after up to 4.13 ,1 of 3 AMD's, (a) the best, fastest and most stable, did that fortran error thingy, put in half a day on a new one & etc. &etc. so I detached it. some notes;
(a) athlon xp 2100 512 Mb R.A.M. crashes, ran @ 10.4 sec./ts win98
(b) athlon M 1800 256 Mb - 64 Mb for onboard vid .runs great @ 4.8 s/ts win98
(c) Duron 550 Mhz takes 14.4
seti classics in same order 3:15 / 4:00 / 24:00
I Build these things for a living (PI is easy), (a) is stable.

(a) is the only machine doing also SETI, LHC, and Pirates. Set to keep in Memory when suspend'd.

so;
1) why (a) so slow ?
2) why (a) crash ?
3) How 2 fix eh ?

P.S. (a) was awarded 2500 bonus cobbles ? for its first crash ??

<img src="http://www.setisynergy.com/images/stats/3638cpdn.png">
<img src="http://www.setisynergy.com/images/stats/comb-162.jpg">

Well, 1 of my AMD machines is doing the same. The crash thing.

It already crashed two results. I use boinc 4.13 aswell. I noticed there is a new version of hadsm (4.04), Windows 98/ME/NT4/2000/XP 4.04 29 Sep 2004 12:17:40 UTC.
Ok so my first result to go bye bye was at about 80% complete, been working it for quite some time now. it died on 18 Oct 2004 12:58:28.

It downloaded a new one and after 3 mere tricklets died again. at 25 Oct 2004 15:13:49

I have a feeling this is going to be a trend. Oh I had to terminate BOINC btw (clicking away the error message wasnt enough) once that happened. AND restart the computer since restarting boinc still left it gittery (it probably just didnt terminate hadsm but i didnt wanna bother with that).
I do not have a superman complex; for I am God, not superman!

well i now got a similar error message, but this time for an LHC application.

I am downgrading to boinc 4.09 to see if it solves the problem
I do not have a superman complex; for I am God, not superman!

ok, i downgraded to 4.09 and it STILL does this, I am thus temporarily detaching from CP.
I do not have a superman complex; for I am God, not superman!

&gt; ok, i downgraded to 4.09 and it STILL does this, I am thus temporarily
&gt; detaching from CP.

My other machine a few mounths ago, errored on more than one project, which pointed me to a dos mode scandisk (full /surface) which turned up a bad sector that windows couldnt find.

My machine however Has eaten many lhc seti predicator &amp; pirates it passes burnin programs like "pi"

Its winter for the last month my machines haven't hit 30 C cpu temp
I stopped overclocking it so I could do cpdn on it (even though
1. cpu was overclocked by default (very slightly) on E.C.S. mainboard.
2. ran bulletproof 2 years @ 14% over (did 3:15 per SETI classic {Athlon 2100))

The only "wildcard " factor I can see was that after 3.5 weeks I upgraded to 4.13 right at the time it was swapping out lhc or seti for CPDN (I noticed it was swapping right at the hour and looked at the time when it went "crash"
after reboot n stuff it ran for 3 hours "crash" Right after it sent that unit and got another (that wu was about 15% done and I got points maybe it was an iceball ?)

I wont downgrade the boinc cince lhc and seti work I have two machines doing just CPDN I wanna keep at least 1 machine to hunt gravity waves ,aliene, fold protines n stuff, or I would try CPDN only on this machine to "prove" if it is the swapping /resume process.

oh one other Q for ya,
My other machine (m8) is overclocked substantially, its one of those semprons that come with a mainboard for about 75$ its a Athlon 1800+ M parading in overclocked form as a 3300+ 2Ghz actual
after this wu (its at 65%)would You reclock it ?

<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=372&amp;trans=off">

I noticed it happens on extreme HDD activity, uploading and downloading 30 things at once.
I do not have a superman complex; for I am God, not superman!

Check this thread:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=1193
I do not have a superman complex; for I am God, not superman!

Well, Back again.. I threw my newest machine across all projects, and it wotks just fine, I guess the problem must be the athlon doesn't like cpdn , hopefully
the two athlons comming in this week (also "palimino's") will be cpdn happy.
an off topic thanks to cpdn for win 98 support,my xp box at 100% cpdn has taken 66 days to reach 47% , my win 98 machine is at 99.97% in 43 days, ever wonder what the xp does with all those cpu cycles ?

<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=372&amp;trans=off">
<img src="http://www.setisynergy.com/images/stats/3638cpdn.png">

&gt; Well, Back again.. I threw my newest machine across all projects, and it wotks
&gt; just fine, I guess the problem must be the athlon doesn't like cpdn ,
&gt; hopefully

Not, that Intel 3.2 Ghz started crashing after timestep 4 , after f@##'ng up a few I detached it, my two machines working ok CPDN + Pirates My other 5 (here) all run SETI,LHC,Pirate, and all work fine.
No running the timer while suspended, or seti's running for hours at 100% , which is easily observable on all win98 machines with both seti &amp; CPDN hopefully these problems will be addressed , and a notice alerting us to sharing problem fixxed, the huge 6 wu uploaded in crashed state is unacceptable and expensive to contemplate across a dozen or so machines.

---

Ha well my win98 machine, CPDN &amp;pirates is at 50% on its secomd wu. so that combo runs on win98. my other win98 machines still wont share between seti &amp; CPDn.
my new machine has no problem (seti,lhc,seti,pirate,) running windows xp pro x64 (64 bit edition), too bad for cpdn the majority of my machines will continue to run win98.

---

Im running cpn on around 10 different computers, mostly Intel, some g4 and an Athlon xp-m. The Athlon is the only one that the cp model keeps crashing on, the machine is stable with any other program, it’s on all the time, dose not get rebooted for months so I don’t think it the machines problem. CP just seems a little flaky on AMD processors.

Its Interesting to watch the CP open day vids when the bloke who developed seti and boinc says that they don’t have any AMD boxes to test there software on and is relying on the AMD crowed to optimise there own versions. Maybe there is a problem as it was not tested thoroughly (or at all) on AMD processors.

I would really like this sorted out but i know the guys running this project are very busy, funds are limited and are trying to cater for the vast majority (i.e. Intel crowd). Hopefully future releases will be tested for stability on AMD platforms and it would be nice if they could use a compiler that enables SSE/SSE2 on AMD processors as well as on Intel processors.

I will let my Athlon try another CPN model but if this one crashes then i will put this computer back on my other projects as its just wasting CPU time.

---

&gt; get rebooted for months so I don’t think it the machines problem. CP just
&gt; seems a little flaky on AMD processors.
&gt;
&gt; Its Interesting to watch the CP open day vids when the bloke who developed
&gt; seti and boinc says that they don’t have any AMD boxes to test there software
&gt; on and is relying on the AMD crowed to optimise there own versions. Maybe
&gt; there is a problem as it was not tested thoroughly (or at all) on AMD
&gt; processors.
&gt;
Hi Craig,

It seems to work fine on all my AMDs, 14 completed vs one aborted on an A64 3400+ (classic and BOINC) , 3 completed with no aborted on an XP 3000+ (BOINC), 10 completed vs. 1 aborted on an A64 3200+ (classic and BOINC), 7 completed vs. 1 aborted on an XP 3200+ (classic models on that one).

All alpha and beta testing are done on both Intel and AMD processors (and sometimes Apple). If it wasn't stable in beta testing on one or the other platforms, it wouldn't be released.

But I agree that it would be nice to have true optimizations for AMD processors. Intel got a huge boost in performance going into BOINC because of the Intel compiler, while XP's stayed about the same, and A64s slowed down some. A newer version of the Intel compiler supposedly doesn't penalize A64s, and can optimize for AMD as well (I'm sure not as much as for Intel however).