ICEpdf
  1. ICEpdf
  2. PDF-809

Refactoring of ImagePool and inline image caches

          Details

          • Type: Improvement Improvement
          • Status: Closed
          • Priority: Major Major
          • Resolution: Fixed
          • Affects Version/s: 5.1
          • Fix Version/s: 5.1.1
          • Component/s: Core/Parsing
          • Labels:
            None
          • Environment:
            -d64
            -XX:+UseConcMarkSweepGC
            -XX:+CMSClassUnloadingEnabled
            -XX:+HeapDumpOnOutOfMemoryError
            -XX:MaxPermSize=256m
            -Xms256m
            -Xmx1024m
            -XX:PermSize=256M
            -Dfile.encoding=UTF-8

            Description

             5.1 had improvements in the rendering time of images but it in the provided sample a client is still running into memory issue during a page capture.

              Activity

              Ascending order - Click to sort in descending order
              Patrick Corless created issue -
              Patrick Corless made changes -
              Field Original Value New Value
              Fix Version/s 5.1.1 [ 11770 ]
              Hide
              Permalink
              Patrick Corless added a comment -

              Top image is heap for an 18 page run, button image is the same run but after the cache changes.

              Show
              Patrick Corless added a comment - Top image is heap for an 18 page run, button image is the same run but after the cache changes.
              Patrick Corless made changes -
              Attachment cache_changes.png [ 17391 ]
              Hide
              Permalink
              Patrick Corless added a comment -

              After doing some profiling it was clear that the ImagePool and inline image cache were holding on to too many image objects. And as a result the garbage collection times went up significantly as the capture proceeded. I refactored the ImagePool and inline image cache to use a WeakHashMap. The results where quite significant, render time dropped from 16 to 4 minutes. And the maxed used heap dropped from 1038MB to 559MB. The attached graphs are worth 1000 works as they show the before and after heap profile of the capture.

              Show
              Patrick Corless added a comment - After doing some profiling it was clear that the ImagePool and inline image cache were holding on to too many image objects. And as a result the garbage collection times went up significantly as the capture proceeded. I refactored the ImagePool and inline image cache to use a WeakHashMap. The results where quite significant, render time dropped from 16 to 4 minutes. And the maxed used heap dropped from 1038MB to 559MB. The attached graphs are worth 1000 works as they show the before and after heap profile of the capture.
              Hide
              Permalink
              Patrick Corless added a comment -

              I have a little more testing to do but so far really happy with the results. Marking as fixed.

              Show
              Patrick Corless added a comment - I have a little more testing to do but so far really happy with the results. Marking as fixed.
              Patrick Corless made changes -
              Status Open [ 1 ] Resolved [ 5 ]
              Resolution Fixed [ 1 ]
              Patrick Corless made changes -
              Status Resolved [ 5 ] Closed [ 6 ]

                People

                • Assignee:
                  Patrick Corless
                  Reporter:
                  Patrick Corless
                • Votes:
                  0 Vote for this issue
                  Watchers:
                  1 Start watching this issue

                  Dates

                  • Created:
                    Updated:
                    Resolved: