Article View: gnu.gcc.help
Article #53152Re: Benchmarks of GCC 2.0
From: David Chase
Date: Fri, 28 Feb 1992 07:35
Date: Fri, 28 Feb 1992 07:35
42 lines
2435 bytes
2435 bytes
[This consists entirely of "documented ways to run Sun compilers to get better numbers than plain old -O4." If you plan to use bizarre flag combinations for one compiler in a comparison, you ought to make an effort for the other compiler, unless you already know the answer that you are looking for.] In article <1992Feb25.184349.23760@cc.ic.ac.uk> jensting@ee.ic.ac.uk (Jens Tingleff) writes: >I tried compiling a large FORTRAN numerical package, using Sun Fortran 1.4, Sun >CC 1.1 (unbundled) Sun CC and GCC 2.0, computer SStation IPX. ... >Compiler command lines > >1) Sun bundled /bin/cc -O4 -c .. >2) Sun unbundled /usr/lang/cc -O4 -c .. >3) F77 f77 -O4 -c .. >4) GCC 2 gcc -O2 -fstrength-reduce -funroll-loops -c .. If you're in the mood to mix and match flags and tinker with options, try using some or all of the following with the Sun compilers: -fast -fast -Bstatic -fast -O3 -fast -O3 -Bstatic <-- I recommend this for a first try at speed. -fast -O4 -fast -O4 -Bstatic I don't know if gcc2 uses dynamic or static binding -- depending upon the program, it can make a difference. Most of the official benchmark numbers quoted for Sun make use of static binding. "-fast" is a documented convenience option for people who want their code to run fast. Suffixing -O3 and -O4 both crank up the optimization level further. And yes, the Fortran compiler knows something that f2c+cc do not. Generally speaking, it knows things that f2c and cc will never know, unless the C language is changed. That's one reason people still use Fortran. David Chase Sun
Technical Details
Message-ID:
<kqrpiuINNbup@exodus.Eng.Sun.COM>Path:
arc1-ca.pugleaf.net!sparky!uunet!sun-barr!male.EBay.Sun.COM!jethro.Corp.Sun.COM!exodus.Eng.Sun.COM!peep.Eng.Sun.COM!chasedReferences:
<1992Feb23.213917.13705@murdoch.acc.Virginia.EDU> <1992Feb25.184349.23760@cc.ic.ac.uk>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 11318c,594f013cff35018a X-Google-Attributes: gid11318c,public X-Google-Thread: 1025b4,e63d5b04b0b2c1eb X-Google-Attributes: gid1025b4,public X-Google-ArrivalTime: 1992-02-27 23:38:55 PST Path: arc1-ca.pugleaf.net!sparky!uunet!sun-barr!male.EBay.Sun.COM!jethro.Corp.Sun.COM!exodus.Eng.Sun.COM!peep.Eng.Sun.COM!chased From: chased@peep.Eng.Sun.COM (David Chase) Newsgroups: gnu.gcc.help,gnu.misc.discuss Subject: Re: Benchmarks of GCC 2.0 Summary: Comparative benchmarking information; how to run them. Message-ID: <kqrpiuINNbup@exodus.Eng.Sun.COM> Date: 28 Feb 92 07:35:58 GMT References: <1992Feb23.213917.13705@murdoch.acc.Virginia.EDU> <1992Feb25.184349.23760@cc.ic.ac.uk> Organization: Sun Microsystems, Mt. View, Ca. Lines: 42 NNTP-Posting-Host: peep