Bug 506354 - Experiment with various levels of caching at evaluator nodes
Summary: Experiment with various levels of caching at evaluator nodes
Status: CLOSED FIXED
Alias: None
Product: Viatra
Classification: Modeling
Component: Query (show other bugs)
Version: unspecified   Edit
Hardware: PC All
: P3 normal
Target Milestone: 1.5.2   Edit
Assignee: Gabor Bergmann CLA
QA Contact:
URL:
Whiteboard:
Keywords: performance
Depends on:
Blocks: 398765
  Show dependency tree
 
Reported: 2016-10-21 11:08 EDT by Gabor Bergmann CLA
Modified: 2017-03-08 05:17 EST (History)
1 user (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Gabor Bergmann CLA 2016-10-21 11:08:45 EDT
In many cases, it would probably make more sense to experiment with eliminating result caching in with check / eval Rete nodes, as opposed to implementing even more caching.

There could be a separate indicator (annotation or hint) to apply on exceptional evaluations that are very costly (e.g. difficult string operations?) and should be cached.
Comment 1 Eclipse Genie CLA 2016-10-21 12:41:29 EDT
New Gerrit change created: https://git.eclipse.org/r/83711
Comment 3 Gabor Bergmann CLA 2016-12-01 03:01:52 EST
Evaluator nodes are now non-caching by default. This might save some memory. A Rete hint has been introduced to override this default at planning time.

Future work: expression-specific intelligence?

Future work: query planning for better caching (in case caching is needed).
Comment 4 Istvan Rath CLA 2017-02-17 06:57:05 EST
As this has caused performance regressions, please change the default behavior back as it was before.
Comment 5 Eclipse Genie CLA 2017-02-27 10:53:20 EST
New Gerrit change created: https://git.eclipse.org/r/91919
Comment 7 Eclipse Genie CLA 2017-03-06 03:14:20 EST
New Gerrit change created: https://git.eclipse.org/r/92340
Comment 8 Eclipse Genie CLA 2017-03-06 05:43:05 EST
New Gerrit change created: https://git.eclipse.org/r/92351
Comment 10 Zoltan Ujhelyi CLA 2017-03-08 05:11:28 EST
Is anything remaining for this ticket or can it be closed?