My main desktop PC (AMD Duron) isn't doing very well, it's got 5.11s/TS, does any one else have an AMD Duron they can show me the performance of? I don't think mine is running at it's best. My notebook is running at 7s/TS aswell, but that's a mobile AMD AthlonXP 1900+.

Are those the right speed or is it going slower than usual? My notebook is totally dedicated to BOINC, I even set it to "normal" priority (not sure if it changed back to low like is usually does).

2,27
amd3000+ 512 l2 cache win xp

live long and prosper ;-)

Also a Duron with 1400Mhz - 5.0s/TS

> Also a Duron with 1400Mhz - 5.0s/TS
>
>

How much memory and OS has that PC got? Your s/TS seems to be lower than mine, although mine is 400mhz faster.

P4 2.4mhz (1024mb RAM) = 2.87 s/TS (windows XP)
P4 2.6mhz (768mb RAM) = 3.22 s/TS (linux)

Wow - about 500 boxes on Carl's CPU list.
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html

Carl, how was the list harvested?
I really don't mind my box(es) being on the list - just currious how is was created. btw, any G5s so far?

I can also see on my trickle page [29 of them so far] that is started at 1.9 sec/TS and settled down after 7-10 of them at about 1.79-1.81 sec/TS.

btw, i really like those new phase temp/precips graphs using PNG format.

> Wow - about 500 boxes on Carl's CPU list.
> http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html
>
> Carl, how was the list harvested?
> I really don't mind my box(es) being on the list - just currious how is was
> created. btw, any G5s so far?
>
> I can also see on my trickle page [29 of them so far] that is started at 1.9
> sec/TS and settled down after 7-10 of them at about 1.79-1.81 sec/TS.
>
> btw, i really like those new phase temp/precips graphs using PNG format.
>
>

Hey carl, could you make a XML file of that table?
<img src="http://www.funkymonkey.org/boinc/sig.php?id=87&amp;proj=cpdn">

Hy Scott,

Duron 1400 Mhz, 256MB, W2k, 5.0 s/TS

&gt;&gt; http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html

Hi, that page is just something I threw together for pc's that have trickled at least once, it's generated once a day. I put an .xml.gz (gzip compressed xml) in the /stats dir

There's no user info so I figured it's "safe" in that it would be pretty hard to figure out who is who based on the info provided?

it appears that cpdn/boinc is Pentium4/WindowsXP-"centric" so far! I will experiment with other options. Unfortunately Mac's are a bit underrepresented due to the sluggishness of the Mac Fortran versus Intel.

Thx Carl!!
Have now installed Linux on my AMD 3200+ 64 and have 2,17 s/Ts in 32 Bit mode
<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=13">

I have ~ 3,16s/TS on my Athlon XP2000+ with Win2k pro.
Is that not a little too fast?

&gt; &gt;&gt; http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html
&gt;
&gt; Hi, that page is just something I threw together for pc's that have trickled
&gt; at least once, it's generated once a day. I put an .xml.gz (gzip compressed
&gt; xml) in the /stats dir
&gt;

&gt; it appears that cpdn/boinc is Pentium4/WindowsXP-"centric" so far! I will
&gt; experiment with other options. Unfortunately Mac's are a bit underrepresented
&gt; due to the sluggishness of the Mac Fortran versus Intel.
&gt;
Hi Carl,

Something curious about this table. My dual Opteron 246 running Linux with 2.14s/TS and 2.21s/TS seems to have slowed down from 2.175s/TS to 6.167.

Flying the flag for Linux :)

&gt; &gt; &gt;&gt; http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html
&gt; &gt;
&gt; &gt; Hi, that page is just something I threw together for pc's that have
&gt; trickled
&gt; &gt; at least once, it's generated once a day. I put an .xml.gz (gzip
&gt; compressed
&gt; &gt; xml) in the /stats dir
&gt; &gt;
&gt;
&gt; &gt; it appears that cpdn/boinc is Pentium4/WindowsXP-"centric" so far! I
&gt; will
&gt; &gt; experiment with other options. Unfortunately Mac's are a bit
&gt; underrepresented
&gt; &gt; due to the sluggishness of the Mac Fortran versus Intel.
&gt; &gt;
&gt; Hi Carl,
&gt;
&gt; Something curious about this table. My dual Opteron 246 running Linux with
&gt; 2.14s/TS and 2.21s/TS seems to have slowed down from 2.175s/TS to 6.167.
&gt;
&gt; Flying the flag for Linux :)
&gt;
&gt;

Linux rules. I have a much better performance on it!!! i hope 64 Bit Client will appear soon!!!!!!!!!!!!
<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=13">

My in-house Linux vs. M$ comparison:

Abox P4 2.8 GHz Linux SuSE 9.0  ~11:10 time/Trickle, ~3.8 sec/TS, ~1.9 sec/TS throughput
Bbox P4 3.0 GHz Linux SuSE 9.0  ~11:10 ................ ~3.8 .......... ~1.9
Dbox P4 3.0 GHz WinXP ............ ~ 8:40 ................ ~2.9 .......... ~1.45 !!! (2.5 hour per Trickle improvement over Linux is significant, IMO.)


Classic CPDN:
Abox throughput w/classic ................ ~2.53 sec/TS
Cbox (also 3.0 GHz) w/classic Model ... ~2.40
Dbox w/classic Model ...................... ~2.38


I don't get it at all.  Similar Intel machines; major OS difference in performance.  As Carl notes, apparently Intel Fortran loves M$/Windoze more than Linux (a love affair shared by many mfg).

If this holds, when Beta is done, if not before, I'll have three boxes in
Windoze again.  (Bbox is Linux-only; a mistake, apparently.  [My faith in the
efficacy of Linux took a big hit!])  Which brings up a question --
whether runs can be copied from Linux to M$ in the same manner as between
upgrades.


I checked 'Benchmark' values to compare M$ to Linux:
Mem. bandwidth is the same across the board (953.67).
U/L &amp; D/L rates differ.

..................... Float.Pt.--- Integer
Abox P4 2.8 GHz 534.07 --- 1143.2  Linux
Bbox P4 3.0 GHz 682.76 --- 1249.3  Linux
Dbox P4 3.0 GHz 1639.2 --- 1891.1  WinXP -- Hmmm....


(Bit of a pain to make tables here, eh?)
________________________________________________
It is impossible to enjoy idling thoroughly unless one has plenty of work to do.
-- Jerome K. Jerome (1859, 1927)

Strange my client rusn faster on Linux!!!!
<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=13">

&gt; I checked 'Benchmark' values to compare M$ to Linux:
&gt; Mem. bandwidth is the same across the board (953.67).
&gt; U/L &amp; D/L rates differ.
&gt;
&gt; ..................... Float.Pt.--- Integer
&gt; Abox P4 2.8 GHz 534.07 --- 1143.2  Linux
&gt; Bbox P4 3.0 GHz 682.76 --- 1249.3  Linux
&gt; Dbox P4 3.0 GHz 1639.2 --- 1891.1  WinXP -- Hmmm....

Emma.. XP 2.6 GHz 3203 ----- 5668... WinXP 1.95 secs/timestep.. :-?

The BOINC benchmarks don't seem to agree very well with CP-boinc performance. :-?

<a href="http://www.nmvs.dsl.pipex.com/"><img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=6&amp;team=off&amp;trans=off"></a>

&gt; The BOINC benchmarks don't seem to agree very well with CP-boinc performance.
&gt; :-?

yeah it is kind of odd, which I guess is good that I count credits as trickle/timesteps completed, so it's even across the board. I'm not quite sure why the benchmarks vary so much, and there's little correlation to CPDN, I suppose having to do with the Intel compiler. I mean the AMD64's are way ahead on "mega-flops" and "mega-integer-ops" but not way ahead on seconds per timestep (although very respectable at 1.8s/ts).

I'm hoping to get a chance to experiment more with compiler settings, especially on the Intel Linux compiler as that doesn't seem to be consistent as people are reporting it's slower than Windows, but all of our machines here in "the Dungeon" show that the Linux CPDN/BOINC is faster than the Win one.

&gt; &gt; ..................... Float.Pt.--- Integer
&gt; &gt; Abox P4 2.8 GHz 534.07 --- 1143.2  Linux
&gt; &gt; Bbox P4 3.0 GHz 682.76 --- 1249.3  Linux
&gt; &gt; Dbox P4 3.0 GHz 1639.2 --- 1891.1  WinXP -- Hmmm....
&gt;
&gt; Emma.. XP 2.6 GHz 3203 ----- 5668... WinXP 1.95 secs/timestep.. :-?
&gt;
&gt; The BOINC benchmarks don't seem to agree very well with CP-boinc performance.
&gt; :-?

Hi, Nick,

Exactly; the numbers are all but meaningless. (However, Emma &amp; Dbox are similar on floating point. P4 numbers are per virtual CPU, double them for total.)

Cheers,
Jim
________________________________________________
It is impossible to enjoy idling thoroughly unless one has plenty of work to do.
-- Jerome K. Jerome (1859, 1927)

&gt; I can't get the visualizations to work, but my trickles page shows three
&gt; trickle results. Two 1.76, and a 1.77.
&gt;
Where's the trickles page? I can't find it

*edit* Found them but can I say the site is very hard to navigate to find such things, like going through a maze ;)

P4 2GHz 3.16 - 3.22 was fastest when I first started, slowed down for some reason now

&gt; Hy Scott,
&gt;
&gt; Duron 1400 Mhz, 256MB, W2k, 5.0 s/TS
&gt;
&gt;
&gt;

How confusing, I have double the memory, Windows XP, and a Duron 1800MHZ and your s/TS is lower?

Do you use the computer? or is it a dedicated system?