14.3 Errors & Troubleshooting
When running on supported hardware then the most likely cause of errors can be configuration of these parameters. The settings will have to be adjusted for the type of analysis that you are running and the system you have installed the VE Accelerad instance on.
The best approach we have found would be to take a standard image type (such as illuminance) and through trial and error work with the parameters described to best match the output achieved from a regular RadianceIES calculation. When happy with the configuration note these parameters then proceed to the next type of analysis you are interested in.
IES are not able to guide or support on this part of the setup or to comment further on the appropriate hardware to use or parameters to apply but some online forums are available to discuss issues and setup:
Internally IES spent time configuring VE 2019 installation on a laptop running with an NVIDIA(R) GeForce(R) GTX 1050Ti. Common errors that you may encounter:
1. Artefacts on external images, will be seen in produced images and can be improved by reducing –aa parameter down to zero. This can increase simulation times
2. Artefacts on internal images, will be seen in generated images and can be improved by reducing parameter –aa down to 0. This can increase simulation times
3. Corrupt image error, will find a CUDA error in the Rdaiance.log file (e.g. CUDA Error 4: unspecified launch failure) can occur during image generation. If observed it is recommended to reduce simulation quality from max to a lower and run again.
4. Corrupt image error, will find a CUDA error in the Radiance.log file (e.g. Unknown error (Details: Function "_rtContextLaunch2D" caught exception: Encountered a CUDA error: Kernel launch returned (719): Launch failed, [6619200])) can occur during image generation. If observed it is recommended to reduce simulation quality from max to a lower and run again.
5. Stack overflow on Dynamic simulations, can be observed as cell lux levels showing as zero or extremely low values when compared to adjacent cells. Dynamic simulation log file shows accelerad_rcontrib: warning - Stack overflow occurred 93 times in sensor. May be resolved by running less cells or a smaller room.
6. If you are not running supported hardware then an error can be found in the Radiance.log alerting to this. Accelerad_rpict: 0 rays, 0.00% after 0.0000 hours
Accelerad_rpict: internal - A supported NVIDIA GPU could not be found
(D:\nljones\Radiance\src\rt\optix_radiance.c:181)
accelerad_rpict: 0 rays, 0.00% after 0.0000 hours
