performance testing of UNWIND kernel option

Gennady Kupava gb at bsdmn.com
Tue Mar 8 15:01:42 CET 2011


Hi, list.

Today I noticed the following change in SHR:
http://git.openembedded.org/cgit.cgi/openembedded/commit/?id=1516588acd3c4b4dd4add71d06ab8ce0d1bafa02 (by Denis 'GNUtoo' Carikli <GNUtoo at no-log.org>) and decided to lmbench it.

Here are results: http://www.bsdmn.com/lmbench/unwind_summary.txt

You can see comparison of:

34def -> kernel with CONFIG_FRAME_POINTER
unwind -> kernel with CONFIG_ARM_UNWIND
default -> for reference, old debugging kernel

The unwind option provide clear benefit of 5%-10% in almost every area.

Nice spot Denis!

Gennady.




More information about the community mailing list