I have just completed my first 50 hours of processing using BOINC for climateprediction.net on an old Windows 98SE machine. The Computer has a 933MHz Celeron (Tualatin core) CPU, 256MBs RAM ATA66 6GB HDD (with very little on it) 896MB fixed Page File Size and after several hours of testing is a very stable machine. It has antivirus software, a firewall, spyware removal, registry maintenance tools, all the usual software and consequently the OS and hardware are very stable. I don't actually use the machine myself, so when it is on (which is about 12 hours each day) it is used solely for BOINC and climateprediction.net processing.
Sufficed to say I have been receiving an error once roughly every 6 to 7 hours of use. The title of this error is "Visual Fortran run-time Error" and it contains 9 or 10 lines of error information within its error message box. The information means very little to me and contains the word “unknown” many times within its content, so it probably wont mean much to many other people either, but I was wondering if anybody else knew of a commonly occurring "Visual Fortran run-time Error" and if there are any recognised or known to work solutions to receiving this error that I could apply. I think the error may be stopping the upload of my processed data to climateprediction.net, as my user area shows “client error” as the “outcome” of all of my work units so far. I’m 99% sure that my firewall (ZoneAlarm) is properly configured to allow the BOINC software to download and upload when it wants so I don’t think that is an issue here.

Some additional information has just come up: The error that occurs shows the following information:
“forrtl: severe (30): open failure, unit 6, file CONOUT$”
It follows this line with: “HADSM3UM_4.12_WIN 008C765B Unknown Unknown Unknown” It repeats this line six (6) times but on each line it gives a new alphanumeric code such as “008B132A” or “008B0039” and so on. After these six lines it then says “KERNEL32.DLL” followed by another alphanumeric code and “Unknown Unknown Unknown” again. It shows this new line 3 times (it could be four times, but probably three) and again each time changes the alphanumeric code.

Any Advice on this error and how to prevent it from occurring would be most appreciated.

Thanks A lot,

Tom

---

(isn't the 933 a Coppermine?)

There's a very similar thread <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=2381">here</a> although it sounds more like it couldn't open the model specific stderr file for writing in your case. But maybe it's a related problem.

There seem to be a lot of those quite unspecific -5 errors lately :-/

If you're on BOINC 4.25, you could try to get a model over that problem with BOINC 4.19. You need to uninstall BOINC completely for this change. From what I heard the uninstall will not affect the project folder but it would be good if someone could confirm that.

A shutdown of the virus scanner might help too and as you say you don't use the machine much for anything else there shouldn't be a big risk - for a while at least.

(Sorry, Les, my DSL was gone for a moment and I hadn't seen that you had already posted about the AV software)

---

Hi Tom
Welcome to the forum and the project.

I'm afraid that Win98 was a last minute addition to the range of op systems before one of the 2 programmers left last year.
It got broken again when all the versions of hadsm got upgraded to match the requirements of the recent major upgrade to BOINC.
The remaining programmer is up to his eyebrows with problems affecting the bulk of the os users, as well as working on the new experiments.

There is strong evidence building, that free-av causes a conflict with cp.
See <a href="http://www.climateprediction.net/board/viewtopic.php?t=2895&amp;postdays=0&amp;postorder=asc&amp;start=0"> this</a> thread.

Also, there is an ongoing problem at Oxford affecting trickle uploads.
See <a href="http://www.climateprediction.net/board/viewtopic.php?t=2887&amp;postdays=0&amp;postorder=asc&amp;start=0"> this</a> thread.

Let us know about your av, and we may be able help some more.

Les

Thanks for responding guys.

Les: AV software is Avast AV Free Edition. I haven't tried turning it off yet so I will give that a go and see if it solves the problem.

Ananas: The CPU is a skt 370 1400MHz Celeron (Tualatin core) with 100MHz FSB in a socket 370 to slot 1 converter. Unfortunately my old Dell T450 XPS Box's PSU (or some other element) cant seem to handle the full 1.4GHz so (not being able to access the multiplier) I under-clocked the FSB to 66MHz (14x multiplier) which leads to 933MHz. This way everything is very stable and the the computer is still faster then when I had an old 600MHz slot 1 P3 with 100MHz FSB. It's not ideal, but Powerleap assured me the 1.4GHz would work (which it seemed to at 1st) but then it started to freeze up the PC after 5 or 6 hours of usual operation. Unfortunately, when this started to happen I had missed the money back/exchange period so I just underclocked it and everything has been fine since. I would have put a new PSU in the box (the current is only 200W) but there seems to be a special power connector on the PSU/Motherboard (probably for the AGP 2x graphics card) and none of the modern PSU's I have seen seem to have this extra power lead and I cant find any old Dell XPS PSU's with more than 200W on eBay. I have a 3GHz Preshott 800MHz FSB 1MB L2 Cache HT enabled system with all the other nice bits on it that is my main PC anyway, so I'm not too bothered if the old Dell isn't quite as fast as it could be...

If running things with the AV off doesn't solve the problem, I'll try and find, download and install BOINC 4.19 to see if that'll work.

Thanks a lot for your quick responses!

Tom

---

I've just had a look and I am actually running version 4.19 already. Perhaps I should try the 4.25 version to see if that works better? Might it be as simple as that? Probably is.

I'll report back when I've tried it out.

Thanks All!

---

No one has gotten WinME/98 to work with hadsm version 4.1x yet. Tolu has admitted that support for those OS's got broke when he created versions greater than 4.04 hadsm. You can try BOINC 4.2x but that is not the problem, it is the compiled CPDN hadsm executable.

&gt; ... I under-clocked the FSB to 66MHz (14x multiplier) which leads to 933MHz. ...

This explains it of course. I have a few Tualatins and the 933 confused me - could have been some exotic mobile one which would have added some extra thrill to finding the problem. All mobiles are exotic to me ;-)

_________________

I've got an idea for this Win98 problem - as many seem to be related to file open thing, could fiddling with FILES and BUFFERS in config.sys help maybe?

Maybe someone who has a backup of a crashed model from just before a crash could experiment with that - so he could tell if it changed the behaviour.

---

&gt; This explains it of course. I have a few Tualatins and the 933 confused me -
&gt; could have been some exotic mobile one which would have added some extra
&gt; thrill to finding the problem. All mobiles are exotic to me ;-)

Confuses me a bit too as 66 x 14 is 924 not 933... But maybe it rounds-up or something... I'm not an expert... :-)

Have turned off the active AV, but so far the software hasn't tried to upload anything so I can’t report any results yet.

&gt;No one has gotten WinME/98 to work with hadsm version 4.1x yet. Tolu has &gt;admitted that support &gt;for those OS's got broke when he created versions &gt;greater than 4.04 hadsm. You can try BOINC &gt;4.2x but that is not the problem, &gt;it is the compiled CPDN hadsm executable.

The statements directly above suggest that nothing I do with regards to AV software, or different versions of BOINC will fix the problem... As it appears that the problem is stopping processed results from getting from my PC to climateprediction.net, its not actually worth running BOINC for climateprediction.net unless I upgrade Win98SE to Win2000 or XP. Upgrading will cost (at least) £30.00/$57.00 and most probably £50.00/$95.00 (using eBay) and more again in the high-street shops.

Surely somebody can find at least a temporary work around for this problem. I know Win98 machines account for only a small percentage of the processed work achieved for climateprediction.net, but having old machines kicking about the house, its much nicer to use them to help the environment then to just dump them, or use them as door-stops. There must be some smart programmers or windows experts out there that could lend a hand with this issue?

Anyway, thanks for the responses. I'll keep you posted as to what develops with turning off the AV and changing BOINC version.

Tom.

---

its not actually worth running BOINC for
&gt; climateprediction.net unless I upgrade Win98SE to Win2000 or XP. Upgrading
&gt; will cost (at least) £30.00/$57.00 and most probably £50.00/$95.00 (using
&gt; eBay) and more again in the high-street shops.
&gt;
&gt; Surely somebody can find at least a temporary work around for this problem.

For people who only want to run the CPDN program, and not other DC projects under BOINC, the solution is to run the classic program. See <a href="http://www.climateprediction.net/board/viewtopic.php?t=1977"> this thread</a>.

Otherwise it is a matter of waiting until either the team (basically one guy at the moment) has this at the top of his priorities, or it fortuitously fixes itself in the course of another upgrade. The chances of the first of these happening looks slim at the moment, so you may have to hope for the second. It's not that nobody cares about Win98 users, it is just that the number of things that need to be done grows all the time.

---

&gt; For people who only want to run the CPDN program, and not other DC projects
&gt; under BOINC, the solution is to run the classic program. See <a> href="http://www.climateprediction.net/board/viewtopic.php?t=1977"&gt; this
&gt; thread</a>.

Thanks for that info Andrew. I have now moved over to the CPDN classic software and am awaiting the outcome of my computers work with it.

Hopefully, as you say, this will solve the problem and allow my Win98 machine to help produce some useful results.

If I have any problems with the classic CPDN software I will post them here.

Thanks to all of you who helped.

Tom

---

&gt; If I have any problems with the classic CPDN software I will post them here.

It is best to use the main community forum for that, as well as general and even BOINC issues. There is a better chance of a quick response, and a fund of information in existing threads.

---

&gt; &gt; This explains it of course. I have a few Tualatins and the 933 confused
&gt; me -
&gt; &gt; could have been some exotic mobile one which would have added some extra
&gt; &gt; thrill to finding the problem. All mobiles are exotic to me ;-)
&gt;
&gt; Confuses me a bit too as 66 x 14 is 924 not 933... But maybe it rounds-up or
&gt; something... I'm not an expert... :-)

Just thought I'd shed a little light on the clock speed here ... it's not 66.0Mhz * 14 ... it's all in multiples of 33.3333333333333Mhz (or 16.66666666MHz if you go back even further) (that's why the 486's were 33.33333, 66.66666, and 100 MHz rather than 33, 66, and 99Mhz - the 33 and 66 were just rounded down so it was easy to say), so yours is actually 66.666666Mhz * 14 = 933.333333Mhz, so the 933 measurement is correct.

As for the thread topic ... I can't get past a percent or 2 with CPDN now on my win98se box (a Celeron 1.7GHz, my main box) so it seems I might have to stop using it on this box, since I run other BOINC (4.19) projects on it. Thank goodness CPDN didn't upgrade the client version to 4.12 until it had finished my first unit. Fortunately we have just got another pc (that I won't be using since it's not my box) in the house that (as the resident IT guy) I have setup and just happen to put BOINC on that. It a 2.8GHz P4 with XP and I put BOINC 4.25 on it. So far it has apparently had issues with 2 units ... the third has been running for about 2.5 days now (along with various units from SETI and Predictor) and is at 2.57% so hopefully it will keep going a little more smoothly from now on. I'd hate to have to pull the plug on CPDN completely, but if it's not gonna get anywhere near through a unit then I might have to.

On the subject of AV progs ... my 98se box is running old old McAfee 4.03 (and I've been a little lax lately in updating the virus definitions ... I really should get around to that), I have recently excluded the BOINC folder and subfolders from scanning - doesn't appear to have changed anything though. The XP box is running a version of Norton Internet Security that came on the motherboard CD, and has Norton AntiVirus (OEM) 11.0.9.16 in it. I haven't excluded anything from scanning yet, but it sounds like I should exclude the BOINC folders and see how that goes.

Anyway ... this is starting to get a little long, so I'll leave it there.

bb from Oz.

---

I think we have to regard the Boinc version of CPDN as basically broken now for Win 9x/ME/NT, possibly for good. The problem is partly that most machines running these OS are at the limits for CPDN anyway. As CPDN Classic is still available for them, and probably will be until the money runs out, then it's the old issue of competing priorities for the tech's time.

Problems under WinXP are a different matter. A lot of them have proved to be the result of overheating, flakey memory etc, which are only really apparent when the PC is asked to undertake a heavy duty application like this one. But the rest are almost certainly down to software conflicts, and there are several threads on the community forum discussing these things. See especially <a href="http://www.climateprediction.net/board/viewforum.php?f=4">this section</a>. Contributions welcome. Of course, what is troublefree for one person may crash another's machine, depending on the particular combination of hardware and software they are running, but the shared effort has been rewarding so far.

---

Tom

I setup my win98 SE machine with no AV, anti spyware etc.

CPDN crashes at about 8 hrs.

I can happily run Boinc 4.19, 4.25 &amp; 4.32 with seti/einstein/protein predictor on it. Thus the problem seems to be the later cpdn models only.

I gave up on cpdn under boinc on my older computers and instead run boinc and the other projects. I run cpdn/boinc on a more powerful machine with win xp. Good luck running the classic cpdn SW and please let us know how it goes

Dave

---

This is now official : the compatibility with Win98 and Millenium is broken since Cdpn 4.10
http://www.climateprediction.net/board/viewtopic.php?t=2923


Didn't know that a Barton 2500+ was at the limit !
I had No problem with Win98SE and Cdpn before the update.

---

&gt; Didn't know that a Barton 2500+ was at the limit !
&gt; I had No problem with Win98SE and Cdpn before the update.

It may be that the problem will be fixed eventually. But the then Chief Software Engineer announced last autumn that he was having to withdraw support for Win 98/ME/NT. The problem at that time was found and put right despite that, but he has since left and the list of things to be fixed has got longer. The harsh truth is that there are all sorts of hardware and OS combinations out there and it would be nice if all could take part in the project, but it just isn't going to be possible.

There are, of course, a fair number of upgraded and home built machines to a high spec running a legacy Win OS, and some organisations still load it on new equipment, usually because they are running old bespoke software that cannot be upgraded. But the number of those that then want to run CPDN under BOINC is unlikely to repay the effort of tracing the problem when assessed against other priorities.

---

thks for your answer :)

I'm probably lucky that i don't try to put a Beos machine under Cdpn...
When i look Seti 1, many platforms were supported (like OS2, Beos...), i doubt that i will see the same "soon" here (or elsewhere also).

Anyway Cdpn has really too many problems with the software, it's really beta.

Removing the support for Win9x and ME will not improve this situation anyway, i think.

So, i detached my Win9x/ME machines from this project and if the situation is not improved, i will do the same with my Win2k machines.

Some 1500 people are left out (according BoincStats), this is perhaps not a lot but this is enough for me.

---