diff --git a/doc/sphinx/00_intro/SSNI-baseline.xlsx b/doc/sphinx/00_intro/SSNI-baseline.xlsx new file mode 100644 index 00000000..d18e8773 Binary files /dev/null and b/doc/sphinx/00_intro/SSNI-baseline.xlsx differ diff --git a/doc/sphinx/00_intro/introduction.rst b/doc/sphinx/00_intro/introduction.rst index 794eb62a..5db50e60 100644 --- a/doc/sphinx/00_intro/introduction.rst +++ b/doc/sphinx/00_intro/introduction.rst @@ -294,9 +294,9 @@ Note: % of device memory is approximate please note actual memory footprint use SSNI Baseline =================================== -The SSNI Baseline spreadsheet linked below provides draft FOMs and example calculations of FOMs for two hypothetical systems with different primary node types and a third hypothetical system with a secondary node type using a subset of the SSNI benchmarks. +The SSNI Baseline spreadsheet linked below provides FOMs and example calculations of FOMs for two hypothetical systems with different primary node types and a third hypothetical system with a secondary node type using a subset of the SSNI benchmarks. -:download:`SSNI-baseline-draft-v2.xlsx ` +:download:`SSNI-baseline.xlsx ` System Information diff --git a/doc/sphinx/05_mlmd/mlmd.rst b/doc/sphinx/05_mlmd/mlmd.rst index 68548b22..e4be0ac4 100644 --- a/doc/sphinx/05_mlmd/mlmd.rst +++ b/doc/sphinx/05_mlmd/mlmd.rst @@ -261,10 +261,10 @@ Results from MLMD are provided on the following systems: Training HIPNN Model -------------------- -For the training task, only a single FOM needs to be reported, the average epoch time found in the ``model_results.txt`` file. +For the training task, only a single FOM needs to be reported, the 1 / average epoch time found in the ``model_results.txt`` file. -* On Chicoma using a single GPU - 1 / FOM Average Epoch time: 1/0.24505662 = 4.05709 -* On Crossroads using a single node - 1 / FOM Average Epoch time: 1/1.67033911= .5986808 +* On Chicoma using a single GPU - 1 / Average Epoch time: 1/0.24505662 = 4.05709 +* On Crossroads using a single node - 1 / Average Epoch time: 1/1.67033911= .5986808 Simulation+Inference -------------------- diff --git a/doc/sphinx/06_umt/umt.rst b/doc/sphinx/06_umt/umt.rst index 5c0d3dba..79655d01 100644 --- a/doc/sphinx/06_umt/umt.rst +++ b/doc/sphinx/06_umt/umt.rst @@ -23,7 +23,7 @@ The benchmark problem is a single node sweep performance problem (SPP) on a 3D u - SPP 1, a configuration with a high number of unknowns per spatial element with 72 directions and 128 energy bins to solve per mesh cell. - SPP 2, a configuration with a low number of unknowns per spatial element with 32 directions and 16 energy bins to solve per mesh - cell. SPP 2 is still a work in progress. + cell. Figure of Merit