Date: Tue, 11 Nov 2003 04:25:42 -0000 From: Alexander Leidinger <Alexander@Leidinger.net> To: des@des.no (Dag-Erling =?ISO-8859-1?Q?Sm=F8rgrav?=) Cc: cvs-src@freebsd.org Subject: Re: cvs commit: src/sys/i386/i386 pmap.c Message-ID: <20031111041443.31952.qmail@exxodus.fedaykin.here> In-Reply-To: <xzp3cdgb6am.fsf@dwp.des.no> References: <20031026064145.18F0E2A8D5@canning.wemm.org> <xzp3cdgb6am.fsf@dwp.des.no>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 26 Oct 2003 12:13:37 +0100
des@des.no (Dag-Erling Sm=F8rgrav) wrote:
> Is there any way to teach (or trick) gcc to generate a branch which
> the p4 will predict correctly?
First of all:
---snip---
`-fno-guess-branch-probability'
Do not guess branch probabilities using a randomized model.
Sometimes gcc will opt to use a randomized model to guess branch
probabilities, when none are available from either profiling
feedback (`-fprofile-arcs') or `__builtin_expect'. This means that
different runs of the compiler on the same program may produce
different object code.
In a hard real-time system, people don't want different runs of the
compiler to produce code that has different behavior; minimizing
non-determinism is of paramount import. This switch allows users
to reduce non-determinism, possibly at the expense of inferior
optimization.
The default is `-fguess-branch-probability' at levels `-O', `-O2',
`-O3', `-Os'.
---snip---
Except instrumenting the code with "expect" or using "-fprofile-arcs"
there's only the possibility to modify the code of gcc to match our
existing model...
Bye,
Alexander.
--=20
I believe the technical term is "Oops!"
http://www.Leidinger.net Alexander @ Leidinger.net
GPG fingerprint =3D C518 BC70 E67F 143F BE91 3365 79E2 9C60 B006 3FE7
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031111041443.31952.qmail>