based on just the processor speed, I would expect things to take close to half as long.  I also am surprised the graphics card doesn't seem to help at all.

[Jmstana51]

I am trying to modify an STL file from a 3D scan.  Most of the features are too complicated to model, hence the need for the 3D scan.  I just need to clean up a few defects and replace one cylindrical feature with a more robust one.  I had been experiencing freezes and slow performance when trying to add the two models together. 

For my next attempt, I have reduced the STL file even more without affecting the accuracy of the model.  Converting to a solid model went much faster.  I am only just starting to manipulate or edit the file to see if I run into the same issues.  I am just shocked that more RAM would not at least prevent freezes.  How do I tell the true model size to see how big the model really is?  Is saved Fushion file size an indicator?

[temple_dawson · Accepted solution]

I agree that a processor with twice as fast single-threaded performance should see a much larger improvement. Along with the added memory which should help if the model is pushing Fusion to the limits of what your old system has. It may be possible that you are pushing the limits of what Fusion itself can handle, regardless of local processing power.

I personally have run into a couple of designs that stressed out Fusion to the point of crashing regardless of local processing power (one with a 3600XT and another with a 13700K). One was a large part with a sketch pattern of various sized holes (well over 1000 sketch objects) from a DXF inserted into the design that would make Fusion hang and often crash. I ended up having to divide the DXF and make multiple sketches for the design to work.

Yes, regarding the GPU, all it does it display. The way you would tell if your GPU is under-spec is by rotating around the model and checking for stutters, and while doing so have Task Manager or a monitoring software open to watch GPU usage.

[TrippyLighting]

---

@temple_dawson wrote:

I personally have run into a couple of designs that stressed out Fusion to the point of crashing regardless of local processing power (one with a 3600XT and another with a 13700K). One was a large part with a sketch pattern of various sized holes (well over 1000 sketch objects) from a DXF inserted into the design that would make Fusion hang and often crash. I ended up having to divide the DXF and make multiple sketches for the design to work.

---

There are hundreds of threads on this forum that advise folks not to import DXF (and SVG) files 😉