MantisBT - MEGA
View Issue Details
IDProjectCategoryView StatusDate SubmittedLast Update
0003382MEGA[All Projects] Feedbackpublic2022-01-18 03:162022-02-01 09:44
Reporterguest 
Assigned Togstecher 
PrioritynormalSeverityminorReproducibilityhave not tried
StatusresolvedResolutionwon't fix 
PlatformOSOS Version
Product VersionMEGA-CC 11 (command line version) 
Target VersionFixed in Version 
First NameDiego
Last NameGalagovsky
Emaildgalagovsky@ice.mpg.de
Confirm Emaildgalagovsky@ice.mpg.de
Summary0003382: Use of CPU cluster for tree bootstrap
DescriptionHello,
I am running a Maximum-Likelihood Tree analysis on a 32 CPU cluster. It comes from 310 nucleotide sequences of around 1.2 kb each. I used the prototype tool. I set the Bootstrap value to 500 and the thread value to 32, since we have a cluster of 32 CPUs. It is that it is running very slow, takes almost a week. It is not using the 32 CPUs but only one. The software can use all because for the Model Selection it did use all 32 CPUs. What can be the issue?
Steps To Reproduce
Additional Information
TagsNo tags attached.
Relationships
Attached Files
Issue History
Date ModifiedUsernameFieldChange
2022-01-18 03:16guestNew Issue
2022-02-01 09:44gstecherNote Added: 0004792
2022-02-01 09:44gstecherStatusnew => resolved
2022-02-01 09:44gstecherResolutionopen => won't fix
2022-02-01 09:44gstecherAssigned To => gstecher

Notes
(0004792)
gstecher   
2022-02-01 09:44   
Hi there,

I am writing in response to your recent question regarding the MEGA software. The current version of MEGA has limited capacity to use multiple CPUs for its maximum likelihood tree search. It uses a fine-grained approach for multithreading when doing branch swapping. This is different than the model selection analysis which distributes the analysis of different models to different threads. There are plans to improve the maximum likelihood tree search parallelization and optimize it for many CPUs but it will be a while before that will happen .

Best regards,
Glen Stecher
Institute for Genomics and Evolutionary Medicine
igem.temple.edu