Skip to main content

Table 4 Average performance of 1000 queries against 1 million targets

From: The chemfp project

#bitsMethod#Tanimotos (M)chemfp 1.5chemfp 3.3
Avg. time (ms)TTanimoto (ns)Bandwidth (GiB/s)Avg. time (ms)TTanimoto (ns)Bandwidth (GiB/s)
166k = 191.80.252.688.340.192.0810.7
166k = 10005882.203.745.971.853.157.10
166T = 0.706881.722.508.931.422.0710.8
881k = 11461.5010.310.21.228.3512.5
881k = 10004855.6411.68.974.739.7510.7
881T = 0.705545.7010.310.24.708.4712.3
1021k = 11131.3011.510.40.867.5615.8
1021k = 10007439.2512.59.586.258.4114.2
1021T = 0.704895.5111.310.63.647.4516.0
2048k = 13567.7621.811.05.2914.816.1
2048k = 100093921.222.610.614.615.515.4
2048T = 0.4092019.921.611.113.614.816.1
  1. The timings use three different search methods to search the four different fingerprint types from the chemfp benchmark data set. The total number of Tanimoto evaluations is less than 1 billion because of BitBound pruning. TTanimoto is the average time per Tanimoto evaluation, including storing the hits. The effective read bandwidth is calculated as #Tanimotos * storage_size (24, 112, 128, and 256 bytes respectively)/TTanimoto. Note that while shorter fingerprints are faster and more compact, longer fingerprints tend to have better scientific usefulness