... shortened ...

I think this flexible interpretation of the law is one of the causes that being a lawyer is such a wealthy profession in some countries...

As my mother said:

Du kriegst nicht Recht, du kriegst ein Urteil.

You don't get justice, you get a judgement.

---

Aloha, Uli

... shortened ...

I think this flexible interpretation of the law is one of the causes that being a lawyer is such a wealthy profession in some countries...

As my mother said:

Du kriegst nicht Recht, du kriegst ein Urteil.

You don't get justice, you get a judgement.

I think she may be right! But do tell me where the flexible bit is?

---

I am really glad this matter is sololved to everyones satisfaction.

May I make one suggestion, that we review the naming convertion of the optimized clients?

As someone mentioned before, we may have some confusion as and when Berkeley release a 4.11 SETI client. We are not really following Berkeleys naming lead.
The official windows release name is in the form "sciencename_ver_opsys_processor.exe" (e.g. setiathome_4.09_Windows_Intelx86.Exe")

The only bit that the optimized apps really change is the "_Intelx86" bit, which indicates generic x86 processors.

To me, it would make so much more sense if this was the only bit of the name that we vary. e.g. "setiathome_4.09_windows_P4sse3.Exe" or "setiathome_4.09_windows_P3vanilla.Exe", "setiathome_4.09_windows_AMD64.Exe" etc

When looking at what version is installed on a given system it is so much easier to see what version is actually running on that CPU.
I've already done this to some of mine, and it only required an edit of the app_info.xml to allow the new name.

---

I think she may be right! But do tell me where the flexible bit is?

I try:

1) and 2) are ok for me.

3) You stated you got the files via email. Ok, then in my point of view it's not Tetsuji's fault but the fault of your 'friend', sending them to you without mentioning it's limited use time.

4) I thought mmciastro referred to Tetsuji as being a friend and to support him instead of accusing him.

As you see, it depends from where you look at the situation.

---

Aloha, Uli

Your arguement for those that "unwittingly used the object code" might be better supported if you weren't in here making the arguement that you didn't know you weren't supposed to continue using the code.

I'm just saying.

LoL.

I think she may be right! But do tell me where the flexible bit is?

I try:

1) and 2) are ok for me.

3) You stated you got the files via email. Ok, then in my point of view it's not Tetsuji's fault but the fault of your 'friend', sending them to you without mentioning it's limited use time.

4) I thought mmciastro referred to Tetsuji as being a friend and to support him instead of accusing him.

As you see, it depends from where you look at the situation.

Sure OK. Well I dont know where he got the zips from or indeed if they are the Tetsuji ones anyway. They may be someones else's...she uses source forge a lot and may be they were legal.

I was never commenting on any situation I may have. I was looking at the legalities of this and trying to reassure people in possession of these clients that actaully I did not think they were exposed as they ARE NOT AND CANNOT BE part of the original licence agreement.

---

To liven up an old thread, I am now builing SETI (hopefully) and have corrected errors down to the 1 error that is now nagging me!

The error is:

c:\programmer\boinc\seti_boinc\client\seti_header.cpp(111) : error C2664: 'std::_Vector_const_iterator::_Vector_const_iterator(const std::_Vector_const_iterator &)' : cannot convert parameter 1 from 'int' to 'const std::_Vector_const_iterator &'
        with
        [
            _Ty=coordinate_t,
            _Alloc=std::allocator
        ]
        Reason: cannot convert from 'int' to 'const std::_Vector_const_iterator'
        with
        [
            _Ty=coordinate_t,
            _Alloc=std::allocator
        ]
        No constructor could take the source type, or constructor overload resolution was ambiguous

Any ideas?

---

I really wish i could help you with this. I know there are some here who could tell how to correct this error.

Please help with the open source solution to a faster client!!

---

98SE XP2500+ @ 2.1 GHz Boinc v5.8.8

And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer.

My two cents on SETI@home with open source programs....if someone will continue this.

I suspect fftw should be the bottleneck. VC++6,7,7.1 produce wrong codes for fftw. See this fftw document.
I've also tried w/o Intel compiler, but the seti client failed in calculation. It stops in a few minutes with wrong results. So I'm not sure VC++2005 will do it correctly. That's why I stuck to Intel C++ compiler 8.1.

If cygwin compiled S@h, it would be fine, and if gcc-4.0 were on cygwin (maybe it's easy) it would be much better...but I won't try it. Programming/compiling on Windows is often a real pain in the ass.

My father's will is "Windows is for games. Development should be done on unix-like systems."

EDIT: The reason I made optimized clients for Windows is my wife cannot do her work without Windows....that's the only reason. I once thought I could have run an optimized Linux client in coLinux on Windows, but it should slow down the crunching speed.

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

Another good reason to do one for Windows is that all "normal"
people use Windows.
All of my friends have Windows, actually nobody I know in RL have
anything else than windows. *lol*

Unix/Linux is for work computers and for nerds only *lol*

yes, I admit it. I may be one of the nerds....lol. There are lots of nice applications for Windows while Linux applications are being developed; only some of them are available. Windows is the de facto standard. Unfortunately it's the truth. I don't want a flame war on which OS is better, Window$ or Linux. But *personally*, I prefer Linux for development, though I usually use Windows for games.

And one truth is....compiling fftw library (which boosts optimized Windows clients) on Windows is currently restricted to Intel compiler (VC++ can't), while on Linux both gcc and Intel can and that both are free. Free develop environment on Windows has now only limited capabilities...cygwin/dev-x or any others.

Unfortunately as I wrote before Intel compiler for Windows is $399, and I could buy it thanks to Mr.X...

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

Wow, I just tried out Prognatus version seti-p3.zip
on my AMD64 3000+ and it cut down the WU time by over a whole hour. From about 2.45 per WU to 1.38.
Could this really be legit ?

It feels completely incredible.

Yes, it's totally legit. Basically we use optimized Fast Fourier Transform (FFT) routines in place of the standard routines used by the Berkeley client. So the same maths is performed, just that it's performed by more highly optimized code so it's considerably faster.

In addition, the optimized clients are specific for particular CPU types and can take advantage of things like SSE that have been introduced since the introduction of the original Pentium Pro (i686) CPU. Processors have developed quite a bit since then and the original berkeley client didn't use any of these additional capabilities.

Ned

---

*** My Guide to Compiling Optimised BOINC and SETI Clients ***
*** Download Optimised BOINC and SETI Clients for Linux Here ***

For the Athlon XP you should use seti-p3.zip
cause it doesn't have SSE2 instructions,
only SSE1.
For other processors get a cpu identification
utility like WCPUID or CPU-Z. Both are free.
That programs will tell you, what instructions
are understood by your cpu.

Also have look over there:
'Legal' optimized clients

---

Aloha, Uli

How to compile with VS6 for AMD64 platform..

They "AMD" have a little guide here!!

http://www.devx.com/amd/Article/21313

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

So far I've just downloaded the boinc-445-sse2.zip file from this http://www.marisan.nl/seti/ web site and installed it as per instructions.
My benchmarks have gone from

Measured floating point speed 1566.95 million ops/sec
Measured integer speed 1827.77 million ops/sec
to
Measured floating point speed 1899.18 million ops/sec
Measured integer speed 2711.9 million ops/sec

Comp specs are 3.2GHzP4 running oc'd to 3.67GHz with 1GB PC3200 RAM on an ASUS P4P800-E-Deluxe mobo runing XPSP2.

Overall this has dropped the estimated completion times on Einstein from 6.5hrs to 5hr50min, which in reality will cause me problems since the
actual wu completion time will not change (since the Einstein code has not been optimized). Looks like it's time to drop Einstein and reattach to
SETI on this machine.

Live long and crunch!

---

Paul
(S@H1 8888)
And proud of it!

Just did the same on my laptop P4-Mobile (not centrino) (only 512kb L2 cache @2.2GHz)

Benchmarks have gone from
Measured floating point speed 1130.82 million ops/sec
Measured integer speed 2353.14 million ops/sec
to
Measured floating point speed 1500.09 million ops/sec
Measured integer speed 3155.49 million ops/sec

Nice.

Also did the app_info.xml and I downloaded the file seti-p4-no-sse3.zip (1,13 MB) for P4 w/o SSE3 (also P4M w/o SSE3) and followed the instructions. This
file extracts as 4.11 whereas BOINC is still using 4.09 on the seti wu's. Do I need to do something else here?

Paul.

Also did the app_info.xml and I downloaded the file seti-p4-no-sse3.zip (1,13 MB) for P4 w/o SSE3 (also P4M w/o SSE3) and followed the instructions. This
file extracts as 4.11 whereas BOINC is still using 4.09 on the seti wu's. Do I need to do something else here?

Depending on the app_info.xml you used, it will most likely work through all the 4.09 WUs you already had on your system, then any newly downloaded work will use 4.11.

If this isn't happening, you did put the app_info.xml in the S@H project directory didn't you?

OK I finally tried the optimized seti & boinc. I didn't wait and empty all my 4.09 wu. just copy the necessary files and restarted boinc.
The Boinc manager says wu with 4.09, but the task manager says 4.11. the new wu that were downloaded says 4.11.
FYI

---

Fred
BOINC Alpha, BOINC Beta, LHC Alpha, Einstein Alpha

Also did the app_info.xml and I downloaded the file seti-p4-no-sse3.zip (1,13 MB) for P4 w/o SSE3 (also P4M w/o SSE3) and followed the instructions. This
file extracts as 4.11 whereas BOINC is still using 4.09 on the seti wu's. Do I need to do something else here?

Depending on the app_info.xml you used, it will most likely work through all the 4.09 WUs you already had on your system, then any newly downloaded work will use 4.11.

If this isn't happening, you did put the app_info.xml in the S@H project directory didn't you?

Yep...and just attached my P4 and it downloaded SETI 4.09...hmmmm....interesting

Paul.

check out what is actually running in task manager.

When I first put on the optimised code, I saw the same thing you do - SETI 4.09 appeared to be still running.

HOwever, upon checking task manager, I saw that it was actuall 4.11 exe that was doing teh crunching. Maybe something similar happening with you?

---

Wanna visit BOINC Synergy? Click my stats!

Join BOINC Synergy Team

Yeah...saw that. The wu's that were initially downloaded (pre-upgrade) were 4.09 and so the xml files showed them as that in the work tab. New wu's
downloaded after the upgrade are 4.11. The drop in crunch time is absolutely amazing, which is partially offset by the increase in benchmarks so overall my
daily claimed credit is going to increase by upwards of 30%!

And it's all legal!

Live long and crunch.

---

Paul
(S@H1 8888)
And proud of it!