James Elliott

Let me check and see. I did `nm` on libsci to vet that they were coming from the lapack/blas side. I strongly suspect it is coming from metis not declaring...

Not metis/parmetis. Weird, grepping all the libs I have from my spack build, ``` [jjellio cce-14.0.0]$ for f in $(ls -1 */lib/*.so); do v=$(nm -ls $f 2>/dev/null | grep -n...

It's not really solved. The grief happens if you unwrap Cray PE, and use CrayFTN directly w/optimizations turned on (O2/O3 seemed to do it). The result is, Cray Fortran transformed...

My view - is there is zero wrong with your Spack package or SEACAS. I added a lapack dependency (and use libsci) in spack - but that just let's me...

Are yous setting `--offload-arch=gfx90a --rocm-path=/opt/rocm-5.2.0 -x hip ` ? and the proper link flag: `-x none` These are the relevant bits of the CMake I use w/Trilinos ``` "-DKokkos_ARCH_ZEN3:BOOL=ON" \...

Because I don't want to use hipcc - (I build through Cray as well as AMD)

This blocks being able to use CXX=CC (Cray PE wrappers) regardless of AMD/Cray. I also build directly w/clang (AMD or Cray) - this is essential to setting non standard flags...

@Rombur - How is this intended to work? I think what you are saying is, 1. set amdclang++ as CXX (but don't specify -x hip/none or ROCM_PATH) 2. Cmake should...

Here's the familiar error: ``` -- Architectures: -- ZEN3 -- VEGA90A CMake Error at packages/kokkos/cmake/kokkos_tpls.cmake:29 (MESSAGE): Enabled TPL ROCM inside TriBITS build, but this can only be enabled in a...

It's generally been because Kokkos/Trilinos Cmake isn't 100% compatible. Before you invest too much time, let's see if Ross has some comments (I was hoping his big TRIBITS update was...