Throughout a semcode run, memory consumption can get very large.
Early on mmapped files (gix mapping pack files) represent most of the consumed memory, but later on anonymous memory grows by a lot:
sec RSS_GB Anon_GB FileMapped_GB
33 36.0 6.6 29.4 ← peak RSS
39 36.0 7.5 28.5
42 11.6 8.9 2.7 ← mapped pages released, heap keeps climbing
78 21.5 20.1 1.4
Investigate where all that memory goes, and how we can reduce it.
Throughout a semcode run, memory consumption can get very large.
Early on mmapped files (gix mapping pack files) represent most of the consumed memory, but later on anonymous memory grows by a lot:
sec RSS_GB Anon_GB FileMapped_GB
33 36.0 6.6 29.4 ← peak RSS
39 36.0 7.5 28.5
42 11.6 8.9 2.7 ← mapped pages released, heap keeps climbing
78 21.5 20.1 1.4
Investigate where all that memory goes, and how we can reduce it.