Christian Trott
Christian Trott
Hm I thought we had an exponential back-off at some point lets check this again.
@janciesko can you please check this, it looks in the code like yield is called here: https://github.com/kokkos/kokkos/blob/71a9bcae52543bd065522bf3e41b5bfa467d8015/core/src/Threads/Kokkos_ThreadsExec.cpp#L132C1-L136C2 But maybe its not doing it for every thread in the thread pool??
Agreed I think we start with just not having tests for Kokkos itself enabled in Trilinos and then go from there.
I don't think this is possible in general. I.e. uniqueness is NOT enough to guarantee that there IS an inverse mapping you can express without explicitly generating it, by computing...
In particular if it is not exhaustive.
We have the license stuff resolved !! I.e. we finally got license change to Apache2 with LLVM exceptions approved, including the wording about containing code with other licenses in subdirectory.
We are currently doing the license update right now in core and kernels.
@cwpearson can you review this
closed in favor of #142
Its not quite clear to me why the point_t default constructor is changed, considering that there is a `reduction_identity` overload. Maybe the team reduce does not use reduction identity??