Carl Pearson
Carl Pearson
> The implementation looks reasonable to me but I don't fully understand why we need to forward the functor size to tools. Can you provide some examples? The only use...
Would you prefer a separate profiling interface function for each separate piece of information we might want to capture? That's a relatively easy change (and how I envisioned it originally).
@lucbv says this is reported to Intel, in the mean time we can apply beta=0 to the input vector ourselves as a workaround
I think we already have `mode_kk_to_onemkl` so we don't need `trans_mode_kk_to_mkl`.
Also I clearly need to fix the commits in this PR...
``` 9: [ RUN ] sycl_test.gemm_double 9: [ OK ] sycl_test.gemm_double (1414 ms) ```
@ndellingwood I'm vaguely concerned about what's going on in the autotester for this PR (a conflict between an old branch I had in my fork which no longer exists, and...
I think this is the offending line: https://github.com/kokkos/kokkos-kernels/blob/b87dc951f64463fe5d6d4d886c0496ee804b2715/sparse/impl/KokkosSparse_spmv_impl_merge.hpp#L312 If I'm reading `scal` correctly, it just just `y = beta * y` for this case, so if `y` is NaN it...
I think `KokkosKernels_ENABLED_COMPONENTS` is just used in `kokkoskernels_tribits.cmake` to check if all the components requested by Trilinos (?) are inside this variable. It probably makes more sense for this code...
Any insight re the Trilinos integration @ndellingwood?