Thanks Gecko!

Let me quote a part of the email I got back from Eric Korpela after asking him about specifics on optimizing and distributing S@H.

My question:
> In addition, I'd like to ask how you feel about someone paying for an
> educational-type license for ICC and IPP and offering X86 builds (and maybe OS X, as
> they now support it) and using this compiler and library package to release the
> official BOINC-distributed Apps. This License would be made out to the Seti project,
> of course. Build scripts to automate this already exist.

Eric's response:
That's a good idea, but I'm not sure whether SETI@home falls into the
academic use license terms. Before spending any money the buyers
should check the license terms.

So that should explain where I personally would like this to go. Level playing field, anyone? Preaching to the choir, I know...

In related matters, I'm trying to find contacts at Intel - since they make the compiler and tools we're using to optimize, it seems logical to ask them whether they would like to sponsor the project
(or rather, provide us with the necessary licenses). Hardware would be neat too, but that's probably a pipe dream :o)

If anyone has contacts at la Intella, please don't hesitate to pitch this idea to them (and/or get in touch with me).

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

---CUT---
..............In related matters, I'm trying to find contacts at Intel - since they make the compiler and tools we're using to optimize, it seems logical to ask them whether they would like to sponsor the project
(or rather, provide us with the necessary licenses). Hardware would be neat too, but that's probably a pipe dream :o)

If anyone has contacts at la Intella, please don't hesitate to pitch this idea to them (and/or get in touch with me).

Regards,
Simon.

Hello!

Maybe when yes, they would like to have some place on S@H for some commercials... :-)


Greetings!

:-)

---

"Little blue men searching for little green men that look like little grey men"?

:o)

You do remember the "Blue Man Group" commercials Intel used to do, right...hilarious stuff.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

"Little blue men searching for little green men that look like little grey men"?

:o)

You do remember the "Blue Man Group" commercials Intel used to do, right...hilarious stuff.

Hello!

Hey I remember, of course! :-)

Really funny boys... :-)


Greetings!

:-)


Maybe they will produce some new specially commercial for us... S@H...! :-)

---

Thanks Gecko!

Let me quote a part of the email I got back from Eric Korpela after asking him about specifics on optimizing and distributing S@H.

My question:
> In addition, I'd like to ask how you feel about someone paying for an
> educational-type license for ICC and IPP and offering X86 builds (and maybe OS X, as
> they now support it) and using this compiler and library package to release the
> official BOINC-distributed Apps. This License would be made out to the Seti project,
> of course. Build scripts to automate this already exist.

Eric's response:
That's a good idea, but I'm not sure whether SETI@home falls into the
academic use license terms. Before spending any money the buyers
should check the license terms.

So that should explain where I personally would like this to go. Level playing field, anyone? Preaching to the choir, I know...

In related matters, I'm trying to find contacts at Intel - since they make the compiler and tools we're using to optimize, it seems logical to ask them whether they would like to sponsor the project
(or rather, provide us with the necessary licenses). Hardware would be neat too, but that's probably a pipe dream :o)

If anyone has contacts at la Intella, please don't hesitate to pitch this idea to them (and/or get in touch with me).

Regards,
Simon.

It is my opinion (and IANAL) that virtually all of the licenses for software like this say that it can be installed and used on a single computer.

Now, if someone loaded up the compiler, and then somehow made a box that would accept jobs and run the compiler and return the results (errors or binaries) then it is my opinion that the Intel license would not be violated.

.... so if we could find someone with ICC and IPP and etc. who'd be willing to make it available by some means, that'd work too.

-- Ned

---

[url]http://setiathome.berkeley.edu/workunit.php?wuid=82161613[url]

Just one funny bit it is claiming less than 10 (maiby this is coz i replace the worker program at 90%)

will see what hapen to the next one it will be workunit id 82214369

my first program i compiled was icc and no optimation options jyst sse in win-config.h
compiled lighting fast compare to with full optimation.

to disable the graphics we must // define BOINC_APP_GRAPHICS only problem i could not find the spot where it is defined for win32
so what i did renamed all instances of it to "disBOINC_APP_GRAPHICS" and that did the trick not sure mayby problem on source code


Conrad

---

In related matters, I'm trying to find contacts at Intel - since they make the compiler and tools we're using to optimize, it seems logical to ask them whether they would like to sponsor the project
(or rather, provide us with the necessary licenses). Hardware would be neat too, but that's probably a pipe dream :o)

If anyone has contacts at la Intella, please don't hesitate to pitch this idea to them (and/or get in touch with me).

I remember this post a few weeks ago in the AMD vs. Intel thread. He made his point and employer well known.

http://setiathome.berkeley.edu/forum_thread.php?id=31183#320990

Team: http://setiathome.berkeley.edu/team_display.php?teamid=26483

Maybe worth a friendly "hello" since he ( & they) denote their hard-earned CPU cycles for the same cause?

I think my reading assignment tonight is:
"Intel license provisions & exclusions."

---

Simon, you seem to have done the hard work of creating optimized apps for x86 Linux and x86 WinXP that are actually better than Crunch3r's were. I salute you.

Alex Kan seems to have done the same thing for G4/5 MacOS X.

Now, the Big Question.
How do we get those apps blessed by S@H and merged into the official CVS tree so that users can download "official" platform specific optimized apps?

The other value I see out of getting all this work into thte official CVS tree is cross-polination. It seems plausible that ideas used to make these optimized apps could very well led to advances in the algorithms / data structures / coding techniques / compile options / etc used for all of S@H and perhaps even fertilize BOINC at bit.

---

[url]http://setiathome.berkeley.edu/workunit.php?wuid=82161613[url]

Just one funny bit it is claiming less than 10 (maiby this is coz i replace the worker program at 90%)

will see what hapen to the next one it will be workunit id 82214369

my first program i compiled was icc and no optimation options jyst sse in win-config.h
compiled lighting fast compare to with full optimation.

to disable the graphics we must // define BOINC_APP_GRAPHICS only problem i could not find the spot where it is defined for win32
so what i did renamed all instances of it to "disBOINC_APP_GRAPHICS" and that did the trick not sure mayby problem on source code


Conrad

I can tell you where to set that :o)
It's in "Preprocessor Options". The first line has stuff in all caps; if you put "XXX" there, it's the same as if you do a "#define USE_XXX" in win_config.h. The client has this (BOINC_APP_GRAPHICS) set in the preprocessor options. Delete that word, you're done.

Also, the stuff you commented out to make it compile - there's a way to fix the "can't call malloc from a catch block" error. Since it's an error message it should not be removed. My sources compile fine with the fix here -
http://setiweb.ssl.berkeley.edu/beta/forum_thread.php?id=527#5419.

I'll be uploading a copy of my sources for Linux and Windows shortly.

The Windows one now compiles with the default MS compiler as well as ICC.

Regards,
Simon.

P.S: Yes, doing an optimized build takes *forever* compared to a normal one. It does however also give a *lot* of performance gain.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

i did un rem that maloc error found the fix to it futher down this thread thought thread was complete but was too lone so some posts was hidden

i still strugling to compile boinc itself with intel compiler but gave up for now

Conrad

---

Simon, you seem to have done the hard work of creating optimized apps for x86 Linux and x86 WinXP that are actually better than Crunch3r's were. I salute you.

Alex Kan seems to have done the same thing for G4/5 MacOS X.

Now, the Big Question.
How do we get those apps blessed by S@H and merged into the official CVS tree so that users can download "official" platform specific optimized apps?

The other value I see out of getting all this work into thte official CVS tree is cross-polination. It seems plausible that ideas used to make these optimized apps could very well led to advances in the algorithms / data structures / coding techniques / compile options / etc used for all of S@H and perhaps even fertilize BOINC at bit.

Well, thanks.

However, my Windows builds aren't at the speed of my Linux ones (yet, I'd hope) so don't give me credit where it's not due.

As to this getting merged into the official sources - you might not be aware of this, but all I changed were small errors that were keeping it from successfully compiling.
I did not change the source in any significant way. I am not even capable of that, in fact. I don't know C/C++, but I am a programmer, so basic syntax and logic constructs apply (and I can do basic "proofreading" of source).

So no, it doesn't need to be merged. ALL, and I repeat, ALL performance gains come SOLELY from using a custom compiler and math library package, namely Intel's ICC/IPP/MKL. This only applies to Linux and Windows, not Alex Kan's excellent optimized client for OSX - AFAIK, he hand-optimized the assembler code using a profiling tool.

I'm trying to somehow orchestrate getting the Seti@Home project a license for said compiler and libraries so the default clients that *everyone* gets via BOINC are already optimized ones for their specific computer.
The BOINC client can figure out the host's capabilities, in fact this has just been added to the code if I'm informed correctly.
So when it downloads a client, why not have it download the one that best fits your host?
That's my ultimate goal - making it available not to a select few, but to everyone.

I believe it's in the spirit of DC and any community-driven project to do it this way.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

"KSWN" wrote:

ALL, and I repeat, ALL performance gains come SOLELY from using a custom compiler and math library package, namely Intel's ICC/IPP/MKL. This only applies to Linux and Windows, not Alex Kan's excellent optimized client for OSX - AFAIK, he hand-optimized the assembler code using a profiling tool.

That's actually good news because it implies that there may be some more performance to be had by using Alex's PPC mods as inspiration and attacking the x86 code with VTune.

I'm trying to somehow orchestrate getting the Seti@Home project a license for said compiler and libraries so the default clients that *everyone* gets via BOINC are already optimized ones for their specific computer.

How can I help? I may know some folk within Intel who can be useful.

The BOINC client can figure out the host's capabilities, in fact this has just been added to the code if I'm informed correctly.

Yes. see comments by Rom Walton in his blog and elsewhere.

Side Note:
You may have problems getting the Win apps to perform as well as the Linux apps.
Among other things, Win threads should be called "ropes" they are so heavyweight, and I do not know if the s@hE app or boincmgr Win ports can or do use what M$ calls "fibers" appropriately.
There are other issues as well with things like the memory management system.
How big is the performance difference?

---

Well,

I pasted a passage from an email to/from Eric Korpela a few posts ago. The ideal thing would be to get a license for ICC, IPP and MKL for all supported platforms - Windows, Linux and OS X.
Yes, OS X too since new Macs are just PCs with a nice shell :o)

The problem is, those software packages do cost quite a bit of money. We could maybe organize a fund drive to get all the needed cash, but getting it direct from the source would open up new vistas of possibility.

The problem with the non-commercial versions is their license. It says for non-commercial PERSONAL use.
I don't read legalese so well, and I don't know whether that means you cannot distribute what you compile with it, but I would assume it might be a problem, and therefore not something the project staff would likely do.

Also, it might alert people at Intel to the project (more than Team Intel, they have 20 people or so), and it might net the project some hardware and/or publicity.

It should be stressed that the project is non-commercial in nature and is donation-driven to a large part - also, that Intel CPUs generally do better at Seti@Home than AMD ones (hey, it's true ;)).
Also, it can't hurt mentioning how many people are involved in Seti@Home...it's not like it's just hundreds or thousands...hundreds of thousands of active crunchers -
not to mention the MILLIONS of people who already once participated and could again.
Big target audience.

So I sincerely hope you can help in this matter.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

The problem with the non-commercial versions is their license. It says for non-commercial PERSONAL use.
I don't read legalese so well, and I don't know whether that means you cannot distribute what you compile with it, but I would assume it might be a problem, and therefore not something the project staff would likely do.

I also came accoss several references related to Academic licenses and prohibitions from receiving ANY kind of funding, grants or compensation related to the use of the software, even if used non-commercially. Also, must be single license holder, multiple hosts OK as long as no more than one host is operational at any given time. I don't believe Academic will work, but this is a question better suited for copywrite attnys. Anyone w/ legal connections?

---

You folks really don't have a clue, and I really think that anything you might try to release would be questional in quality at best.

But again, that doesn't surprise me, as Seti/Boinc has no quality control over their code.

I'd not be surprised if the new cruncher detected an intelligence in Wash DC, but flagged it as an ET.

"But, but, but, all I did was comment out errors I didn't understand!"

---

So umm... "questional" isn't even a word. But aside from that, what is being discussed in the last few posts is somehow procuring a license for Berkeley to use. Then no one but Berkeley would be releasing anything. Keep up with the thread woody!

And we have covered this source control issue before as well. Berkeley has full control over their source. They have also made it available to the public via the GPL with full knowledge that people exactly like this would do exactly what is happening here and yet they officially support development of optimized clients! Straight from the top. Deal with it.

If you don't like the way the project is being run then leave.

---

A member of The Knights Who Say NI!
For rankings, history graphs and more, check out:
My BOINC stats site

Everyone's entitled to your opinion, huh azwoody :o)

Why don't you make some sense instead of making unprovable allegations? (who said I don't understand what I'm doing? Oh yeah...you.)

Questional? You must mean questionable. Just like your motives...

No quality control? Are you aware of how CVS works?

And, no, Washington is not currently beset with an overabundance of reason and intelligence. Who knows, maybe George Bush is really an alien...seriously though, you need to relax.

Quotes you *should* be using...

"but but but, I did not read the thread Simon just created and don't know what I'm saying"
"but but but, I never look closely at what I'm criticising, but make up for it in volume"
"but but but, I have the time to rip on people but not to check my facts"

mmkay?

So go read up and download the sources.
Do a diff. Tell me what you divine.

Until then, you might as well consider yourself ignored. Check the facts - then flame, if you must - although that does violate forum rules.

Regards,
Simon.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

"But, but, but, all I did was comment out errors I didn't understand!"

If you took five minutes to look at the piece of code in question, you'll realize that a coding error was commented out, not an error message.

Regards Hans

P.S: Yes, I reported your post, now buzz off!

---

Hi,
I just checked out the download statistics for the private test packages that I made available to a very few people.

The terms of gaining access to this very first testing and benchmark build were communicated very clearly to everyone who received a binary (or link to one).
They stated, in no uncertain terms, the strict prohibition of distribution or otherwise making accessible the content of the test package to ANYONE but themselves.
As the packages did not include source code, there was clearly no legal way to do it in the first place, but I felt I had to make sure anyway.

Well, honour system is nice as long as it works :o)

So, baracutio@gmx.de - why did you share the link with someone else? I'm sure you didn't have to download the package 74 times all for yourself and the ONE host you had my authorization to test it on -
OFFLINE using my included test-scripts.
You said you didn't even have Linux yourself but shared it with someone else anyway.

We are not amused.

Saenger posted a link to a german SETI/BOINC forum in the "new optimized apps" read-only thread that was at first offering Crunch3rs clients.
You were warned by others and myself that you were doing something illegal -
the admin of that forum (or maybe just with access to its downloads) is you, baracutio, isn't it?

Why do I know it was you and noone else? Simple - everyone received a different download package.
No URL was the same - I may mean well, but you can't seriously believe I'm that naive after what happened recently.
Also, you received the link later than others, and the date of the downloads jibes with that.

And now for the worst part, which I make you, baracutio, fully responsible for -
since this was my very first test package that I sent out to a total of 3 people, I had no comparable results from someone else yet.
It was only after Hans Dorn notified me that the results did not validate that I found an overaggressive compiler flag and removed it - this fixed the results not validating.

Those binaries produce invalid results. They were not meant to be shared, they were meant to be TESTED on your platform.

That's why I expressly forbid you to use them to crunch for credits. They were simply not tested enough for that.

Anyone who received the link from baracutio or from elsewhere and is using the client to crunch online, you will see it gets you nowhere.
You're hurting other people who return valid results. Cease and desist, or follow my how-to guide to make your own.

Thank you for violating the GPL!

May the FSF haunt your dreams.

Regards,
Simon.

P.S.: You can see my code now generates valid results (see host links in previous posts) -
somehow it's a good thing my name would not appear in results that someone might try to crunch on BOINC using the package you got (I had simply not gotten to inserting it at that stage).

P.P.S: Be sure that this in no way affects my dedication to this project or to what I'm doing - my skin is thicker than that.

---

Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information

Hi Simon,

what Microsoft Compiler Version did you use for testing ?
Visual Studio 2003, or 2005 ?