Forums: Troubleshooting and Problems
Hi Carlo,
Yes, SNOMED validation is resource consuming due to large size of SNOMED files. However increase of validation time from 1 minute to 12 hours looks suspicious. I should be ~1.5 minutes instead. My guess is that you run out of memory. Try to use 4+ GB. See http://www.opencdisc.org/projects/validator/performance-and-scalability-guide
Sergiy
Hi Sergiy,
I was running at 4 GB, 4 threads.
I increased the setting to 6 GB (of 8 GB system), ensured I restarted OpenCDISC, and let it run for 10 minutes without a result.
I'll try running again later and see what impact the 6 GB setting has on the full run time.
Regards,
Carlo
Sorry, I cannot reproduce your case.
I run some sample study. Validation of TS domain with SNOMED check itself takes 8 seconds. While running all study datasets, SNOMED validation adds 30 seconds.
OpenCDISC Community package does not include SNOMED files due to the dictionary license limitation. How did you get or create SNOMED files used in your validation?
Hi Sergiy,
We programmatically process the SNOMED source to create a file, sct2_Description_Full-en.txt, and place it in ...\components\config\data\SNOMED\2015-01-31. The process is much the way it is identified here:
http://www.opencdisc.org/forum/snomed-install
We'll take a look and see if there may be anything going on on that end of things.
Regards,
Carlo
I've noticed in the past that configuring the validator (i.e., in version 2.0) to use the SNOMED dictionary results in an increased processing time. For a study that might have taken less than a minute to run without it, incorporating SNOMED can cause it to run for 2-3 minutes, not necessarily a bad deal when the number of records to be processed is considered.
In the latest version of the validator, the same study still takes less than a minute to run without SNOMED, but now takes over 12 hours to run with it.
Regards,
Carlo