Are you utilizing Distributed Computing in OpendTect? Our Distributed Computing feature allows you to run batch processes on remote computers within your network, whether they a...
Are you utilizing Distributed Computing in OpendTect?
Our Distributed Computing feature allows you to run batch processes on remote computers within your network, whether they are running on Windows, Linux, or macOS, workstations or servers. One especially useful feature is that it saves intermediate results that can be restarted (fault tolerance), and is thus recommended for use for any processing that runs for more than one hour.
Distributed Computing can vastly accelerate your processing time and will facilitate optimal performance in operational settings. It may also be used without any configuration, on the local host only, to benefit from the fault tolerance capabilities. It reduces the project duration by preparing filtered data sets more quickly.
Here’s a quick overview of our Distributed Computing steps:
1. Setup Distributed Computing (see our online help: https://lnkd.in/e3X8yu5D and this video: https://lnkd.in/eeMdKG5k)
2. The interpreter then launches the distributed processing of an attribute or volume builder workflow from the OpendTect main window.
3. The job is outlined in a main parameter file, split into N similar files, representing N jobs required to complete the task.
4. One job per node at a time is allocated, sending the next job as the previous one finishes. Failed/unstarted jobs can be restarted.
5. The distributed processing controller, od_SeisMMBatch, operates independently from the OpendTect main window, allowing that to be closed.
6. The output of all jobs is merged into one final output.
Special thanks to Sadaqat Ali from aramco for inspiring this week’s post!
If there’s a specific topic you’d like me to blog about, let me know
dGB Earth Sciences #OpendTect #seismic #geoscience #distributedcomputing
