Well that just sounds about 500 times easier than getting it to run under windows.

Nah, windows is about the same, stop boinc, drop the optimized app in C:\\\\program files\\boinc\\projects\\setiathome.berkeley.edu

then fire boinc back up, look for above message.

LOL you're right of course, I'm just without thinking - factoring in the "make antivirus ignore boinc directories", "open ports for remote control", "receive multiple errors and retry with nographics version" then dismantle machine to improve cooling.

Well that just sounds about 500 times easier than getting it to run under windows.

Ok, I down loaded the optimixer app from the number crunching thread and un zipped it. But When I try to run it I get an error code that to the effect that there is no application to run the optimizer. Even though I have 3 intel macs running, I am a PC person. Can any help me get my mac's running faster?

Many thanks, Fuzzy (and everybody) for your help. Fuzzy, is it legal to transform locked tunes (like iTunes) into unlocked tunes where I can still play the unlocked iTunes (copied from storage CDs, etc) in case there are no more iPods, etc, 20 years from now and my iPod then breaks? Also is there any noticeable loss of fidelity when transforming an iTune into regular unlocked format?

Thank all the hard-working optimizers, as well :o)

I just compile their code, no wizardry there...

They are, in no particular order,

Rev-2.0
Josef W. Segur
Ben Herndon
Alex Kan

Rev 1.41
Michael Harrer

Rev 1.3
Hans Dorn, me (small code fixes, not performance related)

Also, the numerous members of the pre-release test community - without their help in verifying performance and validity on as many platforms as possible, we couldn't have gotten there either.

Lastly, thank the fact that I'm not married and hence have enough free time to devote to this project ;o)

Ni!
Simon.

I would also like to thank you and all of those that put in time on optimizing, but for 1 quick question, will there be a version 2.0 for linux?

Thank all the hard-working optimizers, as well :o)

I just compile their code, no wizardry there...

They are, in no particular order,

Rev-2.0
Josef W. Segur
Ben Herndon
Alex Kan

Rev 1.41
Michael Harrer

Rev 1.3
Hans Dorn, me (small code fixes, not performance related)

Also, the numerous members of the pre-release test community - without their help in verifying performance and validity on as many platforms as possible, we couldn't have gotten there either.

Lastly, thank the fact that I'm not married and hence have enough free time to devote to this project ;o)

Ni!
Simon.

...

OK, so lots of "boinc" in there. That little lot assumes that you've installed into the boinc user home directory in a directory also called boinc. Also, you need to create ...

And instead of manually hacking all that lot native-style, I've just discovered that some kind souls have 'packaged' boinc for some of the main Linux distros so that you can 'install' boinc (one-click style) just like any other application. The ones mentioned are:

Debian (use "apt-get")
Gentoo (use "portage")
Ubuntu, Kubuntu, Xubuntu (use "apt-get")

(And various boinc/seti@home utilities are packaged for Mandriva "urpmi", but not boinc itself. However, I have vague recollection of there being a Fedora Core Linux rpm package for boinc somewhere... And that should work on most Linux rpm-supported distros ok.)


Happy crunchin',
Martin

[edit] There is also the thread "Linux question" that also discusses this. [/edit]

It's not damaged, it's a "Dash 9" result overflow. This just means that due to noise, RFI, etc. there were more "signals" found than is allowed for storage.

HTH,

Alinator

....and this means what? Is my HD space not enough? Sorry for this stupid question, but i want to know, what happens excatly. In the past i had no such problems. Only since the last few month.

Busy forum here! Join our team or just join our forum or IRC channel for SETI chat, or any chat for that matter!

Since the forum here does not lend itself to formatted text very well, here's a link to detailed instructions on how to get your very own optimized client for Linux compiled.

Instructions

If you should decide to compile your own client based on these instructions or have questions, feel free to post here.

Any changes and updates will be announced here.
This page includes download links to Seti@Home enhanced 5.15 sources that are known to compile.

I will outline all code changes done in a later post in this thread. Right now I need a break :o)

Regards and good luck,
Simon

So far I haven't received any email - I asked for a small test-group for optimized clients. Thought there would be takers, actually :)

So let me ask again:

Are you interested in test-driving a new optimized application?

Email me at savant<you know what goes here>lunabyte.net if you would like to.

I'm almost at the point where I want to distribute the app - it's quick, the credit it claims is okay, and it runs on several machines without complaining (as it should, since it's now a fully static binary).

All I'm missing is a test group :o)

Regards,
Simon.

Do you have a P3 SSE version?

Yes.
i586/MMX/SSE/SSE2/SSE2-P4. No SSE3 because none of my machines support it :) Hans can help in that regard, maybe.

I've also written a few little scripts that I'd like to have running on a few more machines to gather some more meaningful test data.

You run one script, it runs the various clients (optimized vs .standard 5.12 as distributed) with a few test WUs. All of this happens offline and has no impact on BOINC or credits or anything. It's purely a performance and compatibility measure, for now - which is also why I'll only be sharing this version with a very few people.

All of which I'm hereby asking to keep what they get to themselves for now. Please do NOT distribute further.

Hans, a copy is underway.
tekwyzrd, email me.

All of you others who may be out there, be quick or have patience ;o)

Regards,
Simon.

Has anyone managed to successfully build either from CVS or from Eric's tarballs?

So far, I haven't had any luck either on Linux or Windows, though the Windows version seems to get further before chickening out.

Catastrophic error  could not open source file "..\\..\\..\\boinc\\lib\\exception.C"

...which is true, because that file just doesn't exist there in CVS boinc. However, it needs to have CVS boinc there so it won't throw a fit over the other missing function that Hans noticed. So now I'm caught between a rock and a hard place :)

The file exists in Eric's source tree. So next I'm trying to just copy exception.C/h over, since Eric's tree failed with a different error before on Windows for me too.

In any case, Fedora Core 3 seems to be very nice in regards to Seti and BOINC sources (it didn't complain once about anything missing). Not to plug it here, since RH derivatives still give me the shivers, but building on it vs. Debian or even FC 5 (which I originally got) seems to work much better.

--edit
Whee, just one more error left on the Windows compile :)

..\\worker.cpp(150):  error: '_alloca' cannot be called from within a catch block

...and the code passage it references (when I double click on that error it goes to the line producing it), it states it's

if (retval) SETIERROR(retval,"from initialize_for_wu() in read_wu_state()");

Cute :) An error message throwing an error, fitting I guess (I did remove the dual SETIERROR definition, because without that it would error sooner, leaving the one with the semicolon at the end).

hmmmm....still getting errors with the tarball from your site, the g++ no input files is gone, but it was replaced with :

make[2]: *** [seti_boinc-main.o] Error 1
make[2]: Leaving directory `/home/boog/build_boinc/seti_boinc/client'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/boog/build_boinc/seti_boinc'
make: *** [all] Error 2

the other one is still the same:

I'm hoping the s_util.h fix got rid of this one...

Don't worry too much abouyt the fakedata errors, unless you need to build fake workunits for testing purposes. For people who want to use SETI@home for their own data analysis (attention SETILeague), fakedata.cpp might be of interest. Then again, the splitter might be useful as well for such things.

Eric

make[2]: *** [fakedata-xml_util.o] Error 1

I'm still getting both of those errors using either your tarball, or the nightly build from the 11th

I wouldn't worry about the fakedata error either if it didn't stop the compile process ;)

I'll go back and rebuild boinc to see if that is where my problem lies.

hmmmm....still getting errors with the tarball from your site, the g++ no input files is gone, but it was replaced with :

make[2]: *** [seti_boinc-main.o] Error 1
make[2]: Leaving directory `/home/boog/build_boinc/seti_boinc/client'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/boog/build_boinc/seti_boinc'
make: *** [all] Error 2

the other one is still the same:

I'm hoping the s_util.h fix got rid of this one...

Don't worry too much abouyt the fakedata errors, unless you need to build fake workunits for testing purposes. For people who want to use SETI@home for their own data analysis (attention SETILeague), fakedata.cpp might be of interest. Then again, the splitter might be useful as well for such things.

Eric

make[2]: *** [fakedata-xml_util.o] Error 1

I guess that's enough of "RPM hell" for me.

I'll try building on an older Debian version next...

Regards Hans

Thanks for your help Tetsuji!

I've got all necessary stuff (sources from Eric's links, Intel compiler + IPP + licenses), but I just can't get seti_boinc to configure. It keeps giving me this error (haven't seen that one before):

checking size of long int... configure: error: cannot compute sizeof (long int), 77
See `config.log' for more details.

config.log

configure:42105: checking size of long int
configure:42424: g++ -o conftest  -I/usr/include -I/usr/include/openssl -I/opt/intel/ipp/5.1/ia32/include -I/opt/intel/ipp/5.1/ia32/
tools/staticlib   -L/usr/lib -nodefaultlibs -L/opt/intel/ipp/5.1/ia32/lib conftest.cc  -lssl -lcrypto /usr/lib/libcrypto.a /usr/lib/
libssl.a -Wl,-Bdynamic -ldl -Wl,-Bdynamic -lm /usr/lib/libnsl.a -Wl,-Bdynamic -lrt /usr/lib/libz.a -Wl,-Bdynamic -lgcc_eh -Wl,-Bdyna
mic -lpthread -Wl,-Bdynamic -lc -lippcore -lippsmerged >&5
/tmp/cc6OoaL7.o:(.eh_frame+0x11): undefined reference to `__gxx_personality_v0'
collect2: ld returned 1 exit status
configure:42427: $? = 1     
configure: program exited with status 1
configure: failed program was:
| /* confdefs.h.  */
|
| #define PACKAGE_NAME "setiathome_enhanced"
| #define PACKAGE_TARNAME "setiathome_enhanced"
| #define PACKAGE_VERSION "5.12"
| #define PACKAGE_STRING "setiathome_enhanced 5.12"

...and so on.

Anyone have an idea how to get it to go over that configure check? I'm pretty sure my libs are up to date. Will do an apt-get update/upgrade and try again then, but I'm not very hopeful about that.

Other than that, it compiled the BOINC client using the intel compiler fine.

I can help some for Linux (and for FreeBSD, the same applies.) here. You need gcc-3.3 or 3.4 (or maybe 3.2). 4.x doesn't work for me.

As for ICC & IPP, I usually use this way (sse2 for example, from my memo, I copy & paste these commands. This is a good point in unix-like OS's):

in boinc directory (after ./_autosetup):

1. ./configure --disable-server --enable-maintainer-mode --disable-client

2. make clean

3. make AR=xiar CC=icc CXX=icpc CFLAGS="-sox -xW -O3 -fp-model fast=2 -no-prec-div -no-prec-sqrt -ipo" CXXFLAGS="-sox -xW -O3 -fp-model fast=2 -no-prec-div -no-prec-sqrt -ipo" MYSQL_CONFIG=true

(if you are using icc 9.0, you should use "-fp-model fast")

in seti_boinc directory,
1. ./configure --disable-server --enable-sse2 --with-ipp=/opt/intel/ipp/5.1/ia32 --enable-maintainer-mode --disable-graphics

2. make clean

3. make AR=xiar CC=icc CXX=icpc CFLAGS="-sox -xW -O3 -fp-model fast=2 -no-prec-div -no-prec-sqrt -ipo" CXXFLAGS="-sox -xW -O3 -fp-model fast=2 -no-prec-div -no-prec-sqrt -ipo" LDFLAGS="-L/opt/intel/ipp/5.1/ia32/lib -lsvml -i-static" MYSQL_CONFIG=true

The environment variables "USE_IPP" or "USE_SSE2" or ipp libraries doesn't have to be given explicitly. "./configure --with-ipp=.... --enable-sse2" will give these macros in config.h, Makefile automatically (thanks Eric!!)

Now I'm actively using ICC for linux for other apps like scilab.

I'm just downloading a Centos ISO :o)

Regards Hans

P.S: I'm giving up for now as well...