Daniel Arndt

icon indicating a website link https://www.ornl.gov/staff-profile/daniel-arndt icon indicating an email [email protected]

icon company Oak Ridge National Lab icon location Oak Ridge icon location Large Scale Simulations Computational Scientist. One of the developers of the deal.II finite element library and Kokkos.

Results 839 comments of Daniel Arndt

[WIP] Add example for 2D triangle-point intersection

> LGTM with the understanding of major simplification after APIv2 is finalized. Sure.

4 tests time out and 1 test fails

Looks like we could use `__FreeBSD__`, see https://docs.freebsd.org/en/books/porters-handbook/porting-dads/#porting-versions.

Nightly build failures with intel/19, "missing return statement at end of non-void function"

Ah, but that's different, right? We should just mark `property` as `maybe_unused`.

Guard t_openmp_instance with KOKKOS_ENABLE_DEPRECATED_CODE_3

Retest this please.

Pass FunctorAnalysis to ParallelReduce

Superseded by #5982 and #5976.

#5333: Use scratch space appropriate to small reduction elements

You should probably use a test similar to the one added in https://github.com/kokkos/kokkos/pull/6296 to test the local memory implementation for Cuda and HIP (and small types of course).

#5333: Use scratch space appropriate to small reduction elements

I pushed a fix for SYCL. For some reason, we were not using the same (proper) way of dealing with the `num_teams_done` variable as for `RangePolicy` and `MDRangePolicy`. Essentially, we...

#5333: Use scratch space appropriate to small reduction elements

> Thanks for this! I think there's no need for another PR as long as we can deal with the [CI timeout](https://cloud.cees.ornl.gov/jenkins-ci/blue/organizations/jenkins/Kokkos/detail/Kokkos/14713/pipeline/49): Looks like the `array_reduce` is still not working....

#5333: Use scratch space appropriate to small reduction elements

> Looks like the `array_reduce` is still not working. I'll debug that more and restore that guard for that test. I pushed another fix that somewhat hides the real issue...

#5333: Use scratch space appropriate to small reduction elements

> @masterleinad The only `FIXME` left is for `OpenACC` (`size_t` and `double`). I look forward to getting this merged and dealing with `OpenACC` / adding more tests in a follow-up....