I have two more optimized client versions now.
Thanks to the user who provided these for me. :)

seti-p3-sse2.zip
Pentium M owners should use this one.

seti-pm.zip
Pentium 4 M owners should use this one.

Graham, and others... if you could snag these last two files as well I'd appreciate it. I expect the university bandwidth police to be knocking on my door any day now.
Dig

I have put them online on www.marisan.nl/seti/...

Luckily no bandwidth police here.. :)

Greetings,
Speedy67

---

I've put them up at http://www.esea.dk/esea/optimized.asp

---

Way Cool indeed Dig. Now let's all hope that someone more talented than I can compile one without a Cinderella license or we'll all turn into a pumpkin in a little over two weeks. If not, it was fun (and productive!) while it lasted.
-Regards

Well, i asked about using the MinGW for compilation, thereby also being able to produce optimized clients for Athlons earlier in this thread, but noone answered.

It would be great to see if FFTW and this MinGW compiler could produce some even more efficient AMD Athlon application versions. Is anyone else out there with the expertise willing to give it a shot? I'm simply don't have the programming skills to pull it off myself.

---

It would be great to see if FFTW and this MinGW compiler could produce some even more efficient AMD Athlon application versions. Is anyone else out there with the expertise willing to give it a shot? I'm simply don't have the programming skills to pull it off myself.

Tomorrow @ work i'll download the complete MinGW collection and the Eclipse development environment including the C/C++ developer tools. In the evening i'll install the whole package onto my private computer and will try to compile the whole package. On friday my main problem was, to get a decent makefile for the components. I presume i could use the makefiles for Linux/Unix and modify them for use with MinGW, cause the MinGW environment resembles a Unix environment on windows.

Any help is truely appreciated :)

---

Aloha, Uli

Well I've been running the optimized clients over the weekend, and they really do speed things up like crazy. As long as the completed units are still providing seti with the information they need, then I am happy to recommend them.

My Athlon 2200 without SSE2 (running at 1900Mhz) was crunching a unit every 200 minutes, now it's crunching a unit in 140 minutes. CPU temperatures remained the same. (Using Seti-P3.zip)

And my Intel Dothan laptop (1700Mhz, SSE2) was crunching a unit every 140 minutes, now it's down to 80 minutes per. (Using seti-p3-sse2.zip)

In each case the optimised client reduced the run time by about an hour. Hurrah!

---

seti-p4-sse3.zip

seti-p4-no-sse3.zip

seti-p3.zip

Updated App_Info.xml

referenceunits.zip

seti-p3-sse2.zip
Pentium M owners should use this one.

seti-pm.zip
Pentium 4 M owners should use this one.

Okay, this should be the last of them. I'll keep all of these on my site until at least the 29th. Don't know about after that.

seti-generic-sse2.zip

seti-plainvanilla.zip

Dig

Oddly the bump back to 1.67v brought my times back to averaging below 1hr (Unless I had a very unlucky 12 in a row)!

Last 10: 3496.13; 3,518.50; 2,904.31; 3,584.89; 3,470.45; 3,477.89; 2,871.13; 3,350.22; 2,903.78; 2,983.97

10 from 1.65v: 4,353.09; 4,605.95; 4,546.53; 3,647.80; 4,586.73; 4,545.45; 4,471.81; 4,245.88; 3,707.17; 4,414.03

I'd like to see what this is able to do if/when I upgrade to phase change :)

Eric

---

I thought to give the optimized clients a whirl. Inadvertantly, I had updated the Boinc manager to the new version, 4.43, before dropping in the .xml file, s@h client and the boinc.exe file.

This procedure seemed to work with Boinc manager 4.25, picking up work already d/l'ed, but with 4.43, it flushed the cache. Bummer....

Am I missing something?

Paul

---

Team Starfire World BOINC
IRC- irc//irc.teamstarfire.net:6667/team_starfire

I thought to give the optimized clients a whirl. Inadvertantly, I had updated the Boinc manager to the new version, 4.43, before dropping in the .xml file, s@h client and the boinc.exe file.

This procedure seemed to work with Boinc manager 4.25, picking up work already d/l'ed, but with 4.43, it flushed the cache. Bummer....

Am I missing something?

Paul

Paul,
By dropping in the BOINC.EXE, you have partly reverted to an earlier version. AFAIK The BOINC.EXE supplied in the optimized ZIPs is 4.25(?).
BOINCMgr will probably tell you it's 4.43, but it is communicating with an earlier BOINC.EXE.
This is probably not the reason your cache got hosed, but something you need to sort out. Re-installing 4.43 again should install the right BOINC.EXE again.
To check, the 4.43 files should be:
19/05/2005 10:40 299,008 boinc.exe
19/05/2005 10:43 1,257,472 boincmgr.exe

---

It would be great to see if FFTW and this MinGW compiler could produce some even more efficient AMD Athlon application versions. Is anyone else out there with the expertise willing to give it a shot? I'm simply don't have the programming skills to pull it off myself.

I tried for nearly 2 days (and nights) and now i'm giving up on that, cause i've got real work to do and also have a life ;)
FFTW compiles fine with GCC in MinGW even with optimizations for SSE, SSE2, 3Dnow and 3Dnow+ enabled. But i can't get the boinc core and the seti cruncher near to anything similiar to compile. I don't have a hand on VC++ 7.x, but even with VC++ 6 you can't do anything because of this /censored/ 'Solution'-files. It's a real weird tactic from M$ of hiding what's going on under the hood.
Sorry, but at least i tried...

---

Aloha, Uli

uh guys i downgraded to boinc 2.5 to test this out. im using the p3 zip and followed the directions on the top.

however, in the messages tab, it still says im running with seti version 4.09.

help?

---

uh guys i downgraded to boinc 2.5 to test this out. im using the p3 zip and followed the directions on the top.

however, in the messages tab, it still says im running with seti version 4.09.

help?

actually nvm. i checked in task manager, it's using 4.11.


well anyways. will this work on all versions of boinc? or just 2.5.

---

uh guys i downgraded to boinc 2.5 to test this out. im using the p3 zip and followed the directions on the top.

however, in the messages tab, it still says im running with seti version 4.09.

help?

actually nvm. i checked in task manager, it's using 4.11.


well anyways. will this work on all versions of boinc? or just 2.5.

It'll work with all versions of boinc ;) I'm currently running it on boinc 4.43 ;)

---

alright thankyou! i'll uninstall 4.25 and do the same thing 4.43 after this w/u.

it's running about 30 minutes faster now! =)

---

Hi all.

I have a question, after reading or skipping through the thread I noticed a comment that the seti P3 optimised exe may expire due to licensing obligations, is this true?, and if so has anyone posted one that wont? (in anticipation optimised for Athlon 3200 XP+), only hoping that we don't have to go backwards and also truly gratefull to those who have spent their time compiling these optimised clients!

Hi all.

I have a question, after reading or skipping through the thread I noticed a comment that the seti P3 optimised exe may expire due to licensing obligations, is this true?, and if so has anyone posted one that wont? (in anticipation optimised for Athlon 3200 XP+), only hoping that we don't have to go backwards and also truly gratefull to those who have spent their time compiling these optimised clients!

Currently it looks like the clients would expire on the 29th this month ;( The compiler that created these clients is simply too expensive to get past the expiration... So it looks like we'll have to revert to the official 4.09 at that time - but at least we'll know how much more we can get out of our processors ;)

---

Currently it looks like the clients would expire on the 29th this month ;( The compiler that created these clients is simply too expensive to get past the expiration... So it looks like we'll have to revert to the official 4.09 at that time - but at least we'll know how much more we can get out of our processors ;)

My understanding of the licence expiry, from what I read in another thread by the gentleman who compiled it, is that the binaries cannot be distributed after May 29th. But anyone with an existing binary would be able to continue to use it.

---

Anyone who is interested may read the EULA for yourself.
Look for the section on Evaluation License, especially the part in caps.

Anyone who is interested may read the EULA for yourself.
Look for the section on Evaluation License, especially the part in caps.

Well, I guess that's clear enough!

So what are we to do? Perhaps we could each run an evaluation in turn and compile the sucker. Is there not someone in the world who has bought the product and can compile the thing for us? Hellooooooooo....!?

---

can i have a look at the source code? just curious =)

---