Cyrus Harrison
Cyrus Harrison
To exactly trigger chunking on an 800K byte dataset, I find I need to do something like: ``` hdf5_otps["chunking/threshold"] = 800000-1; hdf5_otps["chunking/chunk_size"] = 800000; ``` where: ``` hdf5_otps["chunking/threshold"] = 800000;...
``` # z0 z1 z2 # +--------+--------+--------+ # | a0 | a1 ___/| | # |___-----|---- | b2 | # | b0 | b1 | | # +--------+--------+--------+ # matsets:...
update comments here: https://llnl-conduit.readthedocs.io/en/latest/relay_io.html#hdf5-i-o-options
Build pypi binary wheels with zlib + hdf5 support so folks can pip install using pre-build modules. Leverage: cibuildwheel infrastructure (https://cibuildwheel.readthedocs.io/en/stable/) in github actions. Here is the way h5py uses...
Even though MPI process are forking and talking amongst themselves, they still trigger firewall popups during development. Here is an example of what these look like: ![Screen Shot 2023-03-21 at...
Add support to read `HOST_CONFIG` to be used with `cmake -C ${HOST_CONFIG}` This would allow us to pass all options w/o having to plumb each one.
Add caliper macros for source code annotations (inspired by exo) that will be noops when caliper is disabled. Add api to setup caliper when host code or env vars are...
Compiler info: gcc 8.3.1 cuda 10.1 or cuda 11.2.0 How we configure vtk-m: https://github.com/Alpine-DAV/ascent/blob/a00c70b30811e72500aef5c1b1bf611d262ebeac/scripts/build_ascent/build_ascent.sh#L340 (we are not using kokkos cuda support, we are using original vtk-m cuda support) A big...
Conduit now has extensive Blueprint to Silo support, we should wire up a silo extract.