At the moment, workflow recipes encode task graph topology patterns, but there is no information for the data footprint. As a result, generated instances/benchmarks have all files have the same size (based on a user-provided total data footprint specification). What would be useful is to have the recipe, for instance, encode what proportion of the data footprint is used between each level of a microstructure... There are likely much better/fancier/more detailed ways of encoding that information...
At the moment, workflow recipes encode task graph topology patterns, but there is no information for the data footprint. As a result, generated instances/benchmarks have all files have the same size (based on a user-provided total data footprint specification). What would be useful is to have the recipe, for instance, encode what proportion of the data footprint is used between each level of a microstructure... There are likely much better/fancier/more detailed ways of encoding that information...