Is Bambu Studio highly dependent of the graphic card performance ?
The PC I’m using to print is a i7-4790 4GHz with 32Gb or ram, and Bambu Studio is very slow to slice and to display.
I’m using the mobo GFX card.
Will a GFX card improve the slicing and display speed to make it pleasant to use ?
I was sure it would be, but I just checked it on my macbook now - slicing a bunch of plates and looking in Activity Monitor. It sometimes hogged all CPU cores (fluctuated between 800-1300%) but barely used the GPU (~3%). Then again I’m not sure that was a good tool for measuring. For all I know it may have monitored the wrong GPU of the two, the one that happen to be inside the intel CPU rather than the AMD Vega (or both), and I also don’t know if there’s some odd GPGPU standard that may be missing under macOS (but I wouldn’t think so).
OK, but I’ve a PC, so hard to compare I think.
What’s strange is that the GPU is used at around 1% when I’m on Prepare pane doing nothing, at around 80-90% when I’m moving, zooming,…
And it is at 100% when I’m on Preview pane when I’m not doing anything at all.
Not at all. We run the very same code.
I don’t think the GPU is used for slicing, just for displaying. Displaying is absolutely heavily dependent on the GPU. That’s an OS thing, though, not Studio.
Slicing is memory intensive. Depending on the size of the model, 32GB may not be very much memory. And the OS is going to use a lot of that. So your system may be going to disk a lot, which will really slow things down even if you have a SSD.
My 5 year old i9-9900K system has a 2070GTX video card and 128GB of memory and Studio (or my CAD software) isn’t sluggish at all. For a really big STL it might take 10-15 seconds to load it all in. But once loaded, everything else is quite zippy.
I’d upgrade to a dedicated graphics card. Built-in GPUs are almost always a compromise that don’t perform well with heavy workloads. The hardware might be as fast as a video card, but it can’t run that fast because it can’t get rid of the heat.
That’s not strange at all. The preview pane is very simple graphically - just shading the triangles on the surfaces. When you are on the preview pane it’s rendering every segment of every gcode instruction in the part. FAR more complex rendering.
OK, but requiring 100% of the GPU when static, and even GPU at 100% when the Bambu Studio window is minimized in the taskbar is not really useful and optimized.
There is room for performance improvement most probably.
I’ve 2 instances of BStudio opened, with objects having more than 12 000 000 triangles total, and still have 18GB available, so I think 32Gb is far far enough for the kind of models I’m printing.
Will buy a 2nd hand GPU card to improve the rendering speed.
The GPU in my laptop (Lenovo P16) is an RTX A4500. I just loaded a very detailed figure with about 5 million triangles. Prepare pane is a little choppy (4K resolution dual monitors) with this model. Preview was quite freewheeling.
GPU load when minimized appears to be higher than if static full-screen … but two instances with 12 million triangles … you’re asking a lot of your system.
So are you trying to slice two items with two diffrent instances running at the same time?
No, I don’t slice at the same time of course.
But as long as we won’t be able to have multiple projects open under the same instance, I’ll open several.
12 million triangles is not that much, our computers should be able to handle much much more without been sluggish.
Friend of mine gave me a 780Ti this afternoon, I’ve installed it, it’s a bit better, 40% GPU usage instead of 100%.
Slicing itself is cpu intensive, showing it to you is gfx intensive.
I use a lenovo yoga of all things to slice- Decent cpu- and on board graphics- Nothing is instant, but I’ve never waited more than a minute or so-
Yes, Bambu Studio benefits from a dedicated GPU. Your integrated graphics are likely causing the slow slicing and display. Upgrading to a dedicated card like a GTX 1650 can significantly improve performance. Based on video card benchmarks, even entry-level GPUs outperform integrated graphics in 3D tasks like slicing and rendering.
Really been struggling at work with the slicer taking a long time, I fill the build plate with many small objects and sometimes it can take 20 minutes to slice.
At work I use a Mac mini M2 with 8GB ram and I know I need to upgrade but don’t want to spend out on another Mac to find the GPU is the main issue.
At home I have a Mac mini m4 with 16GB ram and moving around the slicer, rotating the camera etc is a fair bit better the slicing speed is still taking almost 10 minutes.
I just installed Bambu on my home PC which is a new build, AMD 9800X3D, 32GB RAM and a 5700ti and it’s insanely faster, slices in about 15 seconds and I can double the plate up on screen and move everything around with no issues.
The problem is I print from work and home and all my print files are shared in the cloud on the Mac, so for this reason I want to stick with a Mac but would say a Mac mini with 32GB ram make enough difference as GPU is a weak aspect for Mac’s.
Ok so after some more testing I made sure to match all my slicer settings on the PC to make sure it wasn’t an issue with my settings and turns out it partially is, matching everything and the PC slicing went really slow too, with the performance window open it also showed nothing was being heavily utilised.
Testing my settings one part at a time I think I have found the issue, somehow Avoid Crossing Walls has been enabled in my settings which I never used to have on, this one setting will add 10 or more minutes to the slice time, happens on both Mac and PC.
Some more testing, once the file is sliced moving around the objects zooming in etc on the Mac (16GB) super slow, on the PC it’s super fast but looking at hardware utilisation the CPU did not use much but the GPU would go to 100% at times.
So I think others have said correctly, the CPU does most the slicing but the GPU makes a big difference displaying everything, moving around the object etc.
I guess the ideal machine is a PC with a decent graphics card and 32GB+ of RAM, but if using a Mac then 32GB+ is worth it for projects like I’m doing.
This project did make RAM utilisation on the PC go over 20GB, so on my existing Mac’s they are bottlenecked on RAM also.