Skip to content

GitLab

  • Menu
Projects Groups Snippets
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
    • Contribute to GitLab
  • Sign in
  • fastcodeml fastcodeml
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
  • Issues 12
    • Issues 12
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 0
    • Merge requests 0
  • Deployments
    • Deployments
    • Releases
  • Monitor
    • Monitor
    • Incidents
  • Analytics
    • Analytics
    • Value stream
    • Repository
  • Wiki
    • Wiki
  • Activity
  • Graph
  • Create a new issue
  • Commits
  • Issue Boards
Collapse sidebar
  • phylo
  • fastcodemlfastcodeml
  • Issues
  • #23

Closed
Open
Created Nov 12, 2015 by Iakov Davydov@idavydovGuest

uses all available CPU's even if -nt 1 with OpenBLAS

If fastcodeml is compiled with OpenBLAS it uses all available CPU's irrespective of -nt option.

It's a default behavior of OpenBLAS to use all CPU's. That's especially problematic since

  1. OpenBLAS is the default BLAS implementation in Ubuntu
  2. Most CPUs have hyper-threading which doesn't give a real advantages. Instead huge number of processes is making things significantly slower.

One work-around is to run fastcodeml with OMP_NUM_THREADS=1, but it only works if you need one thread.

I consider this bug to be important, because it prevents most users from using a parallel version of fastcodeml.

Assignee
Assign to
Time tracking