From owner-freebsd-questions@freebsd.org Wed Nov 11 16:19:23 2015 Return-Path: Delivered-To: freebsd-questions@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id A7AD7A2CEA5 for ; Wed, 11 Nov 2015 16:19:23 +0000 (UTC) (envelope-from wam@hiwaay.net) Received: from fly.hiwaay.net (fly.hiwaay.net [216.180.54.1]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 76BAA147F for ; Wed, 11 Nov 2015 16:19:23 +0000 (UTC) (envelope-from wam@hiwaay.net) Received: from kabini1.local (dynamic-216-186-213-32.knology.net [216.186.213.32] (may be forged)) (authenticated bits=0) by fly.hiwaay.net (8.13.8/8.13.8/fly) with ESMTP id tABGJEvH024163 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO) for ; Wed, 11 Nov 2015 10:19:15 -0600 Subject: Re: amd64-gcc question To: FreeBSD Questions !!!! References: <5640D486.5050305@hiwaay.net> <56420A8E.6080409@hiwaay.net> From: "William A. Mahaffey III" Message-ID: <56436A82.3010702@hiwaay.net> Date: Wed, 11 Nov 2015 10:24:44 -0553.75 User-Agent: Mozilla/5.0 (X11; FreeBSD amd64; rv:38.0) Gecko/20100101 Thunderbird/38.3.0 MIME-Version: 1.0 In-Reply-To: <56420A8E.6080409@hiwaay.net> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 11 Nov 2015 16:19:23 -0000 On 11/10/15 09:22, William A. Mahaffey III wrote: > On 11/09/15 11:19, William A. Mahaffey III wrote: >> >> >> I pkg-installed amd64-gcc over the weekend hoping for Graphite >> (auto-loop parallelization) support, but no go. I looked around over >> the weekend & found that there was no port for that package, only the >> pkg. I just did a 'portsnap fetch upgrade' & there is now a port for >> amd64-gcc, but it includes no files & no pkg-descr file. I determined >> over the weekend that the gcc's from about V4.3 on can indeed be >> built w/ Graphite support, but you need to do it manually. I found a >> post dated 2010 from someone who did it under linux: >> http://openwall.info/wiki/internal/gcc-local-build. I see no >> configure files for any of the gcc ports (I have the entire ports >> tree downloaded & local, & freshly updated as of a few min. ago). >> What is the canonical/BPP (FreeBSD 9.3R) way of recompiling a port >> with different config flags ? >> >> >> I did find ports/pkgs for the 2 main components apparently needed for >> Graphite support (cloog & ppl) & pkg-installed them over the weekend, >> so I am ready to go on that front. >> >> >> I have gotten as far as running 'make showconfig' in the various gcc* >> & amd64-gcc directories to see what info I could get on default >> config options. In all cases they gave options & said to run 'make >> config' to change options. I didn't even see a 'config:' entry in the >> Makefiles (probably included from elsewhere, but I didn't chase it). >> I only want to make the minimum # of config mods necessary (trusting >> that pkg/port maintainers probably know more than I about their >> various pkg's & ports) to add the cloog & ppl support & recompile. >> >> >> I have been using pkg almost exclusively to maintain my (now 3) >> FreeBSD 9.3R boxen, except for recompiling the linux-c6 flash plugin >> for this box whenever it get upgraded, so I have *no* experience with >> getting more nitty-gritty w/ FreeBSD ports than that :-/. TIA & have >> a good one. > > > Well, I plunged ahead by modifying the Makefile for > /usr/ports/lang/gcc49 to include '--with-ppl=/usr/local/lib > --with-cloog=/usr/local/lib' in the 'CONFIGURE_ARGS+=' block & kicked > off a 'make install', with all output collected in a file for perusal > & posting if necessary. I grepped through for cloog & ominously found > no occurrences. I tried ppl & found some, so I decided to just go > ahead & try the compiler on my code. I got the same results as over > the weekend: > > > Beginning background make all > Initiated at 08:58:01 AM MCST on Tuesday, November 10, 2015 > > gcc49 -DNDEBUG -DUNDER_SCORE_SYS -DLOSE_GAMMAL -I../include -I~/inc > -I../Properties -I../TEST -I../pre -march=opteron -mtune=opteron -O3 > -floop-strip-mine -floop-blo > ck -fprefetch-loop-arrays -floop-parallelize-all > -ftree-parallelize-loops=4 -fopt-info -c mpi.c mpprime.c > mpi.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be > used (-fgraphite, -fgraphite-identity, -floop-block, > -floop-interchange, -floop-strip-mine, -fl > oop-parallelize-all, and -ftree-loop-linear) > /* > ^ > mpprime.c:1:0: sorry, unimplemented: Graphite loop optimizations > cannot be used (-fgraphite, -fgraphite-identity, -floop-block, > -floop-interchange, -floop-strip-mine, > -floop-parallelize-all, and -ftree-loop-linear) > /* > ^ > *** [../lib/R4/opteron/libmpi.a] Error code 1 > 1 error > `usual' not remade because of errors. > gcc49 -DNDEBUG -DUNDER_SCORE_SYS -DLOSE_GAMMAL -I../include -I~/inc > -I../Properties -I../TEST -I../pre -DP64_BIT -march=opteron > -mtune=opteron -O3 -floop-strip-mine > -floop-block -fprefetch-loop-arrays -floop-parallelize-all > -ftree-parallelize-loops=4 -fopt-info -c mpi.c mpprime.c > mpi.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be > used (-fgraphite, -fgraphite-identity, -floop-block, > -floop-interchange, -floop-strip-mine, -fl > oop-parallelize-all, and -ftree-loop-linear) > /* > ^ > mpprime.c:1:0: sorry, unimplemented: Graphite loop optimizations > cannot be used (-fgraphite, -fgraphite-identity, -floop-block, > -floop-interchange, -floop-strip-mine, > -floop-parallelize-all, and -ftree-loop-linear) > /* > ^ > *** [../lib/R8/opteron/libmpi.a] Error code 1 > 1 error > `usual' not remade because of errors. > MakeMPI: OPTERON up to date. > > > i.e. no Graphite support. As I said above I have (all of ?) the > required other ports (cloog & ppl & their dependencies) pkg-installed > & ready to go. What does a feller have to do to get this to work :-) ? > *ANY* clues appreciated. TIA & have a good one. Jeeeez, *c'mon guys !!!!*, someone throw me a bone here :-). Is this feasible at all ? Just (more) pilot error ? A mix of the 2, or other issues ? Enquiring minds wanna know :-) .... -- William A. Mahaffey III ---------------------------------------------------------------------- "The M1 Garand is without doubt the finest implement of war ever devised by man." -- Gen. George S. Patton Jr.