MoFtZ
MoFtZ
Duplicate of #1220
hi @laurentchougrani. Can you please provide some sample code the reproduces the issue? Ideally, this would be a minimal project that can be compiled and demonstrates your issue.
hi @El-Gor-do. When your kernel causes a failure (e.g. divide by zero), the kernel will stop executing. The error will be reported on the next Cuda API call (e.g. Synchronize)....
@El-Gor-do ok, did some more investigation, and it looks like Cuda will not throw an exception on divide-by-zero. This is consistent with the [exception handling rules](https://en.wikipedia.org/wiki/IEEE_754#Exception_handling) for the IEEE 754...
I have checked the behavior of divide by zero, and it is giving me infinity. Looking back through your example, you are using integers not floating points. Since IEEE 754...
@El-Gor-do Cuda is a native library. It is important to cleanup all the resources you acquire. ILGPU includes Finalizers that will attempt to cleanup the native resources, however, the order...
The exception is an assertion failure in the Cuda kernel. You should be fixing your kernel to avoid triggering the assertion. In release mode, assertions are normally excluded.
An exception being thrown by a call to dispose is a side-effect of the underlying Cuda implementation. An error code can be returned on a subsequent call to a Cuda...
Sounds like a good idea @pavlovic-ivan 👍 Especially if it makes your life easier.
> There will be some setup required in the settings This will need to be handled by @m4rs-mt 🥰