T2 IRC Log: 2009-11-18
This is the log as captured by an IRC bot in the channel. The statements are those of the individual people and might not neccessarily reflect the policy and legal rules as set forth by the T2 SDE Project.
--- Log opened Wed Nov 18 00:00:17 2009
02:01 -!- mqueiros [n=mqueiros@217.70.75.13] has quit ["Ex-Chat"]
02:02 -!- mqueiros [n=mqueiros@217.70.75.13] has joined #t2
02:16 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has quit ["Verlassend"]
04:42 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
04:42 -!- koan [n=koan@unaffiliated/koan] has joined #t2
05:12 -!- TobiX [i=tobias@zoidberg.org] has quit [Read error: 104 (Connection reset by peer)]
08:37 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
08:37 -!- koan [n=koan@unaffiliated/koan] has joined #t2
08:43 < rxr> re
08:55 -!- Netsplit kornbluth.freenode.net <-> irc.freenode.net quits: mj-netbits_, mjungwirth_
09:02 -!- Netsplit over, joins: mjungwirth_
09:02 < rxr> mtr: at least ppc64 and sparc64 regressions due gcc44
09:03 < rxr> namely ppc64/linux26 and sparc64/silo
09:04 -!- mj-netbits_ [n=mj-netbi@213.174.252.112] has joined #t2
09:31 -!- Baldzius [n=aldas@87-198-133-94.ptr.magnet.ie] has joined #t2
09:31 < Baldzius> moin
09:31 < mjungwirth_> rxr have you add the patch to the archtiecture directory ?
09:32 < rxr> mjungwirth_: the atom opts?
09:32 < mjungwirth_> yes and viac7 settings
09:32 < mjungwirth_> i have changed this settings
09:33 < rxr> yeah - wanted to come back to you
09:33 < rxr> the -O2 does not really belog there, will mostly have no effect at all
09:33 < rxr> there is a global T2 optimiation options lazy(O1), speed(O2), size(Os), none(O0)
09:34 < rxr> the via c3-2 once worke
09:34 < rxr> d
09:34 < rxr> even is in the gcc manpage
09:34 < rxr> it was not accepted by gcc(, anymore)?
09:35 < rxr> " c3-2
09:35 < rxr> Via C3-2 CPU with MMX and SSE instruction set support."
09:52 < rxr> why do you differentiate the 64bit atoms 230 and 330 ?
10:02 < mjungwirth_> i found this setting on the gentoo gcc-optimize documentation
10:02 < rxr> I was under the impression the 230 and 330 cores are identical
10:03 < rxr> the 330 just have 2 of them, ....
10:03 < mjungwirth_> and viac7 has an i686 architecture
10:03 < mjungwirth_> hmm
10:03 < rxr> yeah, I know, I just wonder if the c3-2 settings did not work?
10:03 < rxr> what was the problem with the existing c7 optimization =?
10:04 < mjungwirth_> it works but it's the incorrect optimization.
10:06 < rxr> shouldn't that be near identical? have you measured that the i686 + sse et al. produces better code than the c3-2 + sse et al.?
10:07 < mjungwirth_> i can make e test if you want. i think i686 code is faster than i586.
10:07 < rxr> also, does 686 produce CMOV?
10:08 < rxr> I remeber something that CMOV with via only appear late, maybe the first c7 did not have it?
10:15 < rxr> ah, that was C3 Nehemiah (C5XL)
10:15 < mjungwirth_> yes
10:15 < mjungwirth_> via c7 support CMOV
10:15 < mjungwirth_> i can't found different
10:16 < mjungwirth_> Codename C5J
10:20 < rxr> i have a c7 here as well
10:21 < rxr> if you have some time a little bench would be nice
10:21 < mjungwirth_> yes i can make but only at home. this weekend is a good time.
10:25 < mjungwirth_> hmm Freebsd 6.3 says
10:25 < mjungwirth_> CPU: VIA C7 Esther+RNG+AES+AES-CTR+SHA1+SHA256+RSA
10:25 < mjungwirth_> Origin = "CentaurHauls" Id = 0x6a9 Stepping = 9
10:25 < mjungwirth_> Features=0xa7c9bbff
10:25 < mjungwirth_> Features2=0x181
10:29 < mjungwirth_> which benchmark tool you use?
10:30 < rxr> you could try my openbench scripts, but they are a little raw, tricky
10:30 < rxr> just a single CPU intensive tool with changed optimizations would be a good quick test
10:31 < mjungwirth_> ist this a t2-package?
10:31 < mjungwirth_> is
10:33 < rxr> no :-)
10:33 < rxr> best use just some cpu crunching program
10:33 < rxr> but mind to avoid one with assembly optimizations
10:33 < rxr> http://www.exactcode.de/site/open_source/openbench/
10:34 < mjungwirth_> oh nice
10:40 < rxr> but it's rough
10:40 < rxr> used it mostly for the linux magazine gcc articles
11:52 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
11:52 -!- koan_ [n=koan@unaffiliated/koan] has joined #t2
11:54 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has joined #t2
15:26 -!- digitaloktay_ [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has joined #t2
15:43 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has quit [Read error: 110 (Connection timed out)]
16:39 -!- Baldzius [n=aldas@87-198-133-94.ptr.magnet.ie] has quit ["Leaving"]
16:57 -!- digitaloktay_ [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has quit ["Verlassend"]
17:00 -!- koan_ is now known as koan
17:07 -!- digitaloktay [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has joined #t2
19:36 -!- digitaloktay_ [n=noip@dslb-088-064-175-147.pools.arcor-ip.net] has joined #t2
19:53 -!- digitaloktay [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has quit [Connection timed out]
21:19 < mj-netbits_> someone here
21:19 < mj-netbits_> i think taglib needs cmake
22:07 < mj-netbits_> mc-mp can't compile on x86 system with pentium (i586) optimize
--- Log closed Thu Nov 19 00:00:17 2009
02:01 -!- mqueiros [n=mqueiros@217.70.75.13] has quit ["Ex-Chat"]
02:02 -!- mqueiros [n=mqueiros@217.70.75.13] has joined #t2
02:16 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has quit ["Verlassend"]
04:42 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
04:42 -!- koan [n=koan@unaffiliated/koan] has joined #t2
05:12 -!- TobiX [i=tobias@zoidberg.org] has quit [Read error: 104 (Connection reset by peer)]
08:37 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
08:37 -!- koan [n=koan@unaffiliated/koan] has joined #t2
08:43 < rxr> re
08:55 -!- Netsplit kornbluth.freenode.net <-> irc.freenode.net quits: mj-netbits_, mjungwirth_
09:02 -!- Netsplit over, joins: mjungwirth_
09:02 < rxr> mtr: at least ppc64 and sparc64 regressions due gcc44
09:03 < rxr> namely ppc64/linux26 and sparc64/silo
09:04 -!- mj-netbits_ [n=mj-netbi@213.174.252.112] has joined #t2
09:31 -!- Baldzius [n=aldas@87-198-133-94.ptr.magnet.ie] has joined #t2
09:31 < Baldzius> moin
09:31 < mjungwirth_> rxr have you add the patch to the archtiecture directory ?
09:32 < rxr> mjungwirth_: the atom opts?
09:32 < mjungwirth_> yes and viac7 settings
09:32 < mjungwirth_> i have changed this settings
09:33 < rxr> yeah - wanted to come back to you
09:33 < rxr> the -O2 does not really belog there, will mostly have no effect at all
09:33 < rxr> there is a global T2 optimiation options lazy(O1), speed(O2), size(Os), none(O0)
09:34 < rxr> the via c3-2 once worke
09:34 < rxr> d
09:34 < rxr> even is in the gcc manpage
09:34 < rxr> it was not accepted by gcc(, anymore)?
09:35 < rxr> " c3-2
09:35 < rxr> Via C3-2 CPU with MMX and SSE instruction set support."
09:52 < rxr> why do you differentiate the 64bit atoms 230 and 330 ?
10:02 < mjungwirth_> i found this setting on the gentoo gcc-optimize documentation
10:02 < rxr> I was under the impression the 230 and 330 cores are identical
10:03 < rxr> the 330 just have 2 of them, ....
10:03 < mjungwirth_> and viac7 has an i686 architecture
10:03 < mjungwirth_> hmm
10:03 < rxr> yeah, I know, I just wonder if the c3-2 settings did not work?
10:03 < rxr> what was the problem with the existing c7 optimization =?
10:04 < mjungwirth_> it works but it's the incorrect optimization.
10:06 < rxr> shouldn't that be near identical? have you measured that the i686 + sse et al. produces better code than the c3-2 + sse et al.?
10:07 < mjungwirth_> i can make e test if you want. i think i686 code is faster than i586.
10:07 < rxr> also, does 686 produce CMOV?
10:08 < rxr> I remeber something that CMOV with via only appear late, maybe the first c7 did not have it?
10:15 < rxr> ah, that was C3 Nehemiah (C5XL)
10:15 < mjungwirth_> yes
10:15 < mjungwirth_> via c7 support CMOV
10:15 < mjungwirth_> i can't found different
10:16 < mjungwirth_> Codename C5J
10:20 < rxr> i have a c7 here as well
10:21 < rxr> if you have some time a little bench would be nice
10:21 < mjungwirth_> yes i can make but only at home. this weekend is a good time.
10:25 < mjungwirth_> hmm Freebsd 6.3 says
10:25 < mjungwirth_> CPU: VIA C7 Esther+RNG+AES+AES-CTR+SHA1+SHA256+RSA
10:25 < mjungwirth_> Origin = "CentaurHauls" Id = 0x6a9 Stepping = 9
10:25 < mjungwirth_> Features=0xa7c9bbff
10:25 < mjungwirth_> Features2=0x181
10:29 < mjungwirth_> which benchmark tool you use?
10:30 < rxr> you could try my openbench scripts, but they are a little raw, tricky
10:30 < rxr> just a single CPU intensive tool with changed optimizations would be a good quick test
10:31 < mjungwirth_> ist this a t2-package?
10:31 < mjungwirth_> is
10:33 < rxr> no :-)
10:33 < rxr> best use just some cpu crunching program
10:33 < rxr> but mind to avoid one with assembly optimizations
10:33 < rxr> http://www.exactcode.de/site/open_source/openbench/
10:34 < mjungwirth_> oh nice
10:40 < rxr> but it's rough
10:40 < rxr> used it mostly for the linux magazine gcc articles
11:52 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"]
11:52 -!- koan_ [n=koan@unaffiliated/koan] has joined #t2
11:54 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has joined #t2
15:26 -!- digitaloktay_ [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has joined #t2
15:43 -!- digitaloktay [n=noip@dslb-084-056-249-088.pools.arcor-ip.net] has quit [Read error: 110 (Connection timed out)]
16:39 -!- Baldzius [n=aldas@87-198-133-94.ptr.magnet.ie] has quit ["Leaving"]
16:57 -!- digitaloktay_ [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has quit ["Verlassend"]
17:00 -!- koan_ is now known as koan
17:07 -!- digitaloktay [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has joined #t2
19:36 -!- digitaloktay_ [n=noip@dslb-088-064-175-147.pools.arcor-ip.net] has joined #t2
19:53 -!- digitaloktay [n=noip@dslb-084-057-228-075.pools.arcor-ip.net] has quit [Connection timed out]
21:19 < mj-netbits_> someone here
21:19 < mj-netbits_> i think taglib needs cmake
22:07 < mj-netbits_> mc-mp can't compile on x86 system with pentium (i586) optimize
--- Log closed Thu Nov 19 00:00:17 2009