polymer
polymer copied to clipboard
ddiakopoulos
engine-scene crash via Intel HD graphics driver
Both samples crash with an access violation here https://github.com/ddiakopoulos/polymer/blob/master/lib-engine/gfx/gl/gl-api.hpp#L572
@lebek hmm can't seem to replicate this one. could you hang a breakpoint and see which of the values is invalid?
On further investigation the crash is happening in the GPU driver (Intel HD Graphics 630). When I force it to use my GTX 1060 instead, it works. The values going into the OpenGL calls seem valid. Let me know how I can help debug. I don't think Intel provides debug symbols.
Hmm, thanks for checking. I tried to debug yesterday on a spare windows machine with Iris 540 but Intel's drivers won't let me install their latest with gl 4.5 support (argh!)
It's also failing here with Intel HD Graphics 530 on latest drivers (exception reading 0x0000000000000028 in ig9icd64.dll) Crashes in the glDrawElements call. The values going to the call seem valid, I've forced Windows to use the Nvidia card on my laptop and the call parameters are exactly the same.
Odd. I finally managed to install the latest Intel Iris 540 drivers on my laptop today and the samples ran. I'll keep trying to find a local machine to replicate this issue on.
I had another look this morning and it fails in the shadow pass (works if I disable the shadow pass). It might be related to updating the uniform blocks in renderer_common.glsl. Oddly when I switch those to packed layout instead of std140, it runs (shadow pass on). I doubt it actually worked since the struct layout isn't setup for packed on the C++ side.. but that's as far as I got.