Direct3D glitches on rendered mode

[jeff (Jeff LaSor)]

The memory ramping up I believe will be a one-time thing on new installs if/when any kind of shader change exists… What you’re most likely seeing is AMD’s shader cache getting cleared and recompiled in the background…there’s nothing Rhino can do about this…

But… please confirm this… Just let Rhino continue to run, load more files and use Rendered mode… and after a short period of time (system dependent) the memory usage should go back down. If/when the memory ramping subsides, close and restart Rhino…you should no longer see the memory usage ramping and subsequent runs of Rhino should also no longer show this on startup… but you may experience this again on future installs…but again, it should only happen once and never again.

Actually I would like you to try those other tests I posted above if you don’t mind…just for my sanity, and too see if any off my latest theories hold…

Thanks,

-Jeff

[jeff (Jeff LaSor)]

@maje90

I didn’t like being the only one reporting this, so I tried to setup another machine.
I bought some week ago a 2nd hand pc for 350 bucks: i7 8700K + RX 5700XT 8GB.
I tried with a clean fresh install of both Windows 11 and Windows 10, and they both have no glitches on Rendered Viewport.
I did fast installation with Windows default GPU drivers, I’ll make more tests but it’s not worth to have you waste time on this.
I’m really sorry Jeff, it seems this specific problem only occurs on my main pc.
Thank you for the support!

@maje90 Just to confirm…

I really only need know 2 things:

1. When the artifacts appear, does dragging a selection rectangle over a blank area of the scene make them go away, or do they remain?
2. Do you still see the artifacts after you’ve closed Rhino, unplugged your 3DConnexion device and restart Rhino?

Thanks,

-J

[jeff (Jeff LaSor)]

@maje90

or I manage to remove them on my main pc , maybe a clean Windows re-install is due

I don’t think that will make any difference. I truly believe you’ve discovered a “hidden” problem, not with Rhino but with specific situations with Navi 2.x devices and certain combinations that Rhino is doing that reveals it. Your other machine using the RX 5700XT is using Navi 1.0 (RDNA1) not Navi 2.x (RDNA2), which uses completely different compression hardware. RDNA1 and RDNA3 might all seem to work just fine, but RDNA2 is where the problem really lies… Thus, I don’t think a fresh Windows install will solve anything… but if my latest theories hold, then there may be something I can do in Rhino to get around it… since it’s highly doubtful that AMD will fix any of this at this point in time.

-J

[jeff (Jeff LaSor)]

@maje90

I can reply to your last questions, tell me if you want to pursue this road… but maybe it’s not relevant anymore?

@maje90 Any news on this? If you’re not wanting to dig into this more, I’ll close out the report and we can reopen it at a later date.

Thanks,

-J

[maje90 (Riccardo Majewski)]

Sorry for late reply, Jeff.

9.0.26216.12303, 2026-08-04
Direct3D , Antialias 4x, 3dconnexion usb unpugged and both plugins disabled.
Gamma 2.2
Ground plane and Shadows are on.

---

Now TestMyResolve behaves differently (or maybe I see this only now).
While it’s active, if no object is visible on viewport (or is visible but selected) even on wireframe viewports i have glitches, but it “swap” on Rendered viewport.

---

@jeff

So I guess the sequence would now be:

1. Rotate the view to clear the artifacts.
2. Mouse wheel one click to zoom in/out slightly… I’m assuming this now will produce the artifacts.
3. Drag out selection rectangle in a blank area in the scene (i.e. don’t actually selecting anything)

Do the artifacts disappear on step 3?

With TestMyResolve, yes, but only while the selection rectangle is occuring, upon release the glitches re-appear.
Step 1 that “clear the artifacts” works only if shadows and groundplane is enabled…

---

@jeff

turn shadows off, turn skylight off… I just want plain rendered mode with gamma correction on… The more we can eliminate as possible causes, the closer we can get to the actual culprit.

With groundplane , shadows and lights off , glitches disappear momentarily only while the camera is being adjusted with Rotate/Pan or while rectangle selection is ongoing, any case will make glitches re-appear on click release.

---

@jeff

do you see the artifacts in a blank scene in Rendered mode? No geometry, just a plain blank scene… can you get the artifacts to appear?

Yes. Even with everything off.

---

@jeff

The memory ramping up I believe will be a one-time thing on new installs if/when any kind of shader change exists… What you’re most likely seeing is AMD’s shader cache getting cleared and recompiled in the background…there’s nothing Rhino can do about this…

But… please confirm this… Just let Rhino continue to run, load more files and use Rendered mode… and after a short period of time (system dependent) the memory usage should go back down. If/when the memory ramping subsides, close and restart Rhino…you should no longer see the memory usage ramping and subsequent runs of Rhino should also no longer show this on startup… but you may experience this again on future installs…but again, it should only happen once and never again.

yeah, seems it’s like you say. It re-trigger again every new update… but ok.

---

@jeff

1. When the artifacts appear, does dragging a selection rectangle over a blank area of the scene make them go away, or do they remain?
2. Do you still see the artifacts after you’ve closed Rhino, unplugged your 3DConnexion device and restart Rhino?

1 while “drawing” the selection rectangle, they are gone. Reappear after you finished the rectangle and release the click/mouse button.
2 for all this, I have removed the 3d connexion dongle and disabled the plugins

[jeff (Jeff LaSor)]

Ok thanks for the quick response…

After having seen this, I’m actually at a complete loss… unfortunately The behavior you show here is not at all what I was expecting, which negates any theories I had.

I’m not sure what to do at this point, but I’ll keep digging around and see if I can come up with more tests and possible ideas.

Thanks,

-J