setiathome is compiled using visual C++

Message boards : Number crunching : setiathome is compiled using visual C++
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Peter Foelsche

Send message
Joined: 3 Dec 99
Posts: 4
Credit: 82,719
RAC: 0
United States
Message 169702 - Posted: 19 Sep 2005, 19:20:50 UTC

I've seen considerable performance boosts when using the intel compiler -- especially for numerical applications. Why don't you try this compiler?
ID: 169702 · Report as offensive
Ulrich Metzner
Volunteer tester
Avatar

Send message
Joined: 3 Jul 02
Posts: 1256
Credit: 13,565,513
RAC: 13
Germany
Message 169707 - Posted: 19 Sep 2005, 19:32:10 UTC

Have you ever read the optimized seti cruncher threads? ;)
Aloha, Uli

ID: 169707 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 169732 - Posted: 19 Sep 2005, 20:32:46 UTC - in response to Message 169707.  

Have you ever read the optimized seti cruncher threads? ;)

Obviously he did not ;)

But as i´ve read the thread name a question came to my mind.
Is it possible to "convert" the c++ parts of the code to C code ?
(maybe a stupid question i´m not a programer)



Join BOINC United now!
ID: 169732 · Report as offensive
Profile [B@H] Ray
Volunteer tester
Avatar

Send message
Joined: 1 Sep 00
Posts: 485
Credit: 45,275
RAC: 0
United States
Message 169733 - Posted: 19 Sep 2005, 20:43:01 UTC

For faster clients compiled with C or C++ visit The Optimized Application info center at BOINC Synergy. BOINC Synergy is a great team to join if you are not in one new.


Pizza@Home Rays Place Rays place Forums
ID: 169733 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 169748 - Posted: 19 Sep 2005, 21:23:08 UTC - in response to Message 169707.  

Have you ever read the optimized seti cruncher threads? ;)

Which threads do you mean ?

ID: 169748 · Report as offensive
Ulrich Metzner
Volunteer tester
Avatar

Send message
Joined: 3 Jul 02
Posts: 1256
Credit: 13,565,513
RAC: 13
Germany
Message 169751 - Posted: 19 Sep 2005, 21:27:22 UTC - in response to Message 169748.  
Last modified: 19 Sep 2005, 21:28:28 UTC

Which threads do you mean ?

Ever cared using the search function:
Thread titles containing "optimized"
Aloha, Uli

ID: 169751 · Report as offensive
Profile Peter Foelsche

Send message
Joined: 3 Dec 99
Posts: 4
Credit: 82,719
RAC: 0
United States
Message 169756 - Posted: 19 Sep 2005, 21:33:18 UTC - in response to Message 169732.  

Have you ever read the optimized seti cruncher threads? ;)

Obviously he did not ;)

But as i´ve read the thread name a question came to my mind.
Is it possible to "convert" the c++ parts of the code to C code ?
(maybe a stupid question i´m not a programer)



C++ code runs as fast as c-code if they both implement the same functionality the same way.
C++ can translate C code.
So if you want performance in C++ you may want to consider which features of C++ you're using and whether they are necessary.
Using C you would have to implement these features by hand and they would rather be slower than if they are implemented by the C++ compiler.
C++ code maybe faster as it is heavily using inlining of code,
a technique not available to C programmers.
ID: 169756 · Report as offensive
Profile Peter Foelsche

Send message
Joined: 3 Dec 99
Posts: 4
Credit: 82,719
RAC: 0
United States
Message 169757 - Posted: 19 Sep 2005, 21:39:51 UTC

I'm still curious why setiathome is distributing a Visual C++ application if they already tried the intel compiler and saw the performance gain available to them.
Or in other words: Why is the boinc manager not downloading an application matching the CPU it is running on? So he could optionally exploit sse3 or sse2 which yield quite a performance boost.
ID: 169757 · Report as offensive
Doom.Rainer

Send message
Joined: 3 Apr 99
Posts: 2
Credit: 104,167
RAC: 0
United States
Message 169759 - Posted: 19 Sep 2005, 21:54:41 UTC - in response to Message 169757.  

I'm still curious why setiathome is distributing a Visual C++ application if they already tried the intel compiler and saw the performance gain available to them.
Or in other words: Why is the boinc manager not downloading an application matching the CPU it is running on? So he could optionally exploit sse3 or sse2 which yield quite a performance boost.



Well if I rember correctly, for awhile they wanted to slow everyone down simply because there connection was getting hammered with results. instead of 12 connections a day, just one connection say for a 24hour wu would benifit them greatly. I'm not justifying slowing us down though, I think the units should be larger, but optimized as per cpu (SSE MMX SSE2 or SSE3 3dnow etc...)
ID: 169759 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 169764 - Posted: 19 Sep 2005, 22:09:42 UTC - in response to Message 169757.  

I'm still curious why setiathome is distributing a Visual C++ application if they already tried the intel compiler and saw the performance gain available to them.
Or in other words: Why is the boinc manager not downloading an application matching the CPU it is running on? So he could optionally exploit sse3 or sse2 which yield quite a performance boost.


The guys @ berkeley did not use the intel compiler! Volunteer developers as TMR,Ned slider and may others have build the clients available on the third party download page and optimized them.




Join BOINC United now!
ID: 169764 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 169794 - Posted: 19 Sep 2005, 23:34:01 UTC - in response to Message 169757.  

I'm still curious why setiathome is distributing a Visual C++ application if they already tried the intel compiler and saw the performance gain available to them.
Or in other words: Why is the boinc manager not downloading an application matching the CPU it is running on? So he could optionally exploit sse3 or sse2 which yield quite a performance boost.

The clients on the third party site are not offical supported by berkeley, and is to use on your own risk.
From Berkeley are only Mac, Linux, Windows and Solaris supported, so this are the only one you can direct download.
ID: 169794 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 169795 - Posted: 19 Sep 2005, 23:37:32 UTC

I believe that one of the problems with the Intel compiler is that it does not treat AMD CPUs fairly with the standard set of optimizations that work well for Intel CPUs. I seem to recall some outrage recently someplace when this was discovered.


BOINC WIKI
ID: 169795 · Report as offensive
Profile FZB

Send message
Joined: 16 Nov 01
Posts: 11
Credit: 1,943,154
RAC: 7
Germany
Message 170597 - Posted: 22 Sep 2005, 6:12:46 UTC - in response to Message 169757.  

I'm still curious why setiathome is distributing a Visual C++ application if they already tried the intel compiler and saw the performance gain available to them.
Or in other words: Why is the boinc manager not downloading an application matching the CPU it is running on? So he could optionally exploit sse3 or sse2 which yield quite a performance boost.


may be as simple as maintenance time needed. rather work on one version than maintenance a large number of apps.
ID: 170597 · Report as offensive
Tetsuji Maverick Rai
Volunteer tester
Avatar

Send message
Joined: 25 Apr 99
Posts: 518
Credit: 90,863
RAC: 0
Japan
Message 170656 - Posted: 22 Sep 2005, 13:36:25 UTC
Last modified: 22 Sep 2005, 14:24:38 UTC

A few days ago I happened to recommend Intel C++ compiler for Linux to Eric Korpela, currently the developer of setiathome_enhanced by email for testing and debugging purposes (but not for official clients, because of its preference for Intel processors, and much more because of problems with SSE/SSE2/SSE3 detection.) Actually it's very fast, but I'm not interested in optimization so much now.

BTW Intel "was" a sponsor of SETI@home a while ago (but not now.) If you invoke seti@home client in a commandline with -version option, you'll see (this is from upcoming setiathome_enhanced 4.06 for Linux)

SETI@home client.
Version: 4.06

The project is also sponsored by the Planetary Society,
the University of California, Sun Microsystems, Paramount Pictures,
Fujifilm Computer Products, Informix, Engineering Design Team Inc,
The Santa Cruz Operation (SCO), Intel, Quantum Corporation,
and the SETI Institute.

SETI@home was developed by David Gedye (Founder),
David Anderson (Director), Dan Werthimer (Chief Scientist),
Hiram Clawson, Jeff Cobb, Charlie Fenton,
Eric Heien, Eric Korpela, Matt Lebofsky,
Tetsuji 'Maverick' Rai and Rom Walton

....not a joke :)
Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.
ID: 170656 · Report as offensive
Tetsuji Maverick Rai
Volunteer tester
Avatar

Send message
Joined: 25 Apr 99
Posts: 518
Credit: 90,863
RAC: 0
Japan
Message 170699 - Posted: 22 Sep 2005, 16:27:42 UTC - in response to Message 170656.  
Last modified: 22 Sep 2005, 16:42:38 UTC

Actually it's very fast, but I'm not interested in optimization so much now.


PS: Personally I like Intel C++ compiler and am still using it to build optimized test versions for beta project for testing and debugging personally. Its optimization capability is amazing.

For this SETI@home, I hope unofficially released clients are very nice enough (don't mind the version number 4.11(Windows) or 4.07(Linux), because math inside is the same as that of 4.18 except that optimized clients don't have graphics.)

As others wrote, I guess Berkeley won't use Intel C++ compiler for official clients. I'm not sure.
Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.
ID: 170699 · Report as offensive

Message boards : Number crunching : setiathome is compiled using visual C++


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.