commit 351e5832e16feddd33d900ce1d30f5d275b400cc
author scottb@google.com <scottb@google.com@8db76d5a-ed1c-0410-87a9-c151d255dfc7> Wed Oct 28 06:15:10 2009 +0000
committer scottb@google.com <scottb@google.com@8db76d5a-ed1c-0410-87a9-c151d255dfc7> Wed Oct 28 06:15:10 2009 +0000
tree 5722c01468e4817296b59381ca7ebab8d8d440db
parent 1a2e05201a5f1430e247817bafe803f181dea3a2

Implements "-XshardPrecompile" for sharding the precompilation.

When enabled, Precompile merely counts the number of permutations, and all the real work happens during CompilePerms.  This net result is more work is done, but if enough processors can run perms in parallel, a compile can finish faster in wall time.

Traditionally, peak memory use (and therefore OutOfMemoryError) tends occur while Precompiling many permutations; enabling precompile sharding will greatly reduce peak memory requirements.

Patch by: spoon
Review by: me



git-svn-id: https://google-web-toolkit.googlecode.com/svn/trunk@6501 8db76d5a-ed1c-0410-87a9-c151d255dfc7