GPU rendering

[Anonymous]

Are there any plans in the pipeline to introduce GPU rendering rather than the current CPU rendering?

[JB-CNC]

Hi @jstadt, thanks for stepping in.

I’d like to highlight that the core frustration in this thread is not about local vs. cloud rendering options. It’s about why Fusion 360’s rendering engine still doesn’t leverage GPU acceleration, and whether there is a clear plan to change that.

Most users have GPUs that Fusion simply doesn't utilise. This means painfully slow render times and a sense that Fusion is leaving significant performance on the table, especially when compared to competitors or even free alternatives. In 2016, this could potentially have been dismissed as a frustration. In 2025, it’s frankly absurd.

Your response, while courteous, doesn’t address this core issue:

• Is GPU-accelerated local rendering on the roadmap?

• If not, why not? Is it architectural, strategic, or something else?

• When can we expect a more transparent update?

With legitimate competitors gaining ground, this issue has become more than a technical gripe, its now a deciding factor when choosing where to invest our resources.

[TrippyLighting]

---

@JB-CNC wrote:

With legitimate competitors gaining ground, this issue has become more than a technical gripe, its now a deciding factor when choosing where to invest our resources.

---

Can you name a legitimate competitor?

When it comes to rendering with CAD software, Keyshot has been the de-facto industry standard for many years.

[jf_guilard]

I'm adding to this post very quickly (and this will be my one and only message). I'm therefore addressing the people of fusion360, whether they are technicians, designers, analysts, or any other internal function:
Here's a concrete example of the importance of this topic, and at the same time, of the occasional uselessness of Fusion360 in specific cases. I'm currently working on a correction for a model containing a huge number of facets. I've been waiting for about an hour for fusion to give me back control because the CPU (Ryzen 9 - 7950X3D, 8 cores 16 threads and 3D cache linked to 128 GB of RAM and 12 TB of NVME) simply can't get the job done, whereas if Fusion had implemented a simple box giving the choice between working with the GPU or working with the CPU, the job would have been done in 30 seconds.
The result? I'm going to stop the Fusion360 session to get back on track, abandon it, and do the job with another software.
For Autodesk, this is very bad publicity because when a friend asks me which software I used to do my design, I won't mention Fusion360. It's a shame because everything else about Fusion360 is great. But what a strange idea to refuse to work with the GPU in 2025... Sometimes the Fusion processing time is so slow that you'd think you'd come back with an Intel 386.
No, frankly, the fact that the Autodesk team refuses to listen to the requests of the many users of this thread is appalling.
I read everything, and sometimes it feels like we're on the front lines of politicians. Not cool.

[Pedro_Bidarra]

From your description you're either talking about viewport rendering (what you see in the viewports when manipulating the model, panning, zooming, rotating, etc.) and not the renders done via the Render workspace.
To be clear viewport rendering ALREADY uses the GPU (DirectX or OpenGL on Windows).

Or, you appear to be talking about the performance hit of working with complex meshes, any CAD program generally has some performance issues when dealing with meshes when compared to mesh 'first' apps like Blender or 3D Max, but again, if this is the case, this is not even a rendering, viewport or otherwise, issue (not related to CPU vs GPU) and it's not on topic.

[nick.somethingunique]

FWIW, even if GPU could be used, it wouldn’t necessarily be faster, and it
might even be slower than CPU. GPU is great for certain classes of problem
and algorithm, but not all. Every software also has its strengths and
weaknesses, and Fusion is simply not good at mesh oriented editing
operations. If you’re trying to operate at the mesh level, poor
performance, and maybe a bad result even if it did complete, is not
uncommon. When I needed to inflate a mesh for example, I tried in fusion,
it was a fail, so I went to blender, did it easily and successfully there,
and then imported back to fusion. Needing to do mesh operations is also
rare IME if designing from scratch, as you’re working at a higher level of
design abstraction where fusion performs well in general. So it simply
sounds like you’re using the wrong tool for the job. ##- Please type your
reply above this line -##

[TrippyLighting]

---

@nick.somethingunique wrote:
... When I needed to inflate a mesh for example, I tried in fusion,
it was a fail, so I went to blender, did it easily and successfully there,
and then imported back to fusion. 

---

Just to be clear (I've used Blender since the early 2000's), Blender does not use the GPU for modeling tasks either.