Article View: comp.os.linux.hardware
Article #863Re: P4 vs Athlon for Linux latexing, is gcc-2.95 responsible?
From: Robert Myers
Date: Sun, 16 Jun 2002 16:13
Date: Sun, 16 Jun 2002 16:13
32 lines
2954 bytes
2954 bytes
"G. Hugh Song" <ghsong@nospam.com> wrote in message news:3D0C8986.5090104@nospam.com... > > > My three-year old Compaq-alpha based 21264 duals beat P4 all the time. > Oh, my! Had compaq sold its Alpha to AMD, we would certainly have had a > much better chance of using a dual true 64-bit 21364 (EV7) by now. > For scientists like me, the way the CPU industry is turning its wheel is > something to damn about. > > Which linux distributors make their RPMs with gcc-3.1? > > Thanks. > You can get a nice summary of available gcc rpm's by going to rufus.w3.org and searching on gcc (the list you get is manageable). I believe that the "Rawhide" distribution is RedHat's not-quite-ready-for-prime-time distribution. It's pretty easy to get icc from support.linux.com. You have to make small but manageagle changes to makefiles to get P4 optimizations. Obviously change CC=gcc to CC=icc. Change CFLAGS to CFLAGS = -tpp7 -xW -ip -ipo -vec_report3 -O3. The vec_report3 will inform you of all the loops that haven't vectorized, even with Intel's own compiler. Then you go to support.intel.com and become an expert on the P4's baroque architecture. I don't see any conceivable way that a compiler could vectorize naive code without making possibly dangerous assumptions. Good luck.
Technical Details
Message-ID:
<AO2P8.65367$6m5.49001@rwcrnsc51.ops.asp.att.net>Path:
arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!canoe.uoregon.edu!csulb.edu!newshub.sdsu.edu!west.cox.net!cox.net!newsfeed1.earthlink.net!newsfeed2.earthlink.net!newsfeed.earthlink.net!wn4feed!worldnet.att.net!204.127.198.203!attbi_feed3!attbi.com!rwcrnsc51.ops.asp.att.net.POSTED!not-for-mailReferences:
<3D0B57CB.5060406@nospam.com> <cVYO8.93284$pw3.4355@sccrnsc03> <3D0C8986.5090104@nospam.com>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: f9e16,ea3300a83f1f851c X-Google-Attributes: gidf9e16,public X-Google-Thread: 102b32,ea3300a83f1f851c X-Google-Attributes: gid102b32,public X-Google-Thread: f65d0,ea3300a83f1f851c X-Google-Attributes: gidf65d0,public X-Google-Thread: 1094ba,b32aa620bb2987b1 X-Google-Attributes: gid1094ba,public X-Google-Thread: 1045b9,ea3300a83f1f851c X-Google-Attributes: gid1045b9,public X-Google-ArrivalTime: 2002-06-16 09:13:21 PST Path: arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!canoe.uoregon.edu!csulb.edu!newshub.sdsu.edu!west.cox.net!cox.net!newsfeed1.earthlink.net!newsfeed2.earthlink.net!newsfeed.earthlink.net!wn4feed!worldnet.att.net!204.127.198.203!attbi_feed3!attbi.com!rwcrnsc51.ops.asp.att.net.POSTED!not-for-mail From: "Robert Myers" <rmyers1400@attbi.com> Newsgroups: comp.os.linux.hardware,gnu.gcc,comp.sys.ibm.pc.hardware.chips,comp.text.tex,comp.lang.fortran References: <3D0B57CB.5060406@nospam.com> <cVYO8.93284$pw3.4355@sccrnsc03> <3D0C8986.5090104@nospam.com> Subject: Re: P4 vs Athlon for Linux latexing, is gcc-2.95 responsible? Lines: 36 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2600.0000 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 Message-ID: <AO2P8.65367$6m5.49001@rwcrnsc51.ops.asp.att.net> NNTP-Posting-Host: 66.30.233.6 X-Complaints-To: abuse@attbi.com X-Trace: rwcrnsc51.ops.asp.att.net 1024244000 66.30.233.6 (Sun, 16 Jun 2002 16:13:20 GMT) NNTP-Posting-Date: Sun, 16 Jun 2002 16:13:20 GMT Organization: AT&T Broadband Date: Sun, 16 Jun 2002 16:13:20 GMT