Messages in this thread | | | From | "Woodhouse, David" <> | Subject | Re: [RFC] arm: use built-in byte swap function | Date | Wed, 13 Mar 2013 13:35:05 +0000 |
| |
On Tue, 2013-02-19 at 20:31 -0600, Kim Phillips wrote: > > > > imx_v6_v7_defconfig: 7672373 7667089 -5284 > > > lart_defconfig: 2941150 2941054 -96 > > > mxs_defconfig: 11091983 11095679 3696 > > > > The savings are good, with some impressive cases. However the > > mxs_defconfig is completely the opposite and by far. Any idea? > > Unfortunately, I don't seem to be able to reproduce this anymore. > Same linux-next, with three different compilers always produces > smaller binaries: > > text data bss dec hex filename > 5239363 280576 5569648 11089587 a936b3 linux-next-mxs-orig-gcc4.7/vmlinux > 5239169 280556 5569648 11089373 a935dd linux-next-mxs-bswap-gcc4.7/vmlinux > > 5262223 280592 5569648 11112463 a9900f linux-next-mxs-orig-gcc4.6.3/vmlinux > 5261909 280584 5569648 11112141 a98ecd linux-next-mxs-bswap-gcc4.6.3/vmlinux > > 5241379 280580 5569648 11091607 a93e97 linux-next-mxs-orig-gcc4.6ubuntu/vmlinux > 5241189 280600 5569648 11091437 a93ded linux-next-mxs-bswap-gcc4.6ubuntu/vmlinux > > So I've since made a more consistent cross-build environment, using > cross tools from Linaro [1,2] instead of via Ubuntu [3].
Andrew Pinski has done some work on GCC to support further optimisations: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55177
If you feel like building with the branch at http://gcc.gnu.org/git/?p=gcc.git;a=shortlog;h=refs/heads/pinskia/bytewiseunop and seeing how that affects the results, that could be interesting.
-- Sent with MeeGo's ActiveSync support.
David Woodhouse Open Source Technology Centre David.Woodhouse@intel.com Intel Corporation [unhandled content-type:application/x-pkcs7-signature] | |