[BUG] Increasing server-threads parameter on keydb-server results in poorer keydb-benchmark results

[DreadfulCode]

Describe the bug

keydb-benchmark results in lower numbers when the --server-threads flag is increased in server being benchmarked

To reproduce

Start up a keydb instance with one thread, keydb-server --protected-mode no --server-threads 1 --save “”

Then run benchmark tool and note results: keydb-benchmark -q Actual results

PING_INLINE: 150375.94 requests per second, p50=0.159 msec PING_MBULK: 158730.16 requests per second, p50=0.159 msec

Restart server with 2 threads: keydb-server --protected-mode no --server-threads 2 --save “”

Then run benchmark tool again: keydb-benchmark -q Actual results: PING_INLINE: 116414.43 requests per second, p50=0.207 msec PING_MBULK: 115340.26 requests per second, p50=0.215 msec

Expected behavior Better performance, higher requests per second with more than one thread.

PING_INLINE: 232828.86 requests per second, PING_MBULK: 230680.52 requests per second,

Additional information

on i5-11400/Ubuntu 20