Skip to main content

Table 5 Multiquery search performance

From: The chemfp project

MethodQuery size1 thread2 threads4 threads
Time (s)Time (s)ScalingTime (s)Scaling
k = 110005.313.931.353.691.44
k = 1Sorted5.243.841.363.501.50
k = 1N × N7130 (= 1 h 58 m)5200 (= 1 h 26 m)1.374640 (= 1 h 17 m)1.54
k = 1000100014.610.51.399.541.53
k = 1000Sorted14.58.421.726.302.30
k = 1000N × N15,300 (= 4 h 14 m)8040 (= 2 h 13 m)1.904690 (= 1 h 18 m)3.26
T = 0.9010002.952.191.352.031.45
T = 0.90Sorted2.921.651.771.042.81
T = 0.90N × N1890 (= 31 m 34 s)999 (= 16 m 39 s)1.90550 (= 9 m 9 s)3.45
T = 0.8010005.524.091.353.771.46
T = 0.80Sorted5.472.961.852.032.69
T = 0.80N × N3490 (= 58 m 9 s)1830 (= 30 m 25 s)1.911010 (= 16 m 47 s)3.46
T = 0.7010008.095.951.365.431.49
T = 0.70Sorted8.074.371.852.802.88
T = 0.70N × N4930 (= 1 h 22 m)2580 (= 42 m 57 s)1.911430 (= 23 m 49 s)3.45
T = 0.40100013.69.991.368.281.64
T = 0.40Sorted13.67.391.834.542.99
T = 0.40N × N7120 (= 1 h 58 m)3710 (= 1 h 1 m)1.922100 (= 34 m 55 s)3.40
  1. Time to search the 1 million 2048-bit Morgan fingerprints from the chemfp benchmark data set, for different numbers of threads. A query size of “1000” indicates that the first 1000 benchmark queries were used, “sorted” indicates the same 1000 queries sorted by popcount, and “N × N” generates the full sparse similarity matrix for the 1 million target fingerprints