From 3f252673788161c9fcdf28f19e22c12c7a6d938b Mon Sep 17 00:00:00 2001 From: apcraig Date: Mon, 13 May 2024 20:54:59 -0600 Subject: [PATCH] Update documentation --- doc/source/user_guide/ug_case_settings.rst | 1 + doc/source/user_guide/ug_implementation.rst | 12 ++++++------ 2 files changed, 7 insertions(+), 6 deletions(-) diff --git a/doc/source/user_guide/ug_case_settings.rst b/doc/source/user_guide/ug_case_settings.rst index 6deab8c11..405e64dc1 100644 --- a/doc/source/user_guide/ug_case_settings.rst +++ b/doc/source/user_guide/ug_case_settings.rst @@ -369,6 +369,7 @@ domain_nml "``maskhalo_remap``", "logical", "mask unused halo cells for transport", "``.false.``" "``maskhalo_bound``", "logical", "mask unused halo cells for boundary updates", "``.false.``" "``max_blocks``", "integer", "maximum number of blocks per MPI task for memory allocation", "-1" + "", "``-1``", "find number of blocks per MPI task automatically", "" "``nprocs``", "integer", "number of MPI tasks to use", "-1" "", "``-1``", "find number of MPI tasks automatically", "" "``ns_boundary_type``", "``cyclic``", "periodic boundary conditions in y-direction", "``open``" diff --git a/doc/source/user_guide/ug_implementation.rst b/doc/source/user_guide/ug_implementation.rst index 7d172e91d..246b2f76e 100644 --- a/doc/source/user_guide/ug_implementation.rst +++ b/doc/source/user_guide/ug_implementation.rst @@ -213,10 +213,11 @@ ghost cells, and the same numbering system is applied to each of the four subdomains. The user sets the ``NTASKS`` and ``NTHRDS`` settings in **cice.settings** -and chooses a block size ``block_size_x`` :math:`\times`\ ``block_size_y``, -``max_blocks``, and decomposition information ``distribution_type``, ``processor_shape``, -and ``distribution_type`` in **ice_in**. That information is used to -determine how the blocks are +and chooses a block size, ``block_size_x`` :math:`\times`\ ``block_size_y``, +and decomposition information ``distribution_type``, ``processor_shape``, +and ``distribution_type`` in **ice_in**. ``max_blocks`` is computed +automatically if set to a value of -1, but it can also be set by the user. +This information is used to determine how the blocks are distributed across the processors, and how the processors are distributed across the grid domain. The model is parallelized over blocks for both MPI and OpenMP. Some suggested combinations for these @@ -227,8 +228,7 @@ but the user can overwrite the defaults by manually changing the values in information to the log file, and if the block size or max blocks is inconsistent with the task and thread size, the model will abort. The code will also print a warning if the maximum number of blocks is too large. -Although this is not fatal, it does use extra memory. If ``max_blocks`` is -set to -1, the code will compute a tentative ``max_blocks`` on the fly. +Although this is not fatal, it does use extra memory. A loop at the end of routine *create_blocks* in module **ice_blocks.F90** will print the locations for all of the blocks on