When a user sorts a table in the UI, e.g. timer aggregation view, rows without a value are handled as MAX_VALUE, so they will appear above the highest value.
When a user is searching for, e.g. methods with a long execution time, the user gets (depending on its view properties) a lot of rows without any values.
I would assume that when I'm ordering a list descending the rows related to the highest values are on the top and rows without a value are below the lowest values.
Example assumption (ordered DESC): 9 - 6 - 5 - 2 - 0 - N/A - N/A
Example current (ordered DESC): N/A - N/A - 9 - 6 - 5 - 2 - 0