Support jumbo for mojo generated code (-42 CPU minutes)
Mojo generates a lot of code that is expensive to compile and
compiling them through jumbo saves roughly 80-90% of that effort,
which is 42 CPU minutes on the reference machine (surprising that
mojo code needs that long to compile).
42 CPU minutes is -7% of the current "fastest" compilation, or
-4.5% of current master use_jumbo_build=true. The wall clock effect
will be similar for a 4c/8t machine but smaller the more cores
the computer have.
This patch includes a change to jumbo.gni to handle targets
with no sources, and streamlines targets with only one file,
because both happen through the mojo template.
Bug: 771153
Change-Id: Ia11a709711600f1b6eedd098bf47c1df2239a7a5
Reviewed-on: https://chromium-review.googlesource.com/697664
Reviewed-by: Bruce Dawson <brucedawson@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Reviewed-by: Ken Rockot <rockot@chromium.org>
Commit-Queue: Daniel Bratell <bratell@opera.com>
Cr-Commit-Position: refs/heads/master@{#507393}
CrOS-Libchrome-Original-Commit: 6190165700c744187b3de8082341705f2f58be5e
1 file changed