CIPRES Portal v 1.07 (version history)

How Long Will My Job Run?

We have compiled the run time statistics of CIPRES Portal Jobs for the past six months.

We have found that for GARLI and RAxML, the run time can be estimated by considering the data type, the number of characters, and the number of taxa.
Plots of the run times as a function of these parameters are shown in Figures 1 and 2 below. The results trend toward a linear relationship between log (ntax*nchar) and log(runtime). There are deviations from this relationship if when there is a large disparity between the two terms. For example, ntax=10 nchar=10,000 will run much faster than ntax=100 nchar=1000.

For RAxML, DNA Datatype; the best fit line is log(runtime) = 1.0[log (ntax*nchar] - 4.29 (Figure1, Green line)

For RAxML Protein Datatype; the best fit line is log(runtime) = 1.23[log (ntax*nchar] - 4.18 (Figure1, Red line)

For Garli, DNA Datatype; the best fit line is log(runtime) = 0.8 [log (ntax*nchar] - 2.90 (Figure 2)

 

Figure 1. RAxML Run times

 

Figure 2. GARLI Run times

With PAUP, the relationship between data set and run time did not seem so obvious.
However, the following data sets did not complete in the allowed time using PAUP.

Please send us your feature requests

CIPRES is a Collaboration of the Following Institutions:
American Museum of Natural History, Florida State University, New Jersey Institute of Technology, North Carolina State University, Rice University, San Diego Supercomputer Center, Texas A&M, University of Arizona, University of British Columbia, University of California at Berkeley, University of California at San Diego, University of Connecticut, University of Pennsylvania, Univ. of South Carolina, University of Texas at Austin, and Yale University. xhtml | css