[Please note that some of the entries are very old.*]

Tuesday 5 June 2012

Large TMs

Big TMs can slow you down considerably. To speed up the process, you can do the following:
  • Check the language codes of the TM, and see to it that they are in the right format.
Click to enlarge
  • Increase the RAM assigned to Java in Edit | Options | Memory. The default is 1 GB. If you set it to 4 GB or more, Mac users (I don't know about Windows and Linux) will also have to change the preferences in the Java app. Go to your Application folder | Utilities, right-click Java Preferences, Show Package Contents, MacOS, Java Preferences, and set 64-bit as first choice. Drawback: Check spelling as you type won't work anymore, since the open office apps are still 32-bit (though there is a 64-bit version for Linux that may solve this for Linux users). Igor will implement a spell check solution that doesn't depend on the various open office suites this summer.
  • Set your TM to "Read Only". Drawback: No new segments will be added. You can solve this problem easily by using a ProjectTM that you merge with the Fat One after finishing your translation.
  • Switch to the "Pretranslation" mode from the default "Automatic" in Memory Setting in Project Info and check the Translation in Review box. When you start the translation, you will se a progress bar that indicates the background pretranslation. There's no need to wait till the pretranslation is ready, just start working. No procrastination allowed!
  • In the Project Info, change the Matching Type field from "Fuzzy and Subsegment" to "Fuzzy". Drawback (if there is one): The statistical function of subsegment matching won't work (but fuzzy and exact phrase matching still works).

It's probably a good idea to play with these solutions to find out which combination of the above works optimally for your big TM.