Example using dynamic parallelism

[lehuyduc]

Can you add an example using -rdc=true ? I've added both -rdc=true, -dlink, it compiles but when the program runs, I have __cudaRegisterLinkedBinary_name error.

Thanks!

[cloudhan]

This one is derived from tf_runtime, and it does not genernally supports dlto, and it is hardly to implement it. The cause is quite deeply rooted.

If you need dlto, you can use the new community effort https://github.com/bazel-contrib/rules_cuda.

[liuliu]

Thanks! @cloudhan this reminds me to switch my repos to use your rules_cuda and archive this repo.