Large meshes for testing

[koldunovn]

For testing of daskification, I suggest to use a set of large meshes of 2, 4, and 8 M nodes. I have those simulations on ollie. They are not the maximum size we will have to handle, but big enough not to comfortably fit into the memory.

MERC10 (8M, 10 km on equator)

• /work/ollie/nkolduno/output_orca12
• /work/ollie/nkolduno/meshes/FORCA12

MERC15 (4M, 15 km on equator)

• /work/ollie/nkolduno/output_orca15
• /work/ollie/nkolduno/meshes/FORCA15

MERC20 (2M, 20 km on equator)

• /work/ollie/nkolduno/output_orca20
• /work/ollie/nkolduno/meshes/FORCA20

D3 (3M)

output_D3_70
This most probably will be our future workhorse for middle resolution runs.

• /work/ollie/nkolduno/D3_70/
• /work/ollie/nkolduno/meshes/D3_70/

Those are generated with the latest version of FESOM. At some later stage it make sense to put parts of the data to DKRZ swift and ISILON to test remote access.

@pgierz and @suvarchal - any thoughts?