Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

Graphs containing certain operators do not utilise all cores

Description

An operator which implements computeTileStack() blocks parallelisation of a graph.
Executing the operator separately has no issue and 100% of the CPU is used.
In a certain use case the CPU usage felt down to 18%.

See forum:
http://forum.step.esa.int/t/batch-processing-of-biophysical-retirevals-using-multiple-cpus/4005

Environment

None

GitHub Work

None

Activity

Marcus Engdahl 
5 December 2019 at 15:57
(edited)

I raised this one to blocker - a fix is needed for 8.0

Marcus Engdahl 
18 December 2018 at 15:27

To be discussed further at IRM1

Marco Peters 
28 June 2017 at 05:03

Maybe related to this one

Marcus Engdahl 
16 December 2016 at 09:55

As discussed over email the gpt-related problems with 5.0 are probably related to an another issue.

Marcus Engdahl 
13 December 2016 at 10:09

This performance-issue needs to be fixed asap with a bugfix-release of 5.0.

Fixed

Details

Assignee

Reporter

Sprint

Affects versions

Components

Priority

Created 12 December 2016 at 16:14
Updated 13 February 2023 at 11:20
Resolved 17 March 2020 at 14:24