Measuring the time it takes to run a script with a certain number of database roundtrips

timings

TIME             moves    placements    swaps    resigns     timeouts    TOTAL    req/s    s/req
================================================================================================
47m 36.012s      18770    18314         225      443         13          37765    13.22    0.075
36m 51.226s       9385    18314         225      443         13          28380    12.83    0.078
26m  6.464s       9385     9157         225      443         13          19223    12.27    0.081
15m 11.009s       4693     4579         225      443         13           9953    10.93    0.091
 9m 42.419s       2347     2290         225      443         13           5318     9.13    0.110
 6m 52.158s       1174     1146         225      443         13           3002     7.28    0.137
 5m 28.929s        588      574         225      443         13           1843     5.60    0.179
 4m 46.100s        295      288         225      443         13           1264     4.42    0.226
 4m 28.729s        295      288         225      222         13           1043     3.88    0.258
 3m 25.534s         74       72           1        1          1            149     0.72    1.379
 3m 21.100s         38       37           1        1          1             78     0.39    2.578

req/s and s/req just look like inverses. But I'd be curious if you've reached the point that the main script code is dominating the runtime -- so that if you averaged just the time of the DB requests, if s/req would trend to a bound.