climateprediction.net home page
Solaris Boinc

Solaris Boinc

Questions and Answers : Unix/Linux : Solaris Boinc
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user1265

Send message
Joined: 26 Aug 04
Posts: 13
Credit: 458,996
RAC: 0
Message 2225 - Posted: 30 Aug 2004, 20:40:50 UTC

Any hope of getting Solaris Sparc binaries?
I followed the links for other platforms, and was
able to build boinc itself, but then it failed to
download the proper cp binary for Solaris Sparc.

ID: 2225 · Report as offensive     Reply Quote
old_user17439

Send message
Joined: 13 Sep 04
Posts: 2
Credit: 43,535
RAC: 0
Message 4036 - Posted: 13 Sep 2004, 16:10:34 UTC

Seti claims to have a Solaris copy of BOINIC:

http://setiweb.ssl.berkeley.edu/download.php

However, I've not had any luck downloading it. It attempts to download the .gz file, then fails.

Rich
ID: 4036 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4054 - Posted: 13 Sep 2004, 20:12:32 UTC - in response to Message 4036.  
Last modified: 13 Sep 2004, 20:26:56 UTC

we would need a Solaris box and proper C++ & Fortran compilers, which unfortunately we lack. At one point Sun UK was going to donate a box but I never heard again. We never seem to get "freebies" -- probably because this climate model doesn't make any computers look like "speed demons!" :-)
There's a lot of things I'd like to do such as Solaris & AMD64 CPDN clients but unfortunately due to lack of resources (hardware & money for software i.e. compilers) it seems doubtful they'll get done.
ID: 4054 · Report as offensive     Reply Quote
old_user6609

Send message
Joined: 31 Aug 04
Posts: 10
Credit: 65,416
RAC: 0
Message 4119 - Posted: 14 Sep 2004, 20:39:52 UTC - in response to Message 4036.  

> Seti claims to have a Solaris copy of BOINIC:
>
> http://setiweb.ssl.berkeley.edu/download.php
>
> However, I've not had any luck downloading it. It attempts to download the .gz
> file, then fails.
>
> Rich
>
>

I am running the Seti Boinc on three sparcs at work. These are machines that get used a couple times a month. Other than they they just sit and take up electricity. I would love to see CP on Sparc also.

An older Ultra 1 should be found pretty cheap on the net. I've heard as little as $50 (us)


ID: 4119 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4194 - Posted: 16 Sep 2004, 13:26:06 UTC - in response to Message 4119.  

> An older Ultra 1 should be found pretty cheap on the net. I've heard as
> little as $50 (us)

I'm not sure that would be able to compile and run CPDN though?

ID: 4194 · Report as offensive     Reply Quote
old_user6609

Send message
Joined: 31 Aug 04
Posts: 10
Credit: 65,416
RAC: 0
Message 4412 - Posted: 21 Sep 2004, 0:22:15 UTC - in response to Message 4194.  

> > An older Ultra 1 should be found pretty cheap on the net. I've heard as
> > little as $50 (us)
>
> I'm not sure that would be able to compile and run CPDN though?
>
>
>

I don't see why not. CP seems to need disk space more than anything. It would probably take longer that on a blade... But I don't see any reason that it couldn't. We use ours for heavy graphical analysis of semi-conductor imaging.

Even though they are slower that the current P4's in some reguards. The sparc compile of the older ultra's should have no problem running on the blades. I now have seti running on 2 ultra 1's one ultra 10, 1 ultra 60, and a blade 1000. They see use here about once a month, so I can have BOINC crunching on them most of the time.

Its not a pretty graphical, just a command line, but hey, It does crunch the data for you.

Since the are workstations for my place of employment, I cannot loan them out for the compiling, or I would...
ID: 4412 · Report as offensive     Reply Quote
old_user6609

Send message
Joined: 31 Aug 04
Posts: 10
Credit: 65,416
RAC: 0
Message 4491 - Posted: 22 Sep 2004, 16:32:28 UTC - in response to Message 4412.  

Another thing I noticed is that SETI binary seems to be using gcc+, and the sun is a sponsor for them. Maybe the will run the make for you on the source.





I don't havethe full gcc it installed on the systems here, but I do have the libraries needed to run seti. If needed I have one I can load up gcc on and run a compile. Feel free to contact me by e-mail.

Since seti seems to be having a major problem with credits right now, I would dearly love to switch my sparcs to cp.
ID: 4491 · Report as offensive     Reply Quote
old_user15302

Send message
Joined: 8 Sep 04
Posts: 5
Credit: 538,121
RAC: 0
Message 4676 - Posted: 26 Sep 2004, 17:13:02 UTC

I got the Solaris Boinc running on a Blade 100 by downloading from here
http://setiweb.ssl.berkeley.edu/download.php
I then sorted out the gcc libraries and it started to run
so I entered the URL and my account id but the client responded

platform 'Sun-sparc-solaris-2.7' not found

Can someone tell me how to get this platform added please?

Dave Bamford.


ID: 4676 · Report as offensive     Reply Quote
old_user15302

Send message
Joined: 8 Sep 04
Posts: 5
Credit: 538,121
RAC: 0
Message 4680 - Posted: 26 Sep 2004, 18:52:52 UTC - in response to Message 4054.  

> we would need a Solaris box and proper C++ & Fortran compilers, which
> unfortunately we lack. At one point Sun UK was going to donate a box but I
> never heard again. We never seem to get "freebies" -- probably because this
> climate model doesn't make any computers look like "speed demons!" :-)
> There's a lot of things I'd like to do such as Solaris & AMD64 CPDN
> clients but unfortunately due to lack of resources (hardware & money for
> software i.e. compilers) it seems doubtful they'll get done.
>
ID: 4680 · Report as offensive     Reply Quote
old_user15302

Send message
Joined: 8 Sep 04
Posts: 5
Credit: 538,121
RAC: 0
Message 4681 - Posted: 26 Sep 2004, 18:55:26 UTC - in response to Message 4054.  

> we would need a Solaris box and proper C++ & Fortran compilers, which
> unfortunately we lack. At one point Sun UK was going to donate a box but I
> never heard again. We never seem to get "freebies" -- probably because this
> climate model doesn't make any computers look like "speed demons!" :-)
> There's a lot of things I'd like to do such as Solaris & AMD64 CPDN
> clients but unfortunately due to lack of resources (hardware & money for
> software i.e. compilers) it seems doubtful they'll get done.

You can download a free 30 day trial of C++ and Fortran compilers
from here.
http://www.apogee.com/download.html

If you need access to our Blade I am sure I can let you access it via the net.
let me know if you want to try it out


ID: 4681 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4712 - Posted: 27 Sep 2004, 12:10:41 UTC - in response to Message 4681.  

thanks, but I'm afraid it doesn't look promising for CPDN under Solaris, we just don't have the available staff to work on this right now. Perhaps if CPDN gets to hire some more computer people someday it would be worthwhile.
ID: 4712 · Report as offensive     Reply Quote
old_user15302

Send message
Joined: 8 Sep 04
Posts: 5
Credit: 538,121
RAC: 0
Message 4731 - Posted: 27 Sep 2004, 20:11:42 UTC - in response to Message 4712.  

> thanks, but I'm afraid it doesn't look promising for CPDN under Solaris, we
> just don't have the available staff to work on this right now. Perhaps if
> CPDN gets to hire some more computer people someday it would be worthwhile.
>

Thats a shame :-( . It sort of defeats the object of Boinc. How much work
is it buliding the CP client --with-boinc?

Dave.

ID: 4731 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4733 - Posted: 27 Sep 2004, 22:01:49 UTC - in response to Message 4731.  

> Thats a shame :-( . It sort of defeats the object of Boinc. How much work
> is it buliding the CP client --with-boinc?

well I don't think it "defeats the object of BOINC" considering nobody else has ever run this model on Windows, let alone Win, Mac, and now Linux (well they have run on Linux before, but usually clusters). It's a hell of a lot of work to get a model running on a different O/S -- it took years to get it to Windows, and from there another year to get it running under BOINC in Win/Mac/Linux. That was with Tolu & myself working 6-7 days a week for most of the year, probably 60 hours each a week at least. And with me leaving at the end of Nov I'm afraid there are other priorities right now, such as the not-so-small task of figuring out and documenting exactly what we did on the client & server over the past year to get everything working!

ID: 4733 · Report as offensive     Reply Quote
old_user15302

Send message
Joined: 8 Sep 04
Posts: 5
Credit: 538,121
RAC: 0
Message 4755 - Posted: 28 Sep 2004, 13:27:19 UTC - in response to Message 4733.  

> > Thats a shame :-( . It sort of defeats the object of Boinc. How much
> work
> > is it buliding the CP client --with-boinc?
>
> well I don't think it "defeats the object of BOINC" considering nobody else
> has ever run this model on Windows, let alone Win, Mac, and now Linux (well
> they have run on Linux before, but usually clusters). It's a hell of a lot of
> work to get a model running on a different O/S -- it took years to get it to
> Windows, and from there another year to get it running under BOINC in
> Win/Mac/Linux. That was with Tolu & myself working 6-7 days a week for
> most of the year, probably 60 hours each a week at least. And with me leaving
> at the end of Nov I'm afraid there are other priorities right now, such as the
> not-so-small task of figuring out and documenting exactly what we did on the
> client & server over the past year to get everything working!
>
The code must be a real monster to take that ammount of time to port
we really appreciate your contribution. I was under the false impression that
moving to a different OS with BOINC was much easier than native porting
as BOINC took most of the OS interfacing and comms out of the equation.

It must be a much more complex piece of code than seti@home.
I believe that CPDN is a much more important issue than seti and I am
surprised that Vendors like Sun were not interested. I suppose as time
goes on more people will sit up and listen to the Global Warming issue.
Probably too late already.

Dave.
ID: 4755 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4756 - Posted: 28 Sep 2004, 13:32:33 UTC - in response to Message 4755.  
Last modified: 28 Sep 2004, 14:26:05 UTC

It's 45MB of source code, about 1.2 million lines, so it's on the scale of making a Linux kernel with most drivers and sending it around the world to run on thousands of computers. It's orders of magnitude more complicated and in terms of computing power than anything else out there that I've seen in the DC community.

Unfortunately vendors don't seem interested in us at all; probably because the complexity of the model means that it's probably not a "glamorous" app as far as showing off their speed etc. Perhaps also they wrongly think it would "politicize" them as "global warming" (we're doing climate research and don't really jump into the whole "global warming" political debate).
ID: 4756 · Report as offensive     Reply Quote
Profile old_user559

Send message
Joined: 8 Aug 04
Posts: 1
Credit: 25,620
RAC: 0
Message 14895 - Posted: 3 Aug 2005, 15:19:44 UTC - in response to Message 4054.  

> we would need a Solaris box and proper C++ & Fortran compilers, which
> unfortunately we lack. At one point Sun UK was going to donate a box but I
> never heard again. We never seem to get "freebies" -- probably because this
> climate model doesn't make any computers look like "speed demons!" :-)
> There's a lot of things I'd like to do such as Solaris & AMD64 CPDN
> clients but unfortunately due to lack of resources (hardware & money for
> software i.e. compilers) it seems doubtful they'll get done.
>

If you ever plan to make the Solaris port, I would be willing to help you in order to test it over more than a dozen heterogeneous SPARC Machines (UltraSparcs, Enterprise Servers, E1000Ks, SunFires 6800 or even SunFires 25K) as well as AMD Opterons on 64 bits on Solaris x86.
<a href="http://Sun-Microsystems.ORG">http://Sun-Microsystems.ORG</a>
ID: 14895 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 14896 - Posted: 3 Aug 2005, 15:30:54 UTC - in response to Message 14895.  

thanks for the offer, but it's pretty unlikely. we may even have to drop the Mac OS X version due to lack of staff &amp; hardware for development.
ID: 14896 · Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : Solaris Boinc

©2024 cpdn.org