In theory it is possible to load a dump and index files produced on another platform (including those from a 64 bit JVM into a 32-bit Memory Analyzer), however there is no guarantee that the memory will be sufficient.
To do this, copy the dump and all the generated index files from the original analysis into a single directory on your workstation, keeping the original file names for the dump and index files.
Then start Memory Analyzer and attempt to load the dump on your workstation. Memory Analyzer should recognise the presence of the previously produced index files and attempt to load these rather than reparse the dump.
This should work provided you have adequate memory for the task, however this is not guaranteed. Failing this you could raise a requirement against MAT on the Eclipse Bugzilla: https://bugs.eclipse.org/bugs/buglist.cgi?product=MAT
to reduce memory requirements for reloading a parsed heap dump.
At the moment if the index file and heap dump have an identical last modified date and time then the dump will still be reparsed.
I think it would be better if the dates were the same the dump should be reopened, as it is more likely that the dump and indexes have been restored or transferred with the current date rather than the dump has been regenerated just at the moment the index is created. Any comments?