litghost
litghost
The flip side is this increases the RAM usage dramatically. How's your runtime with the "create_edges" step right now? "create_edges" doesn't use a database cache either. Rather than bringing all...
In general, the access patterns for the routing import step should be: - index lookups that SQLite should be caching already - Row scans that should be disk read cache...
Those all look good. I believe the issue is that the SQLite cache is too small and results in evicting the b-trees on use, rather than holding them in memory....
I just checked the 3 indicies that get hit by the queries you listed, and they total about ~120 MB of b-trees. I believe if you try setting the sqlite...
Which graph is this on?
I expected this to resolve the hanging Vivado issue we've been seeing, but Vivado 2020.2 still appears to have the issue. This likely points to something else is going on.
Please rebase on master
This test is missing an SDC file to constraint clocks.
The changes in this PR look okay, but I think a couple things are missing. First, there needs to a be a usage document on how to use this flag...
> @litghost is there any reason this pull request hasn't been merged other than a rebase? I have recently run into the issue with LDCEs being generated instead of LDCE_ZINIs...