This query is greatly improved from main to upsert_handle, although, only slightly for the input usr CPU (which is similar), the bulk of the improvement is in the actual DBSP Usr CPU (where the step()
s are called).
database-stream-processor git:(main) cargo bench --bench nexmark --features with-nexmark -- --first-event-rate=10000000 --max-events=10000000 --cpu-cores 1 --query q4
Compiling dbsp v0.1.0 (/home/michael/dev/vmware/database-stream-processor)
Finished bench [optimized + debuginfo] target(s) in 27.61s
Running benches/nexmark/main.rs (target/release/deps/nexmark-c782d17920c2b5ef)