Details
-
Task
-
Resolution: Fixed
-
Neutral
-
None
-
None
-
-
Empty show more show less
-
Empty show more show less
-
Yes
-
Yes
Description
Research in DEV-1381 showed that:
- DL4J's memory consumption can be kept under control with JVM arguments
- the limit of 10 users allowed to have a memory NN is not flexible
- too big for small instances, too small for big instances
- also, 10 in-memory networks lead to a lot of processing on top of them
Although releasing users immediately from Magnolia's internal cache doesn't free memory immediately, it is still an improvement over the current solution. We should implement that while at the same time find out what to do with JVM flags. (see linked GitHub issue)
Checklists
Acceptance criteria
Attachments
Issue Links
- relates to
-
MLEARN-18 Adopt deeplearning4j latest beta version
-
- Closed
-
- links to