Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Sun, May 14, 2017 at 05:22:27AM -0400, Rasmus Thomsen wrote > Hi, > > unwind tables sounds really interesting, but does it affect the > binary size when I compile with -O2 anyway? Yes, because it's a different part of the build. Personal experience; I do the contributed SSE-only Pale Moon build for older (i.e Pentium 3 class) linux machines. To qualify for official Pale Moon branding, I have to stick with standard specs. For personal use at home on a T400 Lenovo, I can do as I please. My home version uses unwind tables, but the official version does not. The ".bz2" compressed tarball is several megabytes smaller for my home version than for the official SSE-only version. It's even more noticable when extracted/uncompressed. Obviously, this reduction only applies to elf-binaries and libraries, not to sqlite databases, JSON, XML, text files, etc. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Sun, May 14, 2017 at 11:05:14AM +0200, Helmut Jarausch wrote > Many thanks for your hints! > > On 05/14/2017 07:13:05 AM, Walter Dnes wrote: > > Out of sheer curiousity, what > > does gcc report as your native cpu when you run... > > > > gcc -c -Q -march=native --help=target | grep march= > > -march=amdfam10 According to the GCC website... https://gcc.gnu.org/onlinedocs/gcc-6.3.0/gcc/x86-Options.html#x86-Options GCC 6.3.0 "-march=amdfam10" ("-march=native" on your machine) supports MMX, SSE, SSE2, SSE3, SSE4A, 3DNow!, enhanced 3DNow!, and ABM. GCC builds binaries to use these instructions where appropriate. When building for the machine that's doing the build, I always use "-march=native". This gets the full available speedup, without having to figure out what the exact cpu family is. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
Hi, unwind tables sounds really interesting, but does it affect the binary size when I compile with -O2 anyway? Regards, Rasmus Original Message On 14 May 2017, 07:13, Walter Dnes wrote: On Sat, May 13, 2017 at 10:21:19AM +0200, Helmut Jarausch wrote > > I have switched to gcc-6.3.0 at the end of last year. > I have (re-)compiled all my packages including icewm-1.3.12-r1. > Everything works just fine. > > I did have some segfaults (with gimp) until I recompiled each package > with gcc-6.3.0 > > I have > CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" > > in /etc/portage/make.conf Why aren't you running "-march=native"? You're probably missing quite a few features of your cpu. BTW, "-march=native" implies "-mtune=native" and "-msse3" (if the cpu supports it). Out of sheer curiousity, what does gcc report as your native cpu when you run... gcc -c -Q -march=native --help=target | grep march= > and quite a e few > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > (for my aged AMD64 Phenom II machine) > > in /etc/portage/env/... "-O3"... yikes! Not really recommended. If you want speedup, use "-march=native" instead. In make.conf, I have... CFLAGS="-O2 -march=native -mfpmath=sse -fomit-frame-pointer -pipe -fno-unwind-tables -fno-asynchronous-unwind-tables" The 2 "unwind_tables" flags reduce code bloat. See http://comments.gmane.org/gmane.linux.busybox/36695 for details. The guys at busybox do what they can to reduce code bloat. Knocking 15% off the size of an executable or library helps. This results in faster loading at startup (less to load), and less stuff in ram cuts down on swapping. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
Many thanks for your hints! On 05/14/2017 07:13:05 AM, Walter Dnes wrote: Why aren't you running "-march=native"? You're probably missing quite a few features of your cpu. BTW, "-march=native" implies "-mtune=native" and "-msse3" (if the cpu supports it). Out of sheer curiousity, what does gcc report as your native cpu when you run... gcc -c -Q -march=native --help=target | grep march= -march=amdfam10 > and quite a e few > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > (for my aged AMD64 Phenom II machine) > > in /etc/portage/env/... "-O3"... yikes! Not really recommended. If you want speedup, use "-march=native" instead. I didn't have any problems with that. I only use these setting non compute intensive packages which are not system critical. In make.conf, I have... CFLAGS="-O2 -march=native -mfpmath=sse -fomit-frame-pointer -pipe -fno-unwind-tables -fno-asynchronous-unwind-tables" The 2 "unwind_tables" flags reduce code bloat. See http://comments.gmane.org/gmane.linux.busybox/36695 for details. The guys at busybox do what they can to reduce code bloat. Knocking 15% off the size of an executable or library helps. This results in faster loading at startup (less to load), and less stuff in ram cuts down on swapping. Many thanks, again Helmut
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Sat, May 13, 2017 at 10:21:19AM +0200, Helmut Jarausch wrote > > I have switched to gcc-6.3.0 at the end of last year. > I have (re-)compiled all my packages including icewm-1.3.12-r1. > Everything works just fine. > > I did have some segfaults (with gimp) until I recompiled each package > with gcc-6.3.0 > > I have > CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" > > in /etc/portage/make.conf Why aren't you running "-march=native"? You're probably missing quite a few features of your cpu. BTW, "-march=native" implies "-mtune=native" and "-msse3" (if the cpu supports it). Out of sheer curiousity, what does gcc report as your native cpu when you run... gcc -c -Q -march=native --help=target | grep march= > and quite a e few > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > (for my aged AMD64 Phenom II machine) > > in /etc/portage/env/... "-O3"... yikes! Not really recommended. If you want speedup, use "-march=native" instead. In make.conf, I have... CFLAGS="-O2 -march=native -mfpmath=sse -fomit-frame-pointer -pipe -fno-unwind-tables -fno-asynchronous-unwind-tables" The 2 "unwind_tables" flags reduce code bloat. See http://comments.gmane.org/gmane.linux.busybox/36695 for details. The guys at busybox do what they can to reduce code bloat. Knocking 15% off the size of an executable or library helps. This results in faster loading at startup (less to load), and less stuff in ram cuts down on swapping. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On 170513-17:25+0200, Helmut Jarausch wrote: > On 05/13/2017 04:07:52 PM, Miroslav Rovis wrote: > > On 170513-10:21+0200, Helmut Jarausch wrote: > > > On 05/13/2017 05:56:16 AM, Walter Dnes wrote: > > > > On Wed, May 10, 2017 at 08:38:56PM -0400, Walter Dnes wrote > > > > > "Walter's Excellent Adventure Continues"... Now that I've got > > ICEWM > > > > > built, I'm trying to start X. I copied over my ~/.icewm > > cirectory, > > ... > > > > > cursor... for half a second or so. Then X exits immediately. I > > ... > > > > This message is coming to you from an almost entirely GCC 6.3.0 > > > > machine. I had to build ICEWM with GCC 5.4.0 to stop it > > segfaulting. > > > > The rest of Gentoo, including ICEWM's dependancies, is built with > > GCC > > > > 6.3.0. > > > > > > > > > > I have switched to gcc-6.3.0 at the end of last year. > > > I have (re-)compiled all my packages including icewm-1.3.12-r1. > > > Everything works just fine. > > > > > > I did have some segfaults (with gimp) until I recompiled each > > package > > > with gcc-6.3.0 > > > > > > I have > > > CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" > > > > > > in /etc/portage/make.conf > > > > > > and quite a e few > > > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > > > (for my aged AMD64 Phenom II machine) > > > > > > in /etc/portage/env/... > > > > > > Helmut > > > > > I have a Phenom II machine as well. And I'd be interested what you > > needed to stick in the /etc/portage/env/, if you could tell me, pls. > > > > Here an example: if you want to set some environment variables for > sci-libs/atlas : > > First, create the directory > > /etc/portage/env/sci-libs > > Second, create the file 'atlas', e.g. by > > echo 'CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a > -m3dnow"' > atlas > I see. Thx. -- Miroslav Rovis Zagreb, Croatia https://www.CroatiaFidelis.hr signature.asc Description: Digital signature
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On 05/13/2017 04:07:52 PM, Miroslav Rovis wrote: On 170513-10:21+0200, Helmut Jarausch wrote: > On 05/13/2017 05:56:16 AM, Walter Dnes wrote: > > On Wed, May 10, 2017 at 08:38:56PM -0400, Walter Dnes wrote > > > "Walter's Excellent Adventure Continues"... Now that I've got ICEWM > > > built, I'm trying to start X. I copied over my ~/.icewm cirectory, ... > > > cursor... for half a second or so. Then X exits immediately. I ... > > This message is coming to you from an almost entirely GCC 6.3.0 > > machine. I had to build ICEWM with GCC 5.4.0 to stop it segfaulting. > > The rest of Gentoo, including ICEWM's dependancies, is built with GCC > > 6.3.0. > > > > I have switched to gcc-6.3.0 at the end of last year. > I have (re-)compiled all my packages including icewm-1.3.12-r1. > Everything works just fine. > > I did have some segfaults (with gimp) until I recompiled each package > with gcc-6.3.0 > > I have > CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" > > in /etc/portage/make.conf > > and quite a e few > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > (for my aged AMD64 Phenom II machine) > > in /etc/portage/env/... > > Helmut > I have a Phenom II machine as well. And I'd be interested what you needed to stick in the /etc/portage/env/, if you could tell me, pls. Here an example: if you want to set some environment variables for sci-libs/atlas : First, create the directory /etc/portage/env/sci-libs Second, create the file 'atlas', e.g. by echo 'CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow"' > atlas
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On 170513-10:21+0200, Helmut Jarausch wrote: > On 05/13/2017 05:56:16 AM, Walter Dnes wrote: > > On Wed, May 10, 2017 at 08:38:56PM -0400, Walter Dnes wrote > > > "Walter's Excellent Adventure Continues"... Now that I've got ICEWM > > > built, I'm trying to start X. I copied over my ~/.icewm cirectory, ... > > > cursor... for half a second or so. Then X exits immediately. I ... > > This message is coming to you from an almost entirely GCC 6.3.0 > > machine. I had to build ICEWM with GCC 5.4.0 to stop it segfaulting. > > The rest of Gentoo, including ICEWM's dependancies, is built with GCC > > 6.3.0. > > > > I have switched to gcc-6.3.0 at the end of last year. > I have (re-)compiled all my packages including icewm-1.3.12-r1. > Everything works just fine. > > I did have some segfaults (with gimp) until I recompiled each package > with gcc-6.3.0 > > I have > CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" > > in /etc/portage/make.conf > > and quite a e few > CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" > (for my aged AMD64 Phenom II machine) > > in /etc/portage/env/... > > Helmut > I have a Phenom II machine as well. And I'd be interested what you needed to stick in the /etc/portage/env/, if you could tell me, pls. -- Miroslav Rovis Zagreb, Croatia https://www.CroatiaFidelis.hr signature.asc Description: Digital signature
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On 05/13/2017 05:56:16 AM, Walter Dnes wrote: On Wed, May 10, 2017 at 08:38:56PM -0400, Walter Dnes wrote > "Walter's Excellent Adventure Continues"... Now that I've got ICEWM > built, I'm trying to start X. I copied over my ~/.icewm cirectory, the > .xinitrc, and local X startup script from my current machine, where it > all works. I run "startx", and the X window pops up and I can see the > cursor... for half a second or so. Then X exits immediately. I don't > see any ominous error messages. I'm attaching the contents of > Xorg.0.log in case it helps. This message is coming to you from an almost entirely GCC 6.3.0 machine. I had to build ICEWM with GCC 5.4.0 to stop it segfaulting. The rest of Gentoo, including ICEWM's dependancies, is built with GCC 6.3.0. I have switched to gcc-6.3.0 at the end of last year. I have (re-)compiled all my packages including icewm-1.3.12-r1. Everything works just fine. I did have some segfaults (with gimp) until I recompiled each package with gcc-6.3.0 I have CFLAGS="-mtune=native -O2 -msse3 -pipe -fPIC" in /etc/portage/make.conf and quite a e few CFLAGS="-O3 -mtune=native -pipe -msse -msse2 -msse3 -msse4a -m3dnow" (for my aged AMD64 Phenom II machine) in /etc/portage/env/... Helmut
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Wed, May 10, 2017 at 08:38:56PM -0400, Walter Dnes wrote > "Walter's Excellent Adventure Continues"... Now that I've got ICEWM > built, I'm trying to start X. I copied over my ~/.icewm cirectory, the > .xinitrc, and local X startup script from my current machine, where it > all works. I run "startx", and the X window pops up and I can see the > cursor... for half a second or so. Then X exits immediately. I don't > see any ominous error messages. I'm attaching the contents of > Xorg.0.log in case it helps. This message is coming to you from an almost entirely GCC 6.3.0 machine. I had to build ICEWM with GCC 5.4.0 to stop it segfaulting. The rest of Gentoo, including ICEWM's dependancies, is built with GCC 6.3.0. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
"Walter's Excellent Adventure Continues"... Now that I've got ICEWM built, I'm trying to start X. I copied over my ~/.icewm cirectory, the .xinitrc, and local X startup script from my current machine, where it all works. I run "startx", and the X window pops up and I can see the cursor... for half a second or so. Then X exits immediately. I don't see any ominous error messages. I'm attaching the contents of Xorg.0.log in case it helps. -- Walter DnesI don't run "desktop environments"; I run useful applications xorg.log.gz Description: Binary data
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Wed, May 10, 2017 at 02:15:23PM -0400, Walter Dnes wrote > On Mon, May 08, 2017 at 07:55:30AM +0200, Raffaele Belardi wrote > > Bringing a 2008 Core2 duo back into use. I ran into major problems > with graphite optimizations. So I re-installed without graphite. > > I copied over most of the world file from my current desktop (GCC > 5.4.0) and tried it on the totally GCC 6.3.0 system (without graphite). > Two problems... > > 1) icewm (This would be a showstopper) > > icesh.cc:249:1: error: narrowing conversion of '4294967295ul' from 'long > unsigned int' to 'long int' inside { } [-Wnarrowing] > }; > ^ > icesh.cc:261:1: error: narrowing conversion of '4294967295ul' from 'long > unsigned int' to 'long int' inside { } [-Wnarrowing] > }; > ^ A bit more Google-fu turned up the answer. Step 1) mkdir /etc/portage/env echo 'CXXFLAGS="${CXXFLAGS} -Wno-narrowing"' > /etc/portage/env/narrowing Step 2) create/edit /etc/portage/package.env adding the 2 lines games-board/xfreecell narrowing x11-wm/icewm narrowing -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Mon, May 08, 2017 at 07:55:30AM +0200, Raffaele Belardi wrote Bringing a 2008 Core2 duo back into use. I ran into major problems with graphite optimizations. So I re-installed without graphite. I copied over most of the world file from my current desktop (GCC 5.4.0) and tried it on the totally GCC 6.3.0 system (without graphite). Two problems... 1) icewm (This would be a showstopper) icesh.cc:249:1: error: narrowing conversion of '4294967295ul' from 'long unsigned int' to 'long int' inside { } [-Wnarrowing] }; ^ icesh.cc:261:1: error: narrowing conversion of '4294967295ul' from 'long unsigned int' to 'long int' inside { } [-Wnarrowing] }; ^ 2) xfreecell A whole slew of stuff like... boundingMask.bm:108:38: error: narrowing conversion of '224' from 'int' to 'char ' inside { } [-Wnarrowing] 0xff, 0xff, 0xff, 0xff, 0xff, 0x03}; ^ boundingMask.bm:108:38: error: narrowing conversion of '255' from 'int' to 'char ' inside { } [-Wnarrowing] Is there a way to turn off the "-Wnarrowing" check? If I can do that, the next step would be to create custom build environments, with modified CFLAGS/CXXFLAGS as per https://wiki.gentoo.org/wiki/Knowledge_Base:Overriding_environment_variables_per_package Is this sloppy programming that should be reported as a bug? -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
I've been running some number theory on my CPU 24/7 for the last month... It's a good burn-in because it uses all my cpu and all my ram. =P The run will complete a month from now, I hope... Kinda stoked about gcc 6.x coming, really looking forward to rebuilding my system, hope things go well by the time I'm ready. =) Walter Dnes wrote: > I have a 2008 dual core Core2 that I gave up on because it doesn't > support QEMU hardware acceleration. Now that I've figured out chroot > for stuff I need, I decided to do a new 32-bit install. I figured this > might be a chance to experiment with GCC 6.3.0. I went whole-hog and > enabled graphite, and used various -floop and -ftree CXXFLAGS. When I > tried updating @world on a new install, it was a Churchillian > "unmitigated disaster", with multiple program builds aborting with > "internal compiler error" messages complaining about something to do > with "loop". I blew away the partially completed install and started > again. This time, it'll be 6.3.0 without graphite. > > I'm busy Tuesday morning. So I'll fire up an "emerge -e @world" before > going to bed, and see where things are when I get back in the afternoon. -- Strange Game. The only winning move is not to play. Powers are not rights.
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On Tue, May 09, 2017 at 12:59:12AM -0400, Walter Dnes wrote > I blew away the partially completed install and started again. > This time, it'll be 6.3.0 without graphite. > > I'm busy Tuesday morning. So I'll fire up an "emerge -e @world" > before going to bed, and see where things are when I get back in > the afternoon. The "emerge -e" finished successfully overnight. Since this is at the initial install stage, there were only 156 packages to build, but at least that's progress. I'll finish the install this afternoon. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
I have a 2008 dual core Core2 that I gave up on because it doesn't support QEMU hardware acceleration. Now that I've figured out chroot for stuff I need, I decided to do a new 32-bit install. I figured this might be a chance to experiment with GCC 6.3.0. I went whole-hog and enabled graphite, and used various -floop and -ftree CXXFLAGS. When I tried updating @world on a new install, it was a Churchillian "unmitigated disaster", with multiple program builds aborting with "internal compiler error" messages complaining about something to do with "loop". I blew away the partially completed install and started again. This time, it'll be 6.3.0 without graphite. I'm busy Tuesday morning. So I'll fire up an "emerge -e @world" before going to bed, and see where things are when I get back in the afternoon. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
On May 8, 2017 1:47:22 PM GMT+02:00, Dalewrote: >Raffaele Belardi wrote: >> During the weekend I 'emerge -e' a couple of ~amd64 systems with >> gcc-6.3.0: >> >> 1. gnome desktop, 1000 packages, all build fine except: >> - net-libs/webkit-gtk, rebuilding it again after world fixed it >> (possibly an issue with -j MAKEOPTS, a similar build failure is >> mentioned in bug 515968 although it's for an older package version) >> - sci-electronics/gnucap (bug 594184) >> - games-action/armagetronad (bugs 610844, 617768) >> >> 2. lxde media center, 850 packages, all fine except: >> - games-sports/trigger >> - games-sports/torcs >> Haven't had time to process these yet. >> >> Looks like it's still better to keep a gcc-5.x version around. >> >> raffaele >> >> > > >I'm glad you posted this. I read on -dev the other day about them >thinking about skipping to gcc 7 or something. I was thinking about >jumping to gcc 6 for now. I was curious as to the problems tho since >it >is not likely to widely used yet. Then here comes your post about it. >You pretty much answered my question except that I use KDE where it >seems you use Gnome. > >You have ESP or something?? lol > >Anyone using KDE and gcc 6 yet? Problems? > >Dale > >:-) :-) Only recently got my desktop converted to 5.4. Next is my laptop. When that is done, I could risk 6 on my desktop. But if Dev is thinking of skipping it -- Joost -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] world rebuild with gcc-6.3.0 - not completely there yet
Raffaele Belardi wrote: > During the weekend I 'emerge -e' a couple of ~amd64 systems with > gcc-6.3.0: > > 1. gnome desktop, 1000 packages, all build fine except: > - net-libs/webkit-gtk, rebuilding it again after world fixed it > (possibly an issue with -j MAKEOPTS, a similar build failure is > mentioned in bug 515968 although it's for an older package version) > - sci-electronics/gnucap (bug 594184) > - games-action/armagetronad (bugs 610844, 617768) > > 2. lxde media center, 850 packages, all fine except: > - games-sports/trigger > - games-sports/torcs > Haven't had time to process these yet. > > Looks like it's still better to keep a gcc-5.x version around. > > raffaele > > I'm glad you posted this. I read on -dev the other day about them thinking about skipping to gcc 7 or something. I was thinking about jumping to gcc 6 for now. I was curious as to the problems tho since it is not likely to widely used yet. Then here comes your post about it. You pretty much answered my question except that I use KDE where it seems you use Gnome. You have ESP or something?? lol Anyone using KDE and gcc 6 yet? Problems? Dale :-) :-)