Empty Cell Filtering

Occasionally it is important or desirable to remove empty cells from output, particularly when layer filtering is employed. Layer filtering can produce large numbers of empty cells. A large number of empty cells will increase file size and may produce inefficiency in downstream processing operations. Thus, provision for removing empty cells is available from the Empty Cell Filter check box group.

Empty cell filtering is recursive, in that it eliminates empty cells, and cells that contain only instances of empty cells. There are two empty cell filtering operations available.

1. The pre-filter uses in-memory per-layer/per-cell statistics gathered during Cell Hierarchy Digest (CHD) creation to identify cells that should be excluded due to layer filtering. This has relatively low overhead. The CHD in use must have been created with per-cell and per-layer counts specified, or this filtering is skipped. If a CHD is implicitly created in processing, i.e., the user is not using a named CHD from the Cell Hierarchy Digests panel, then these counts will be saved automatically.

   This filtering operation is performed entirely in memory and is typically very fast. However, it identifies only cells that are made empty due to layer filtering.

2. The post-filter identifies empty cells by reading the source layout file. This can be rather time consuming, but applies whether or not layer filtering is being used, and will identify all empty cells.

The two check boxes separately enable each of these empty cell filtering operations. If one doesn't care about empty cells, neither box should be checked. If one is using layer filtering and just wants a quick pass to remove cells made empty due to layer filtering, pre-filter should be checked. If one wants to remove all empty cells, both pre-filter and post-filter should be checked. This will generally provide the fastest operation. If not using layer filtering, this will be equivalent to checking post-filter only. When using layer filtering, enabling both filters can be much faster that using post-filtering only.