From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Sat, 5 Feb 2005 12:00:09 +0200 From: Michael Shigorin To: ALT Devel discussion list Message-ID: <20050205100009.GS21863@osdn.org.ua> Mail-Followup-To: ALT Devel discussion list References: <20050203192917.GA14070@lrn.ru> <20050203222916.GN4945@solemn.turbinal.org> <20050203225355.GO4945@solemn.turbinal.org> <20050204073748.GB19859@lrn.ru> <20050204141120.GP4945@solemn.turbinal.org> <20050204164657.GA23386@basalt.office.altlinux.org> <20050204203415.GA3006@lrn.ru> <20050204213341.GB636@basalt.office.altlinux.org> <20050204214218.GG3006@lrn.ru> <20050204225446.GV4945@solemn.turbinal.org> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="AGBBLMjITsWHeOTZ" Content-Disposition: inline In-Reply-To: <20050204225446.GV4945@solemn.turbinal.org> User-Agent: Mutt/1.4.2.1i Subject: [devel] -mcpu? (was: lib32/lib64) X-BeenThere: devel@altlinux.ru X-Mailman-Version: 2.1.5 Precedence: list Reply-To: ALT Devel discussion list List-Id: ALT Devel discussion list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 05 Feb 2005 10:00:16 -0000 Archived-At: List-Archive: List-Post: --AGBBLMjITsWHeOTZ Content-Type: text/plain; charset=koi8-r Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Feb 05, 2005 at 01:54:46AM +0300, Alexey Tourbin wrote: > =EB=D3=D4=C1=D4=C9, =CB=C1=D6=C5=D4=D3=D1, =C1=D2=C8=C9=D4=C5=CB=D4=D5=D2= =C5 i586 =CB=C1=CB =CF=D3=CE=CF=D7=CE=CF=CA =C4=CC=D1 =D3=C2=CF=D2=CB=C9 = =D5=D6=C5 > =C2=CF=CC=D8=DB=C5 5 =CC=C5=D4, =CF=CE=C1 =C5=DD=A3 =D3=CF =D7=D2=C5=CD= =C5=CE iplabs, =CB=CF=C7=C4=C1 =C1=CC=D8=D4=CC=C9=CE=D5=CB=D3=C1 =CE=C5 > =C2=D9=CC=CF. =FC=DC=DC... =CE=D5 =D2=C1=DA =D0=CF=DB=CC=C1 =D4=C1=CB=C1=D1 =D0=D8=D1=CE= =CB=C1... --- http://lwn.net/Articles/121843/ The Fedora developers keep being asked this, and their answer is the same (although I've yet to check the compile flags on the Extras packages). Fedora packages are compiled optimising for i686 (earlier releases) or the Pentium 4 (today). But they normally use the i386 instruction set. For most packages, there is no performance benefit to using the later instructions. And modern processors other than the Pentium 4 seem to benefit as much from Pentium 4 optimisations (Athlons, in particular, do so much reordering that they can optimise for themselves). If you can identify a package that is compiled with 386 instructions but would benefit from more recent instructions, and if you can provide benchmarks ("Distribution X is faster" doesn't count since there are too many variables), then fedora-devel@redhat.com would love to hear from you. Incidentally, AMD K6-based computers count as i586 as far as gcc is concerned. --- =ED=CF=D6=C5=D4, =C9 =CE=C1=CD =CB=C1=CB-=CE=C9=D4=D8 =CE=C1 =C4=CF=D3=D5= =C7=C5(tm) =D0=CF=C9=C7=D2=C1=D4=D8=D3=D1 =D3 =D4=C5=D3=D4=C1=CD=C9 =D0=CF = =DE=C1=D3=D4=C9 -mcpu? =F0=CF=C4 =D2=D5=CB=CF=CA =C5=D3=D4=D8 Pentium I/II/III{,Xeon}/4, A= thlon, AthlonMP, AthlonXP. --=20 ---- WBR, Michael Shigorin ------ Linux.Kiev http://www.linux.kiev.ua/ --AGBBLMjITsWHeOTZ Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) iD8DBQFCBJkpbsPDprYMm3IRAviYAKDeHsAFhsP22rr1B/1EF/VTsk8OAgCeKmvL POdhoGVYBbva2DYWUS4YtfQ= =tnIx -----END PGP SIGNATURE----- --AGBBLMjITsWHeOTZ--