Date: Sat, 14 May 2011 22:43:53 +0200 From: Marius Strobl <marius@alchemy.franken.de> To: Alan Cox <alc@rice.edu> Cc: svn-src-head@freebsd.org, Matthew D Fleming <mdf@freebsd.org>, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r221855 - in head/sys: amd64/include arm/include dev/md dev/null i386/include ia64/include mips/include powerpc/include sparc64/include sun4v/include sys vm Message-ID: <20110514204352.GA33899@alchemy.franken.de> In-Reply-To: <4DCD9766.4010601@rice.edu> References: <201105131935.p4DJZ14Q080511@svn.freebsd.org> <4DCD9766.4010601@rice.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, May 13, 2011 at 03:41:10PM -0500, Alan Cox wrote: > On 5/13/2011 2:35 PM, Matthew D Fleming wrote: > >Author: mdf > >Date: Fri May 13 19:35:01 2011 > >New Revision: 221855 > >URL: http://svn.freebsd.org/changeset/base/221855 > > > >Log: > > Move the ZERO_REGION_SIZE to a machine-dependent file, as on many > > architectures (i386, for example) the virtual memory space may be > > constrained enough that 2MB is a large chunk. Use 64K for arches > > other than amd64 and ia64, with special handling for sparc64 due to > > differing hardware. > > > > Also commit the comment changes to kmem_init_zero_region() that I > > missed due to not saving the file. (Darn the unfamiliar development > > environment). > > > > Arch maintainers, please feel free to adjust ZERO_REGION_SIZE as you > > see fit. > > > > Requested by: alc > > MFC after: 1 week > > MFC with: r221853 > > I would encourage people on platforms with virtually indexed data caches > to do a simple before and after performance test on this change, like > "dd if=/dev/zero of=/dev/null bs=64k", just to make sure that there are > no performance regressions. > Actually, this improved the performance on an old 500MHz USIIe quite a bit, with r221854: v100# dd if=/dev/zero of=/dev/null bs=64k count=100000 100000+0 records in 100000+0 records out 6553600000 bytes transferred in 13.523223 secs (484618204 bytes/sec) with r221855: v100# dd if=/dev/zero of=/dev/null bs=64k count=100000 100000+0 records in 100000+0 records out 6553600000 bytes transferred in 5.235192 secs (1251835639 bytes/sec) Repeated runs vary only marginally in both cases. Marius
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110514204352.GA33899>