Tomasetti Romin
Tomasetti Romin
There are a lot of failing tests and compilation failures. Are they related to the changes induced by this PR ?
We agreed with @dalg24 that I would re-compile the Trilinos SHA we're using and run the tests. This would help us detect any potential issue in downstream codes.
@dalg24 @fnrizzi @masterleinad Would you have a minute to look at this one? Thanks!
There is a recurring pattern here: ```c++ space.fence(" : fence before"); auto result = ; space.fence(" : fence after"); return result; ``` Would a proposal as follows make sense (at...
Hi @masterleinad ! I think this PR is now more focused on the topic "add `exec` argument to rehash, erase, etc". I think I used the same style as you...
I wasn't even aware of `UnorderedMapInvalidIndex`... Indeed, it's very strange to define it "twice" (`Kokkos::UnorderedMapInvalidIndex` and `Kokkos::UnorderedMap::invalid_index`). I always wondered why `UnorderedMapInsertResult` and `UnorderedMapInsertOpTypes` are defined outside of `UnorderedMap`. What...
> > What about moving these two structs into `UnorderedMap` and removing `UnorderedMapInvalidIndex`? Users would then use: > > > > * `Kokkos::Unorderedmap::invalid_index` > > * `Kokkos::UnorderedMap::InsertResult` > > *...
I guess `SYCL` can be tackled by someone else in another PR if there is any interest, right ?
I have explicitly kept the commit with the "complicated test that relies too much on internals", and added a revert commit. @masterleinad Do you think that it's OK / brings...
I just dropped them. It's easy to re-write it if need be.