The new albert runs under FreeBSD in LINUX-COMPAT.
It runs 20% - 25% faster then the old application.
LINUX-COMPAT must be installed, COMPAT4x isn't needed.
Installation:
Wait until all work is finished.
#stop boinc
$ > /usr/local/etc/rc.d/boinc.sh stop
#Log in as boinc:
$ > su boinc
#Change to work dir
$ > cd /var/db/boinc/projects/einstein.phys.uwm.edu
#Get new albert
$ > wget \\ http://einstein.phys.uwm.edu/app_test/linux/albert_4.51_i686-pc-linux-gnu.tar.gz
#unpack
$ > tar -xzf albert_4.51_i686-pc-linux-gnu.tar.gz
#brandelf
$ > brandelf -t linux albert_4.51_i686-pc-linux-gnu
$ > brandelf -t linux albert_4.51_i686-pc-linux-gnu.so
#start boinc
$ > /usr/local/etc/rc.d/boinc.sh start
Thats all. We will distribute a package for easy installation in the next days on our team-site http://www.cmds.tk/
Done! Everybody is welcome to download Martins good work. Please take a look at the ReadMe inside the .tbz Package.
Warning: Some Versions of FireFox (my too) make an error by trying to open the package, it will be a good idea using: right-mouse-click and "save target as"
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
1 Hour 12 min and 33% done - getting close to 3 hours 30-ish per WU !
looking good !
Gray
Well back home again and all 3 apps have finished with times of 3 hours 30-odd minutes as I expected: quite a jump in performance time from the usual 4 hours twenty as my fastest time up to 4 hours 45 minutes as a variation, depending on which flavour of Linux / Window Manager I used.
have done 7 WUs up to now with 4.51.
It has roughly doubled the speed of execution.
That's fine!
But I got two invalid WUs, up to now.
Result- ID 24485779m, for instance.
But no hint in stderr out.
Machine is an AMD Sempron 2200+,
Kernel is 2.6.11-6mdk, Mandriva 2005 standard installation.
Boinc is 5.2.14, an optimized version.
Might that be the troublemaker?
8 WUs on another machine, whith the official boinc, went flawless.
Yep, I got some invalid results from 4.51 too, lately. I made a slightly modified version (4.55) preserving more precision at a certain point - might be slightly slower but give more accurate results. I'll put it up after some internal testing.
Hmm, I thought you're going to do a real pkg_add'able package... just an idea ;)
Everybody can install this prepared apps without consequences to the conistency of officiell ports. It's only a bad hack and i won't spend time to maintain a bad hack for the ports-tree.
If someone else maintain this for the portstree - ok.
Hi,
I got the signal 11 - error, e.g with this WU http://einsteinathome.org/task/24583106 on a Celeron 1300MHz with Suse 8.0.
After 'preparing' the albert_4.51_i686-pc-linux-gnu.so - file it seems to work.
RE: The new albert runs
)
Done! Everybody is welcome to download Martins good work. Please take a look at the ReadMe inside the .tbz Package.
Chris
http://www.cmds.tk
Warning: Some Versions of FireFox (my too) make an error by trying to open the package, it will be a good idea using: right-mouse-click and "save target as"
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
Hmm, I thought you're going
)
Hmm, I thought you're going to do a real pkg_add'able package... just an idea ;)
Still I'll wait for a native app.
RE: Hiya All Quick update
)
Well back home again and all 3 apps have finished with times of 3 hours 30-odd minutes as I expected: quite a jump in performance time from the usual 4 hours twenty as my fastest time up to 4 hours 45 minutes as a variation, depending on which flavour of Linux / Window Manager I used.
Thanks for the effort - Gray
Hello there, have done 7
)
Hello there,
have done 7 WUs up to now with 4.51.
It has roughly doubled the speed of execution.
That's fine!
But I got two invalid WUs, up to now.
Result- ID 24485779m, for instance.
But no hint in stderr out.
Machine is an AMD Sempron 2200+,
Kernel is 2.6.11-6mdk, Mandriva 2005 standard installation.
Boinc is 5.2.14, an optimized version.
Might that be the troublemaker?
8 WUs on another machine, whith the official boinc, went flawless.
Regards, bonnyscott
Argh! The first two of my
)
Argh!
The first two of my pendings with the beta app got 0 credits!
The exit status is allways 0 and outcome is success.???
The AMD 64 shure is SSE capable and I didn't use any graphics.
Any problem with the scheduler?
cu,
Michael
By now I worked 2 WUs with
)
By now I worked 2 WUs with 4.51. Both WUs run invalid. http://einsteinathome.org/task/24468709
http://einsteinathome.org/task/24375621
Proz: Athlon XP1700+
Yep, I got some invalid
)
Yep, I got some invalid results from 4.51 too, lately. I made a slightly modified version (4.55) preserving more precision at a certain point - might be slightly slower but give more accurate results. I'll put it up after some internal testing.
BM
BM
RE: Hmm, I thought you're
)
Everybody can install this prepared apps without consequences to the conistency of officiell ports. It's only a bad hack and i won't spend time to maintain a bad hack for the ports-tree.
If someone else maintain this for the portstree - ok.
CU
Martin
Hi, I got the signal 11 -
)
Hi,
I got the signal 11 - error, e.g with this WU http://einsteinathome.org/task/24583106 on a Celeron 1300MHz with Suse 8.0.
After 'preparing' the albert_4.51_i686-pc-linux-gnu.so - file it seems to work.
Regards, Axel
I got a 45% speedup on my 400
)
I got a 45% speedup on my 400 MHz Pentium II (Deschutes), SuSE Linux 9.3, using Albert 4.51.
Tullio