hi astro, hope the tests go OK! feel free to try this when you get back, the more the merrier (or "the more, the crashed" perhaps!)

> hi astro, hope the tests go OK! feel free to try this when you get back, the
> more the merrier (or "the more, the crashed" perhaps!)
>

Thanks, Carl, appreciated.

I just edited these times into my last message but, as that message is now bypassed...

Abox, P4 2.8, SuSE 9.0, was 3.55 sec/TS, now 2.94 & 3.01 sec/TS

Bbox, P4 3.0, SuSE 9.0, was 3.52 & 3.68, now 2.92 sec/TS



________________________________________________
Video meliora, proboque; Deteriora sequor
I see the better way, and approve it; I follow the worse
-- Ovid (43BC-17AD)

wow that's a good improvement, about 20%. If these sorts of numbers hold, and runs are stable then this may be the way to go. hopefully it will be noticeably better for AMD64 users too, as I think cpdn/boinc is slower than cpdn/classic!

> the option I'm using on the build in the zip's now is: -xK using -ax* seems to allow Intel Fortran programs to "choose" giving the unoptimize d "generic" IA32 code for AMD procs, but xK means everybody running will need PIII-compatibility at least. Perhaps it's a little too "strict" for a 10-20% performance gain? We never really had anyone run with less than a P3 on the old CPDN anyway
>
> Anyway the -xK is supposed to be "Pentium III compatible only", although
> 'eeyore' reported a crash on an AMD Opterton so perhaps that's too "strict?"
> Any other Opteron users? It's chugging along fine on my AMD64 (after only a
> few hours, doing about 2.24 sec per ts versus 2.46 on the regular beta UM).
>

I haven't checked this build (running only Win boxes) but as i can read the post around I guess that several builds under each OS should be available and welcomed from users e.g.
- non-optimized code suitable for PII or older AMD (K6?).
- P4 optimized
- AMD64 optimized.

People having troubles should be adviced to run non-optimized code, others can perform nice 1/4 boots as reported.

<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=56&amp;trans=off">

well I hope to have an AMD64-specific one (for 64-bit Linux &amp; 64-bit Win), but it seems that settings are too precarious for these Fortran compilers &amp; the huge unified model. The problem is you need to run multiple machine types with the "new &amp; improved" settings for at least one and preferable two or more runs, and of course that's months. So I think it's best to go back to stable settings that everyone can run, at worst it's maybe 10% extra time for certain processors but I think overall it will be more reliable. We just don't have the personnel, machines, money, and time to really test &amp; optimize various combinations &amp; different compilers, having to buy various compilers, etc.

&gt; well I hope to have an AMD64-specific one (for 64-bit Linux &amp; 64-bit Win),
&gt; but it seems that settings are too precarious for these Fortran compilers
&gt; &amp; the huge unified model. The problem is you need to run multiple machine
&gt; types with the "new &amp; improved" settings for at least one and preferable
&gt; two or more runs, and of course that's months. So I think it's best to go
&gt; back to stable settings that everyone can run, at worst it's maybe 10% extra
&gt; time for certain processors but I think overall it will be more reliable. We
&gt; just don't have the personnel, machines, money, and time to really test &amp;
&gt; optimize various combinations &amp; different compilers, having to buy various
&gt; compilers, etc.
&gt;
&gt;
That's true for now as the boinc launch is set for this week.
But it would be great, if you could go on trying later when everything is settled down and you have time for something like this.
Thanks for your work, anyway.

The regular model is fine, this one freezes with the following:

Stack size=48.00 MB
Waiting for model startup, this may take a minute...
02lo_100028357 - PH 1 TS 000001 - 00/00/0000 00:00 - H:M:S=0000:00:00 AVG= 0.00 DLT= 0.00

..and there it stays

OS: knoppix (debian-derived) amd xp2 512mb ram.



reiteration.net networks

&gt; The regular model is fine, this one freezes with the following:
&gt;
&gt; Stack size=48.00 MB
&gt; Waiting for model startup, this may take a minute...
&gt; 02lo_100028357 - PH 1 TS 000001 - 00/00/0000 00:00 - H:M:S=0000:00:00 AVG=
&gt; 0.00 DLT= 0.00
&gt;
&gt; ..and there it stays
&gt;
&gt; OS: knoppix (debian-derived) amd xp2 512mb ram.

I should have mentioned, athalon xp2000+ with 512mb ram, abit mobo
reiteration.net networks

The new model takes more memory (54 MB) than the regular (49 MB) on my Pentium II without any improvement in speed but runs OK. SuSE Linux 9.1, kernel 2.6.5, 218 MB RAM. My swap space (512 MB) is not seen by BOINC.

yeah -- if you tried the new model I think it's best to go back to the original one. different model optimizations is something I'm going to have to try out on the sulphur cycle release; it just takes too long on this. Or perhaps we'll switch compilers to something more AMD friendly then.