I have a model in which I am writing a performance metric (average tokens/batch) to a global table at the end of the model run. Is there a way to reference this value directly when using the Experimenter (without the need to copy the global table value to the PerformanceMeasures table)? This used to be possible in older software versions but it seems to have been removed in v. 2021.
Answer by Jordan Johnson ·
The only thing about performance measures that changed in v2021 is that you define performance measures in a Performance Measure table, instead of in a tab in the Experimenter tool. The Experimenter includes all Performance Measures found in the model automatically.
To set a performance measure to the value of a global table, set its value to code like the following:
You could also use the Reference sampler to sample the cell in the table, and then use code like this:
It may also be possible to skip writing to the global table at all. For Batch activities in General Process Flows, you can read that number directly from the batch. Add a performance measure, sample the Batch as the reference, and use code like the following:
getstat(reference, "TokensPerBatch", STAT_AVERAGE)