I tried to compile seti_boinc-client-cvs-2005-08-12 on NetBSD and all i got was this error-message:

gmake  all-recursive
gmake[1]: Entering directory `/home/martin/dateien/software/seti_boinc'
Making all in jpeglib
gmake[2]: Entering directory `/home/martin/dateien/software/seti_boinc/jpeglib'
gmake[2]: *** No rule to make target `all'.  Stop.
gmake[2]: Leaving directory `/home/martin/dateien/software/seti_boinc/jpeglib'
gmake[1]: *** [all-recursive] Error 1
gmake[1]: Leaving directory `/home/martin/dateien/software/seti_boinc'
gmake: *** [all] Error 2

Is this problem known? Are there stable sources which compile?

---

Did the configure script runs successfull ?
You need only to compile all in client/, if you wan't to have a text only boinc client.

---

Did the configure script runs successfull ?
You need only to compile all in client/, if you wan't to have a text only boinc client.

The configure script run successfull. Though configure script from the boinc coreclient doesnt recognize wxGTK (which is installed in /usr/pkg), text only is sufficient for now. Yes, I added "-I/usr/pkg/include" to CFLAGS and CXXFLAGS.

cd client && gmake run successfully. Is it enough to cp setiathome-4.18.i386-unknown-netbsdelf3.99.7 to a projectdir?

---

Did the configure script runs successfull ?
You need only to compile all in client/, if you wan't to have a text only boinc client.

The configure script run successfull. Though configure script from the boinc coreclient doesnt recognize wxGTK (which is installed in /usr/pkg), text only is sufficient for now. Yes, I added "-I/usr/pkg/include" to CFLAGS and CXXFLAGS.

cd client && gmake run successfully. Is it enough to cp setiathome-4.18.i386-unknown-netbsdelf3.99.7 to a projectdir?

Good !

You have to run now your setiathome binaries against the reference workunit to vailidate, if the results wich your binaries computes are OK.
If the binary was to heavy optimized, it could be, that the results are out of range, and the results wil not be vailidated from berkeley.

Ned Slider have a very good documentation on his site, how to run and validate the results : http://www.pperry.f2s.com/boinc-compile-seti.htm
The validation limits are here : http://www.pperry.f2s.com/files/validate.txt

This reference WU should run even, you make changes to the optimation. It's also a good method to let the reference WU run as benchmark. For example with "time ./setiathome...".

Btw. : Do you wan't to share your binaries with other users ? - If you wish, it is posible to add a link from the third party site (http://boinc.berkeley.edu/download_other.php) to your download site. If you wan't to have a link, report it, and I will report it to Berkley. If you don't have webspace, it should be no problem, there are some people, who won't host binaries for other users.

Regards

Lars

---

You have to run now your setiathome binaries against the reference workunit to vailidate, if the results wich your binaries computes are OK.

I hope you mean seti_boinc/client/test_workunits ? Because this is already running... :)

If the binary was to heavy optimized, it could be, that the results are out of range, and the results wil not be vailidated from berkeley.

Ok, i haven't optimized anything in this first build.

Ned Slider have a very good documentation on his site, how to run and validate the results : http://www.pperry.f2s.com/boinc-compile-seti.htm
The validation limits are here : http://www.pperry.f2s.com/files/validate.txt

This reference WU should run even, you make changes to the optimation. It's also a good method to let the reference WU run as benchmark. For example with "time ./setiathome...".

Thanks. I will try them, too.

Btw. : Do you wan't to share your binaries with other users ? - If you wish, it is posible to add a link from the third party site (http://boinc.berkeley.edu/download_other.php) to your download site.

I think it's better to update the already existing ports in pkgsrc. See
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-coreclient/
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-libs/
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-seti/

and

http://www.pkgsrc.org
http://www.netbsd.org/Documentation/software/packages.html#platforms

It couldn't use them because version 3 is too old, as the server told me :) A NetBSD Problem report has been filed already.
[/quote]

---

Btw. : Do you wan't to share your binaries with other users ? - If you wish, it is posible to add a link from the third party site (http://boinc.berkeley.edu/download_other.php) to your download site.

I think it's better to update the already existing ports in pkgsrc. See
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-coreclient/
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-libs/
http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/net/boinc3-seti/

and

http://www.pkgsrc.org
http://www.netbsd.org/Documentation/software/packages.html#platforms

It couldn't use them because version 3 is too old, as the server told me :) A NetBSD Problem report has been filed already.

Hm. Too old. No project support V.3 boinc clients anymore.
But I see the problem, that a user don't know the links to the bsd sites. So it is better to collect the links on a site, where the user can fast find a matching binary for his plattform.

What is this packagesrv ? - It this a collection of needed fixes to compile it correctly ? Found no binary. For which Platfrom are this stuff,.... ????
Sorry, but looks confusing for outstanding (NON NetBSD) people.

---

You have to run now your setiathome binaries against the reference workunit to vailidate, if the results wich your binaries computes are OK.

I hope you mean seti_boinc/client/test_workunits ? Because this is already running... :)

The test_workunit completed, but there are differences between the reference result and my computed result:

http://www.bnro.de/~martin81/seti.diff

However, the reference result seems to be really old. Can the differences be explained by the differing application version with which they were computed?

---

The test_workunit completed, but there are differences between the reference result and my computed result:

http://www.bnro.de/~martin81/seti.diff

However, the reference result seems to be really old. Can the differences be explained by the differing application version with which they were computed?

You have to compare your results with the reference_results, if they are within the validation limit.
But your client detects only one of two gaussians. So far I know is it a problem in the source, and the results are not valid. But I am not shure !

---

But your client detects only one of two gaussians. So far I know is it a problem in the source, and the results are not valid. But I am not shure !

So, if it's a problem in the source-code, which sources should i use then? I only found the cvs snapshots and no stable versions. Which cvs snapshot is known to work correctly?

---