Anyhit.cu in OptiX_Apps

[m001]

there seems to be a typo in the comment of this line:
https://github.com/NVIDIA/OptiX_Apps/blob/master/apps/intro_denoiser/shaders/anyhit.cu#L91
that anyhit program is not for the radiance ray type 0; its for the shadow ray type 1

cause here its set as shadow ray type 1:
https://github.com/NVIDIA/OptiX_Apps/blob/c7ef89010a169f333ea4779661e60debe99c0c14/apps/intro_denoiser/src/Application.cpp#L2214

And I encountered an access violation in my app (during calling optixModuleCreateFromPTX) when using optixSetPayload_1() in an anyhit program since driver 442.19; no exception on driver 436.48 (I upgraded directly from 436.48 to 442.19 so I don’t know anything about the behaviour of drivers inbetween).
I changed the handling so that the shadow ray now also uses the “PRD” struct (as implemented in the OptiX_Apps code). This works now. In the failing version I only updated the payload (see screnshot in attachment)

My system: OptiX 7.0.0 SDK CUDA 10.1.243 GTX 1050 2GB Win10PRO 64bit (version 1809; build 17763.107) device driver: 442.19 VS2019 v16.4.2 (toolkit v140 of VS2015)
NOTE: My app is based on the OptiX Introduction Samples and I ported it to OptiX7 before the OptiX_Apps code was online.

[droettger]

Thanks for spotting the copy-paste error in the comment. I just removed it. The function name says it all.

Code after an optixIgnoreIntersection() or optixTerminateRay() will not have any effect.
“These functions do not return to the caller and they immediately terminate the program. Any modifications to ray payload values must be set before calling these functions.”
https://raytracing-docs.nvidia.com/optix7/guide/index.html#device_side_functions#terminating-or-ignoring-traversal

That code shouldn’t crash during module creation though. That would require a minimal, complete reproducer to analyze.

The static_cast<uint32_t> on the seed is not necessary. The optixSetPayload_*() argument is unsigned int.

[m001]

Thank you for your answer, I sent a reproducer by private message.