Posts by Rene


log in
1) Questions and Answers : Wish list : build without graphics (X11) (Message 875812)
Posted 15 Mar 2009 by Profile Rene
I've made an update to the port (not committed yet) to make X11 optional and to update Astropulse to version 5.03 It can be fetched from [1]. It is also available as a patch to the current port version from [2]

The clients compile but if you don't select X11 with Astropulse, the Astropulse client will crash when starting up :( If anyone knows a *proper* fix (i.e. not adding graphics-only include files) for this, let me know.

[1] ftp://rene-ladan.nl/freebsd/boinc-setiathome-enhanced.tar.bz2 (3346 bytes, MD5 = 7316bdc690fc353225a3d4184e30c318)
[2] ftp://rene-ladan.nl/freebsd/boinc-setiathome-enhanced.diff (7033 bytes, MD5 = a6c6587746d78e9e169d508773f1900b)
2) Questions and Answers : Wish list : exited with zero status but no 'finished' file (Message 868087)
Posted 22 Feb 2009 by Profile Rene
I get these messages from time to time on my FreeBSD box, usually when the cpu is drowned in work (load > number of cores). The excessive load causes the BOINC manager to not pick up the "I'm alive" signal from the science processes, and consequently thinks they have terminated, so it restarts them. The excessive load can also cause the science processes to be asleep for more than 30 seconds, in which case they don't send the "I'm alive signal", but are still running.
3) Questions and Answers : Wish list : build without graphics (X11) (Message 852479)
Posted 11 Jan 2009 by Profile Rene
I figured out that the problem is specific to astropulse, setiathome does respect the --disable-graphics option (and consequently builds a slightly smaller binary :) )
4) Questions and Answers : Wish list : build without graphics (X11) (Message 851313)
Posted 9 Jan 2009 by Profile Rene
After I ported the setiathome 6.03 and astropulse 5.00 to FreeBSD I've got some requests if it would be possible to compile the clients without graphics support. This way it is possible to run the clients on headless stations without having to install X11.

There is a --disable-graphics knob in the configure scripts but currently it is broken, because (as far as I can see) the state information of the math part of the astropulse client is stored inside a graphics structure (why?). This used to work fine in the previous version of the setiathome client (5.27).

I might be able to come up with some patches myself (against SVN head). Is there any interest for this?
5) Questions and Answers : Unix/Linux : Astropulse binaries for other platforms (Message 848172)
Posted 2 Jan 2009 by Profile Rene
I've published the diffs between archives as used by the FreeBSD port and the Berkeley sources at ftp://rene-ladan.nl/pub/freebsd/seti_boinc-r333-port.diff and ftp://rene-ladan.nl/pub/freebsd/astropulse-r354-port.diff
6) Questions and Answers : Unix/Linux : Astropulse binaries for other platforms (Message 846353)
Posted 29 Dec 2008 by Profile Rene
Today the FreeBSD port astro/boinc-setiathome-enhanced got updated. It installs setiathome 6.03 and astropulse 5.00 from source.


This would not have been possible without Dotsch' help. Thanks Dotsch!
7) Questions and Answers : Unix/Linux : Astropulse binaries for other platforms (Message 846140)
Posted 28 Dec 2008 by Profile Rene
Today the FreeBSD port astro/boinc-setiathome-enhanced got updated. It installs setiathome 6.03 and astropulse 5.00 from source.
8) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 846137)
Posted 28 Dec 2008 by Profile Rene
Today the astro/boinc-setiathome-enhanced port got updated. It now installs setiathome 6.03 and astropulse 5.00 (plus the screensavers) from source.
9) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 841401)
Posted 18 Dec 2008 by Profile Rene
I'm now using revision 333 of setiathome and revision 354 of astropulse.

The problem with that astropulse revision is that it effectively kills the --disable-graphics switch of configure, since the developers have made the IMO rather peculiar decision to put the current state into the graphics shared memory. I have some patches to make the switch useful again, but I guess I have to change the source code more than just adding some missing #ifdef BOINC_APP_GRAPHICS ... #endif tags to make it work.

The whole point of releasing setiathome 6 was that the science and graphics part are two separate pieces IIRC. Shouldn't astropulse do the same, for people who cannot or don't want to run the graphics part?

Updated diff to current astro/boinc-setiathome-enhanced port: ftp://rene-ladan.nl/pub/freebsd/boinc-setiathome-enhanced.diff (11695 bytes)
10) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 837825)
Posted 7 Dec 2008 by Profile Rene
Google doesn't know the fold_buf_{long|short}_{pos|neg} functions either.


ap_client_main.cpp: In member function 'void Astropulse::Client::init()':
ap_client_main.cpp:708: error: 'fold_buf_long_pos' was not declared in this scope
ap_client_main.cpp:709: error: 'fold_buf_short_pos' was not declared in this scope
ap_client_main.cpp:712: error: 'fold_buf_long_neg' was not declared in this scope
ap_client_main.cpp:713: error: 'fold_buf_short_neg' was not declared in this scope
Most possible an undeclared variable, because of work in progress. You could always tell svn to download an older revision.


I know. Is there a list of svn revisions used as releases?
11) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 837813)
Posted 7 Dec 2008 by Profile Rene
Google doesn't know the fold_buf_{long|short}_{pos|neg} functions either.
12) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 837811)
Posted 7 Dec 2008 by Profile Rene
Substituting sh with bash helped, but now I get a compile error for astropulse.
This is probably an error in the source code. I'm using SVN for both astropulse and setiathome:

self % svn info
Path: .
URL: https://setisvn.ssl.berkeley.edu/svn/astropulse
Repository Root: https://setisvn.ssl.berkeley.edu/svn
Repository UUID: ce18c45e-0eac-4cad-939e-398e9871e6ff
Revision: 372
Node Kind: directory
Schedule: normal
Last Changed Author: vonkorff
Last Changed Rev: 360
Last Changed Date: 2008-11-24 21:53:02 +0100 (Mon, 24 Nov 2008)

The error log is available at ftp://rene-ladan.nl/pub/freebsd/astropulse-compile-error.txt (42924 bytes)

The new port diff: ftp://rene-ladan.nl/pub/freebsd/sah-371-port.diff (11354 bytes)

Am I right that SVN always contains the latest source code?
13) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 836898)
Posted 4 Dec 2008 by Profile Rene
Strange, astrpoulse passed configure with sh on FreeBSD 7.0

But I will try bash.
14) Message boards : Number crunching : Astropulse binaries for other platforms - Read only (Message 836794)
Posted 3 Dec 2008 by Profile Rene
I've tested compiling my version of Seti@Home / Astropulse on FreeBSD 7.1/amd64 but it fails in the configure stage. This also happens on FreeBSD 7.1/i386.

The log is available at ftp://rene-ladan.nl/pub/freebsd/sah_371.txt (17219 bytes)

The diff to the current port (Seti@Home 5.27): ftp://rene-ladan.nl/pub/freebsd/sah-371-port.diff (11129 bytes)

The source diff for Seti@Home (svn 371): ftp://rene-ladan.nl/pub/freebsd/seti_boinc.diff (2724 bytes)

The source diff for Astropulse (svn 371):
ftp://rene-ladan.nl/pub/freebsd/astropulse.diff (1091 bytes)

Does anyone know what's going on?
15) Message boards : Number crunching : Optimized FreeBSD Builds? (Message 617957)
Posted 11 Aug 2007 by Profile Rene
Hey guys,

I've got a quad Xeon box that's running the amd64 version of freebsd, and I'd like to use the optimized build, but I can't seem to run the optimized 32-bit one for my proc (It always dies and core dumps).

I've tried installing a 32-bit userland, and making sure I have 32-bit compat on in the kernel. I guess what I really need is a 64 bit optimized bin that's geared towards intel instead of amd.

My problem seems to be that the 64 bit builds are geared for AMD's. I was running a non-optimized amd64 build, but it seemed my dual 2.4 proc box was outperforming my quad xeon without even running an optimized version.


You can try the ports version in astro/boinc-setiathome-enhanced (source build). If you set CPUTYPE correct in /etc/make.conf, it automatically optimizes for your box. It's currently at version 5.13, but I'm testing 5.27 now. The 5.13 version easily outperforms the stock Windows client.
16) Message boards : Number crunching : Welcome to the 7 year club (Message 504471)
Posted 17 Jan 2007 by Profile Rene
I just realized that I joined the seven year club - I joinged seti dec 29, 1999.

Christopher.


I have to wait another year...
I joined since January 2001
17) Questions and Answers : Unix/Linux : SETI Enhanced binaries for other Platforms (Message 389926)
Posted 6 Aug 2006 by Profile Rene
When I install ports/astro/boinc-setiathome-enhanced 5.13 on FreeBSD6.1,
I got these error messages...

analyzeFuncs.cpp: In function `int seti_analyze(ANALYSIS_STATE&)':
analyzeFuncs.cpp:351: error: `boinc_ops_cumulative' undeclared (first use this function)
analyzeFuncs.cpp:351: error: (Each undeclared identifier is reported only once for each function it appears in.)
*** Error code 1

Stop in /usr/ports/astro/boinc-setiathome-enhanced/work/boinc-setiathome-enhanced-5.13/client.
*** Error code 1

Stop in /usr/ports/astro/boinc-setiathome-enhanced/work/boinc-setiathome-enhanced-5.13.
*** Error code 1

Stop in /usr/ports/astro/boinc-setiathome-enhanced/work/boinc-setiathome-enhanced-5.13.
*** Error code 1

Stop in /usr/ports/astro/boinc-setiathome-enhanced.


Looks like you need to upgrade your net/boinc-client port. The header of the missing function is defined in boinc_api.h. This file is placed in /usr/local/include/BOINC when you install the net/boinc-client port.
18) Questions and Answers : Unix/Linux : SETI Enhanced binaries for other Platforms (Message 382607)
Posted 30 Jul 2006 by Profile Rene
Some people did not know of the existance and state of the different ports for other, not offical supported ports of SETI Enhanced for other platforms.

Here is a list of the available ports :

http://www.lb.shuttle.de/apastron/boincDown.shtml :
- Solaris x86 on AMD64 (Opteron)
- Linux on AMD64 (Opteron)
- Linux on Itanium (IA64)
- Linux on PPC and Power CPUs
- HPUX 10.20/11.0/11i (V1 and V2.) on PARISC
- HPUX 11i (V.1 and V.2) on IA64 (Itanium)
- FreeBSD i386
- NetBSD i386
- OpenBSD i386

http://www.freebsd.org/cgi/ports.cgi?query=boinc-setiathome&stype=all
- FreeBSD i386

http://groups.yahoo.com/group/seti-warp/message/1467
- OS/2 i386


The ports for AIX, Irix, Linux on SPARC, Solaris x86 for i386 and TRU64 are in work.


For your information, the FreeBSD ports collection now contains a version of the SAH Enhanced client which builds from source code, using the 2006-05-12 (5.13) tarball as astro/boinc-setiathome-enhanced.
Tested to be highly accurate on i386 (a Pentium 4M, with MMX and SSE2), feedback regarding other CPU architectures is welcome (my email address is in the port Makefile).

Regards,
Rene
19) Questions and Answers : Unix/Linux : Freebs 4.10 (Message 245102)
Posted 9 Feb 2006 by Profile Rene
... previously my pc was for server Linux 2.4.2 with -0 Mem :). ...


The Linuxolator lets uname() return 'Linux' (instead of something like 'Linux (FreeBSD)' ). The -0 MB RAM/swap was because you didn't mount linprocfs :)

I'm running the linux client (5.2.13) and get the same SETI@home errors as you. Looking in the results, it seems that there is some bug in the find_pulse() function. LHC@home and Einstein@home run fine (thats why I don't run the native version, LHC@home and Einstein@home don't yet support FreeBSD natively).

Seti results at http://setiathome.berkeley.edu/results.php?hostid=2164224

Rene


Copyright © 2014 University of California