Since upgrading to BOINC 4.05/SETI 4.03 (the most current SETI/BOINC application,) when I begin to process a unit, I get nothing but blue screens which cause me to need to reset the computer. I tried an overlay install of the BOINC client, and a clean install, after making sure all the old folders and files for the previous version of the program are gone.

The error I saw listed on the blue screen is "page fault in non-paged area" which seems to indicate a memory issue. My memory seems to be completely fine beyond SETI/BOINC.

My system is based on a ASUS A7N8X-X motherboard with a AMD Athlon XP 3000+ (2,100 MHz) on a 400 MHz FSB. I have 768 MB Kingston Value RAM, unbuffered. This is the type of RAM prescribed for this motherboard by the motherboard manufacturer. I have been assured by the motherboard manufacturer that all components are compatible. I am using the most up-to-date drivers for my motherboard and all components. I am using Microsoft Windows XP Home Edition SP2.

I really would like a valuable response to this issue, else I will have to no longer run SETI@home.

I've been a member of SETI@home since December 6, 2000.

[EDIT - ADDITIONAL INFO]

This problem goes away when I underclock my CPU to a AMD Athlon XP 2700+ on a 333 MHz FSB, running at about 2 GHz. This CPU configuration actually does not exist as a CPU you could buy from a AMD reseller, but this is how my motherboard recognizes it. I would also point out that the classic SETI didn't work at the 400 MHz FSB either, and I had to do the same underclocking to get that to work also. I just can't see this as an incompatibility issue between the 400 MHZ CPU/RAM FSB speed and the motherboard, otherwise Windows would crash even without SETI running, and obviously, it doesn't. On a side note, I am only losing about 5% of the processor's processing power with this underclock fix.

MJ's Star Trek Portal

Sounds to me like your processor is overheating. Is your cpu fan working properly? Are your case fans etc. working ok?

HH

[Post removed by poster -- poster error]

> Sounds to me like your processor is overheating. Is your cpu fan working
> properly? Are your case fans etc. working ok?
>
> HH
>
>

All fans are working properly. As I stated, the blue screen error code seems to indicate a memory issue.

just a hunch - a lot of ppl have problems with XP's SP2. But it'd probably be widespread knowledge that seti doen't work with it.

> just a hunch - a lot of ppl have problems with XP's SP2. But it'd probably be
> widespread knowledge that seti doen't work with it.
>
>

I have heard this too. I suppose I could uninstall SP2 and see if BOINC works any better. I just don't want to go through the 80 megabyte download again, or wait for the CD to come in the mail, if SP2 is not the issue. The upgrade to the latest BOINC coincided within a week of the release of SP2, so I can't be sure which one of them is the culprit. Since BOINC was down for the whole time between SP2's release and the release of the latest BOINC, and I didn't have any work in my queue to test SP2, I still can't be sure SP2 is the issue in my case.

I do know that clocking down to a 333 MHz FSB/RAM bus speed seems to solve the issue, so I suspect I might have run into a hardware issue, but Windows itself, and none of my other programs malfunction with the 400 MHz FSB/RAM bus speed, so I'm still lost, and suspecting BOINC is the culprit. I don't know, I guess I can live with the clock-down since it only costs me about 5% of my processor's peak processing power, but I can't figure out why SETI is the only program I have that malfunctions on the 400 MHz FSB/RAM bus speed. Absolutely none of my other programs cause this issue, and Windows itself seems happy with it as well.

> > Sounds to me like your processor is overheating. Is your cpu fan
> working
> > properly? Are your case fans etc. working ok?
> >
> > HH
> >
> >
>
> All fans are working properly. As I stated, the blue screen error code seems
> to indicate a memory issue.
>

Well, I have seen similar errors when case temps got too hot. Keep in mind your L1 and L2 cache memory are on your processor die. SP 2 could be the problem as well, I guess. I've put off upgrading till I see how it works for other people. It doesn't really offer anything I need anyway. Other than that I guess I can't help you. Good luck!

HH

> > > Sounds to me like your processor is overheating. Is your cpu fan
> > working
> > > properly? Are your case fans etc. working ok?
> > >
> > > HH
> > >
> > >
> >
> > All fans are working properly. As I stated, the blue screen error code
> seems
> > to indicate a memory issue.
> >
>
> Well, I have seen similar errors when case temps got too hot. Keep in mind
> your L1 and L2 cache memory are on your processor die. SP 2 could be the
> problem as well, I guess. I've put off upgrading till I see how it works for
> other people. It doesn't really offer anything I need anyway. Other than
> that I guess I can't help you. Good luck!
>
> HH
>

Well, according to Motherboard Monitor and my motherboard's hardware monitor, none of my temps are in the red. My motherboard also has a feature that shuts down the system if there is anything wrong with the fans or temperatures. There would be no blue screen with another error of any type. The system would just shut off immediately if anything was out of range.

Here are my readings in Motherboard Monitor with a full load:

Case: 75 F
Motherboard: 77 F
CPU diode: 118 F
CPU Socket: 111 F
CPU Fan: 5,100+ RPMs
Case Fan: 2,300+ RPMs

My motherboard's hardware monitor tends to agree with these readings, with the exception of the CPU temps, which would be absolutely normal, since the motherboard's hardware monitor would read the temps when there is no load on the CPU. The readings in Motherboard Monitor are with a full CPU load and SETI/BOINC running (at the underclocked 333 MHz FSB/RAM bus speed.) The CPU temps when there is no load on the CPU are in the mid and high 90s F, which is well in the safe zone.

Ok, what does your MM read when you're running at 2.1/400 full load? Or, maybe you can't get into it in time to see? My 2800+ will start acting up somewhere above 50C (socket sensor) when it's hot here. Well below damaging temperatures but still hot enough to cause problems. I'm wondering if when you clock it up to where it's supposed to run if thats just enough to cause heat problems?

Have to go here. Again, good luck!

HH

> Ok, what does your MM read when you're running at 2.1/400 full load? Or,
> maybe you can't get into it in time to see? My 2800+ will start acting up
> somewhere above 50C (socket sensor) when it's hot here. Well below damaging
> temperatures but still hot enough to cause problems. I'm wondering if when
> you clock it up to where it's supposed to run if thats just enough to cause
> heat problems?
>
> Have to go here. Again, good luck!
>
> HH
>
>

50C being about 122F, that is right about where my CPU runs in the normal 2.1 GHz/400 FSB/RAM configuration, sometimes spiking up to about 57C/134F. I guess it's possible that, somewhere along the line, the tolerances have been tightened by something. I just don't know where that would be, and I'm not convinced that such a narrow operating limit exists, although I am not a tech guru. I can tell you that earlier versions of XP (SP1 and SP1a) and BOINC (versions just before the current one) never malfunctioned when the temperatures were at the values mentioned above. Once more, though, would a CPU temperature issue masquerade itself as a memory parity issue, and why did it change so radically after the SP2/BOINC 4.05 upgrade? I'm still not convinced SP2 would be the cause. Either way, since there is no way to really remedy the issue, if it is a temperature issue, I'm resigned to keeping the clock down to 333 MHz.

Well, I went through some of the other forum areas and discovered a download link for version 4.06 of BOINC and this seems to have fixed my issues, so, obviously, it was at least partly related to BOINC. I have put my clocks back to 400 MHz and all seems to be fine, although it's only been about 20 minutes since I changed all of this around, so we will see if this holds out for longer.

> Well, I went through some of the other forum areas and discovered a download
> link for version 4.06 of BOINC and this seems to have fixed my issues, so,
> obviously, it was at least partly related to BOINC. I have put my clocks back
> to 400 MHz and all seems to be fine, although it's only been about 20 minutes
> since I changed all of this around, so we will see if this holds out for
> longer.
>
>

Well, a half hour later now, and things are acting up again, same error, so I am back to the 333 MHz FSB/RAM bus speed. There doesn't seem to be any other solution at this time.

I would remove the cover from your case. That will drop all temperatures significantly (cpu, memory, chipset etc.). Then, if it still does it, my hunch is wrong and you'll be rid of me!

Sorry I didnt mention the 4.06 release however, it was primarily developed for Win 9x problems I beleive, but still, was a good idea for you to try.

HH

> I would remove the cover from your case. That will drop all temperatures
> significantly (cpu, memory, chipset etc.). Then, if it still does it, my
> hunch is wrong and you'll be rid of me!
>
> Sorry I didnt mention the 4.06 release however, it was primarily developed for
> Win 9x problems I beleive, but still, was a good idea for you to try.
>
> HH
>
>

I always run with my case covers off. It doesn't change anything. I still do not think it's a temperature issue. It worked fine up until last week. As I said, I don't know if XP SP2 is the issue, or if BOINC itself is faulty, and I am not willing to uninstall SP2 to find out.

You are probably correct. Just my thoughts. If it was working before it's probably not a temperature problem.

I will try to keep an eye on this thread to see if you figure it out. I'm curious.

Good luck!

HH

> You are probably correct. Just my thoughts. If it was working before it's
> probably not a temperature problem.
>
> I will try to keep an eye on this thread to see if you figure it out. I'm
> curious.
>
> Good luck!
>
> HH
>

Well, it seems I am able to clock it up, still on a 333 MHz FSB/RAM clock, but I adjusted the bus clock multiplier to bring the CPU up to ~2,083 MHz (an Athlon XP 2800+ Barton Core.) This is ever so slightly short of the supposed limit of my Barton Core Athlon XP 3000+. It's about 17 MHz, or about 1% short of the maximum rated processing power of the processor. If it holds, I can accept it, since it is a negligible loss. This fix has been in place for about 1 hour and 20 minutes. It should be noted that this fix was not working before I found the BOINC v4.06 update. It would seem that the folks over at the SSL may be working up to finally solving the issue.

[EDIT - ADDITIONAL INFO]

2 hours and 43 minutes into the test with my bus clock multiplier pushed up 1/2x and everything appears to be working properly. The CPU seems to be handling BOINC just fine at 2,083 MHz.

[EDIT #2 - MORE ADDITIONAL INFORMATION]

About one hour after the "additional info" snippet above, my PC gave me the same blue screen and error code, but at least it is much less common. I was working on a web page in a HTML editor at the time (and, coincidentally, lost the changes I made.) This has to be a memory leak, or a memory assignment error.

[EDIT #3 - 09-20-04 - EVEN MORE ADDITIONAL INFORMATION]

Well, I have been running further tests in regard to this issue. The 400 MHz FSB appears to *not* be the issue, and thus it doesn't seem to be a motherboard issue (though I could be wrong).

I am underclocking my CPU 1/2x to 2,000 MHz on a 400 MHz FSB, and I don't see any issues yet. Anything over 2,000 MHz on both the 400 and the 333 MHz FSB is unstable (with SETI-BOINC and the SETI Classic) and causes blue screens with all kinds of errors that do not happen when SETI@home is not running, even when the CPU is clocked to where it belongs (2,100 MHz) on the 400 MHz FSB. Windows and all of my other programs (aside from some non-essential and unimportant games) do not seem to care at all when the CPU and FSB are clocked up where they belong. I am forced to conclude that this issue is in fact some uninvestigated (and unknown) SETI problem with an AMD Athlon XP processor working at greater than 2,000 MHz.

> > You are probably correct. Just my thoughts. If it was working before
> it's
> > probably not a temperature problem.
> >
> > I will try to keep an eye on this thread to see if you figure it out.
> I'm
> > curious.
> >
> > Good luck!
> >
> > HH
> >
>
> Well, it seems I am able to clock it up, still on a 333 MHz FSB/RAM clock, but
> I adjusted the bus clock multiplier to bring the CPU up to ~2,083 MHz (an
> Athlon XP 2800+ Barton Core.) This is ever so slightly short of the supposed
> limit of my Barton Core Athlon XP 3000+. It's about 17 MHz, or about 1% short
> of the maximum rated processing power of the processor. If it holds, I can
> accept it, since it is a negligible loss. This fix has been in place for about
> 1 hour and 20 minutes. It should be noted that this fix was not working before
> I found the BOINC v4.06 update. It would seem that the folks over at the SSL
> may be working up to finally solving the issue.
>
> [EDIT - ADDITIONAL INFO]
>
> 2 hours and 43 minutes into the test with my bus clock multiplier pushed up
> 1/2x and everything appears to be working properly. The CPU seems to be
> handling BOINC just fine at 2,083 MHz.
>
> [EDIT #2 - MORE ADDITIONAL INFORMATION]
>
> About one hour after the "additional info" snippet above, my PC gave me the
> same blue screen and error code, but at least it is much less common. I was
> working on a web page in a HTML editor at the time (and, coincidentally, lost
> the changes I made.) This has to be a memory leak, or a memory assignment
> error.
>
> [EDIT #3 - 09-20-04 - EVEN MORE ADDITIONAL INFORMATION]
>
> Well, I have been running further tests in regard to this issue. The 400 MHz
> FSB appears to *not* be the issue, and thus it doesn't seem to be a
> motherboard issue (though I could be wrong).
>
> I am underclocking my CPU 1/2x to 2,000 MHz on a 400 MHz FSB, and I don't see
> any issues yet. Anything over 2,000 MHz on both the 400 and the 333 MHz FSB is
> unstable (with SETI-BOINC and the SETI Classic) and causes blue screens with
> all kinds of errors that do not happen when SETI@home is not running, even
> when the CPU is clocked to where it belongs (2,100 MHz) on the 400 MHz FSB.
> Windows and all of my other programs (aside from some non-essential and
> unimportant games) do not seem to care at all when the CPU and FSB are clocked
> up where they belong. I am forced to conclude that this issue is in fact some
> uninvestigated (and unknown) SETI problem with an AMD Athlon XP processor
> working at greater than 2,000 MHz.

Well, finally, I think I have licked the problem, and it only took almost 6 months, and a new motherboard to figure it out. The new motherboard didn't really fix the issue. I think, at this time, that I needed to enable "FSB Spread Spectrum" in my BIOS settings. At this time, I have been running SETI/BOINC for about an hour with the spread spectrum setting in the BIOS, and there have been no apparent errors. I have had no more blue screens or any other apparent errors. I will post again after a good long time testing this theory. It may be a day, it may be a week, it may only be an hour, but, I hope this solves my problem, and it might help someone else who has (or had) similar problems.

---

MJ's Star Trek Portal
Motherboard: ASUS A8N-E with nVidia nForce4 Ultra Chipset
Processor: AMD Athlon 64 3000+, ~2,106 MHz, FSB: 468 MHz
RAM: 2,048 MB PC3200 DDR-SDRAM (Dual Channel)
OS: Windows XP Professional Edition Service Pack 2

> I will post again after a good long time testing this theory.

Hopefully it is sorted.

---

Grant
Darwin NT

> > > You are probably correct. Just my thoughts. If it was working
> before
> > it's
> > > probably not a temperature problem.
> > >
> > > I will try to keep an eye on this thread to see if you figure it
> out.
> > I'm
> > > curious.
> > >
> > > Good luck!
> > >
> > > HH
> > >
> >
> > Well, it seems I am able to clock it up, still on a 333 MHz FSB/RAM
> clock, but
> > I adjusted the bus clock multiplier to bring the CPU up to ~2,083 MHz
> (an
> > Athlon XP 2800+ Barton Core.) This is ever so slightly short of the
> supposed
> > limit of my Barton Core Athlon XP 3000+. It's about 17 MHz, or about 1%
> short
> > of the maximum rated processing power of the processor. If it holds, I
> can
> > accept it, since it is a negligible loss. This fix has been in place for
> about
> > 1 hour and 20 minutes. It should be noted that this fix was not working
> before
> > I found the BOINC v4.06 update. It would seem that the folks over at the
> SSL
> > may be working up to finally solving the issue.
> >
> > [EDIT - ADDITIONAL INFO]
> >
> > 2 hours and 43 minutes into the test with my bus clock multiplier pushed
> up
> > 1/2x and everything appears to be working properly. The CPU seems to be
> > handling BOINC just fine at 2,083 MHz.
> >
> > [EDIT #2 - MORE ADDITIONAL INFORMATION]
> >
> > About one hour after the "additional info" snippet above, my PC gave me
> the
> > same blue screen and error code, but at least it is much less common. I
> was
> > working on a web page in a HTML editor at the time (and, coincidentally,
> lost
> > the changes I made.) This has to be a memory leak, or a memory
> assignment
> > error.
> >
> > [EDIT #3 - 09-20-04 - EVEN MORE ADDITIONAL INFORMATION]
> >
> > Well, I have been running further tests in regard to this issue. The 400
> MHz
> > FSB appears to *not* be the issue, and thus it doesn't seem to be a
> > motherboard issue (though I could be wrong).
> >
> > I am underclocking my CPU 1/2x to 2,000 MHz on a 400 MHz FSB, and I don't
> see
> > any issues yet. Anything over 2,000 MHz on both the 400 and the 333 MHz
> FSB is
> > unstable (with SETI-BOINC and the SETI Classic) and causes blue screens
> with
> > all kinds of errors that do not happen when SETI@home is not running,
> even
> > when the CPU is clocked to where it belongs (2,100 MHz) on the 400 MHz
> FSB.
> > Windows and all of my other programs (aside from some non-essential and
> > unimportant games) do not seem to care at all when the CPU and FSB are
> clocked
> > up where they belong. I am forced to conclude that this issue is in fact
> some
> > uninvestigated (and unknown) SETI problem with an AMD Athlon XP
> processor
> > working at greater than 2,000 MHz.
>
> Well, finally, I think I have licked the problem, and it only took almost 6
> months, and a new motherboard to figure it out. The new motherboard didn't
> really fix the issue. I think, at this time, that I needed to enable "FSB
> Spread Spectrum" in my BIOS settings. At this time, I have been running
> SETI/BOINC for about an hour with the spread spectrum setting in the BIOS, and
> there have been no apparent errors. I have had no more blue screens or any
> other apparent errors. I will post again after a good long time testing this
> theory. It may be a day, it may be a week, it may only be an hour, but, I hope
> this solves my problem, and it might help someone else who has (or had)
> similar problems.
>
>
>

Well, it has been a couple of days testing the spread spectrum setting, and it apparently works. An alternative to this is to underclock the front side bus by 1 MHz, to 199 (398) MHz in the BIOS, and keep the front side bus spread spectrum off. This seems to work as well. Both settings reduce the CPU operating frequency by the same amount (in the neighborhood of 10-15 MHz), and everything appears stable, so it seems I have discovered my own fix to this problem. This reduces the overall processing power of the CPU by slightly less than 1% (actually not quite 1% -- about 9/10 of 1%), so overall, in all of the fooling around with this setup that I have done over the last 6 months, this appears to be the very best I can do to solve the issue, and preserve the most of the performance of my computer.

---

MJ's Star Trek Portal
Motherboard: ASUS A8N-E with nVidia nForce4 Ultra Chipset
Processor: AMD Athlon 64 3000+, ~2,106 MHz, FSB: 468 MHz
RAM: 2,048 MB PC3200 DDR-SDRAM (Dual Channel)
OS: Windows XP Professional Edition Service Pack 2

Well, finally, I think I have licked the problem, and it only took almost 6 months, and a new motherboard to figure it out. The new motherboard didn't really fix the issue. I think, at this time, that I needed to enable "FSB Spread Spectrum" in my BIOS settings. At this time, I have been running SETI/BOINC for about an hour with the spread spectrum setting in the BIOS, and there have been no apparent errors. I have had no more blue screens or any other apparent errors. I will post again after a good long time testing this theory. It may be a day, it may be a week, it may only be an hour, but, I hope this solves my problem, and it might help someone else who has (or had) similar problems.

At last, it seems I have found the answer to my problem. Apparently my Athlon XP 3000+ is defective. I am now using a AMD Sempron 3000+ that I have overclocked to the same MHz speed as the Athlon XP 3000+ works at normally, and have had no problems at all with BOINC/SETI@home, or anything else, for that matter. The overclock is 100 MHz above the native clock speed of the Sempron 3000+ and it seems happy with it. Even the temperatures are relatively good, not much more than a few degrees above the normal temperatures when the CPU is working at its native speed. I'm glad this is not an issue with SETI, but I am not happy that my Athlon XP 3000+ is defective. At least I am able to use a different processor.

---

MJ's Star Trek Portal
Motherboard: ASUS A8N-E with nVidia nForce4 Ultra Chipset
Processor: AMD Athlon 64 3000+, ~2,106 MHz, FSB: 468 MHz
RAM: 2,048 MB PC3200 DDR-SDRAM (Dual Channel)
OS: Windows XP Professional Edition Service Pack 2