scale: 10000 records × 10000 tokens (100.0M tokens), pool capacity 100.0M, target ratio 0.5, gc keep 0.5
write: 10000 records / 100.0M tokens in 9.4s
db after write: 10.7 MiB, records dir 384.8 MiB
== reads ==
facets: median 5.58ms p95 5.74ms
list 20: median 5.81ms p95 5.85ms
catalog: median 11.93ms p95 12.01ms
search title term "topic 7": median 7.16ms p95 7.47ms
search ascii "alpha beta": median 10.58ms p95 10.69ms
search CJK "记忆": median 2.09ms p95 2.14ms
grep body "xxxxxxxx": median 15.18ms p95 15.29ms
get record: median 0.04ms p95 0.04ms
== compression ==
compressCold: moved 5000 records, freed 50.00M tokens in 14.45s
hot 50.00M / cold 50.00M tokens; cold files 192.4 MiB; db 28.3 MiB
search after cold (card fields): median 28.33ms p95 28.94ms
catalog after cold: median 14.51ms p95 14.72ms
== cold gc ==
gcCold: removed 2500 records, freed 25.00M cold tokens in 5.36s
db after gc: 28.3 MiB