I installed a seti on XP Pro at work (with permission). This copy has now done 7 units and they all have this error. I don't know what it means or how to fix it. If anyone has any clues and can email me details I'd appreciate it. I am not terribly familiar with XP Pro. It was installed so any user could run it though this particular CPU always logs in as admin....

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

I installed a seti on XP Pro at work (with permission). This copy has now done 7 units and they all have this error. I don't know what it means or how to fix it. If anyone has any clues and can email me details I'd appreciate it. I am not terribly familiar with XP Pro. It was installed so any user could run it though this particular CPU always logs in as admin....

http://setiathome.berkeley.edu/results.php?hostid=2325831
(That's the computer link so you can study the result errors)

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

You're trying to access an address in memory, but your XP has defined it as
an address that you can't access directly..

I suspect your trying to read a hardware register with an I/O instruction.
I/O access implements instructions at the system call level which is usually not available to unpriviledged accounts.

I suspect it is a account priviledge problem. Can you save to your hard drive? Alot of admins, myself included, deny write access on the local drive, and only allow write over to the private user area on a remote NFS.

---

I'm not sure that this is a memory access issue.

The error refers to executing a priviledged instruction at the given address, not a failure to access that address.

Also, I am unconvinced that it is a disk (or other hardware) write problem, as (a) it is normal for compilers to include run-time error handling for this type of error and produce some sort of error message; (b) If such an error handler isn't present, and the practice of write-prtecting a disk (or other hardware) is common, I would have expected the error to be reported by others before.

A Google search for 'Privileged Instruction (0x00000096)' reveals quit a few instances of this error, but no obvious solution in the few I looked at.

BTW, in this case Privileged doesn't refer to the priviledges of an Admin user, but to the mode the CPU is operating in - which should be USER mode. The message suggests that an instruction that can only be executed when the CPU is in KERNEL mode has been attempted in a lower mode e.g. User.

Since the CPU time used per WU before the error varies seemingly from a couple of minues to 10 minutes, the possibility some kind of System Instability, CPU or memory fault cannot be ignored.

/EDIT
Just noticed that 1 workunit did actually run to completion, which lends credence to the notion that the error is transient, not inherrent and stability issues become more likely.

---

What other programs are running at the same time?
Is Boinc set to leave wu in memory.
It's possible that another program overwrites the wu that is temp stored in memory.

---

Fred
BOINC Alpha, BOINC Beta, LHC Alpha, Einstein Alpha

I'm not sure that this is a memory access issue.

The error refers to executing a priviledged instruction at the given address, not a failure to access that address.

Also, I am unconvinced that it is a disk (or other hardware) write problem, as (a) it is normal for compilers to include run-time error handling for this type of error and produce some sort of error message; (b) If such an error handler isn't present, and the practice of write-prtecting a disk (or other hardware) is common, I would have expected the error to be reported by others before.

A Google search for 'Privileged Instruction (0x00000096)' reveals quit a few instances of this error, but no obvious solution in the few I looked at.

BTW, in this case Privileged doesn't refer to the priviledges of an Admin user, but to the mode the CPU is operating in - which should be USER mode. The message suggests that an instruction that can only be executed when the CPU is in KERNEL mode has been attempted in a lower mode e.g. User.

Since the CPU time used per WU before the error varies seemingly from a couple of minues to 10 minutes, the possibility some kind of System Instability, CPU or memory fault cannot be ignored.

/EDIT
Just noticed that 1 workunit did actually run to completion, which lends credence to the notion that the error is transient, not inherrent and stability issues become more likely.

True...my idea was just a stab in the dark. hmm...requires more thought.

---

What other programs are running at the same time?
Is Boinc set to leave wu in memory.
It's possible that another program overwrites the wu that is temp stored in memory.

The computer is running one and only one application besides Boinc. That's UPS WORLDSHIP 8.0

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

What other programs are running at the same time?
Is Boinc set to leave wu in memory.
It's possible that another program overwrites the wu that is temp stored in memory.

How do you select whether to leave WU in memory on a particular machine? I've never seen this choice.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

You can go to your "General Preferences, you can set several different venues (home, work, school, and default). You can then go to "your computers" and have that puter changed from what venue it's assigned, to the one you changed. Note: each host must "call home" to see the change.

tony

---

You can go to your "General Preferences, you can set several different venues (home, work, school, and default). You can then go to "your computers" and have that puter changed from what venue it's assigned, to the one you changed. Note: each host must "call home" to see the change.

tony

It's NOT set to keep in my memory anyway so that's not it.

This is SO frustrating. Someone must have an inkling of this stupidity. I can't see -- some of my units actually FINISH and then get a client error.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

I've picked this result at random from your results:

286722931 2325831 7 Apr 2006 18:21:52 UTC 7 Apr 2006 18:32:07 UTC Over Client error Computing 452.36 1.05 ---
286722932 1303347 7 Apr 2006 18:21:54 UTC 8 Apr 2006 3:05:13 UTC Over Success Done 18,939.64 26.37 31.43
286722933 2046508 7 Apr 2006 18:21:52 UTC 9 Apr 2006 18:06:58 UTC Over Success Done 33,136.31 66.74 31.43
286722934 2211813 7 Apr 2006 18:21:52 UTC 21 Apr 2006 18:21:52 UTC In Progress Unknown New --- --- ---
287053904 1286574 8 Apr 2006 1:32:56 UTC 10 Apr 2006 4:18:33 UTC Over Success Done 9,083.81 31.43 31.43

Notice every other user has crunched it to completion successfully and their returned result validated. From this I must conclude it's something about your machine which is "flaky".

Try running memtest86+ for an hour or so, then try Prime95 for a hour or three, then try Speedfan to check on temp issues.

edit=if all this fails you might want to check the PSU voltages.

NOTE: all the above linked software is freeware.

hope you find the problem

tony

---

Try running memtest86+ for an hour or so, then try Prime95 for a hour or three, then try Speedfan to check on temp issues.

I TRIED. There are no reported problems at all. I am more and more thinking this has got to be some sort of weird permissions error where it's the timing of the read/write. This is being done under 'administrator' with full priviledges.

I give up. It's really really really frustrating though. I'm leaving the system running just on principle -- even if only a few results make it through it's better than none.

This pattern continues and I've made sure Windoze Firewall is disabled BTW which causes all sorts of hell with Seti.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

I TRIED. There are no reported problems at all. I am more and more thinking this has got to be some sort of weird permissions error where it's the timing of the read/write. This is being done under 'administrator' with full priviledges.

As I already explained earlier, in this case 'Privileged' has nothing to do with who or what you are logged on as.

A "privileged instruction" is a CPU instruction that is only allowed to execute when the processor is in KERNEL (and/or maybe Executive, Supervisor) Mode.

Normal user programs run in USER Mode. KERNEL (and the others - which windows don't use) Mode is reserved for Operating System code, driver code and the like.

For example, the x86 instruction set has a HALT instruction, which as it's name suggests, Halts the processor dead - the whole system stops normal operation.

Imagine if just any old program were allowed to execute HALT... Systems would go stopping all over the place. So the HALT instruction is 'Privileged' and is not allowed to happen except in the operating system - in Kernel Mode ("You May Now Turn Off Your Computer"!)

When code is running in User Mode and a HALT instruction is reached, the processor rightly generates an exception and prevents it executing and taking the syste down.

Neither BOINC or the SETI applications will intentionally ever try to execute a priviledged instruction, so your system is falsly thinking it has reached one.

As others have said, you most likely have a bit of flakey or overheating hardware, could be memory, processor (esp. the cache), and/or power supply problems. A good few hours of MEMTEST86+ and Prime95 (google them) should reveal the fault.

What is is NOT is any kind of permissions error to do with timing of read/writes.

---

Try running memtest86+ for an hour or so, then try Prime95 for a hour or three, then try Speedfan to check on temp issues.

I TRIED. There are no reported problems at all. I am more and more thinking this has got to be some sort of weird permissions error where it's the timing of the read/write. This is being done under 'administrator' with full priviledges.

I give up. It's really really really frustrating though. I'm leaving the system running just on principle -- even if only a few results make it through it's better than none.

This pattern continues and I've made sure Windoze Firewall is disabled BTW which causes all sorts of hell with Seti.

Looks like you are having the same problem on 2 machines? Computer ID are 2325831 and 1382587 a 2.6 GHZ and a 2.4 GHZ machine. Confirm if these are the same machine or if you really do have the same problem on 2 machines.

This surely has to be some kind of problem on your machines related to heat or system timings. I can't believe you would leave them running just to get the few random credits.

---

Boinc....Boinc....Boinc....Boinc....

Looks like you are having the same problem on 2 machines? Computer ID are 2325831 and 1382587 a 2.6 GHZ and a 2.4 GHZ machine. Confirm if these are the same machine or if you really do have the same problem on 2 machines.

This surely has to be some kind of problem on your machines related to heat or system timings. I can't believe you would leave them running just to get the few random credits.

How interesting.

The one I've been complaining about (UPS Computer) has to run 24/7 because it polls UPS for updates and all sorts of crap. That has XP pro. Why did it pass all those tests the other guy had me run? This one is in the warehouse and it certainly gets hot out there.

The one you mentioned I didn't even notice was having troubles -- and that must be a new development. That's an old antique machine -- but it only runs 7am to 5pm weekdays at worse case (it's our e-mail distribution system not the actual server). This has XP home on it. This one is in a properly air-conditioned office.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

It seems you have two puters returning junk, your celeron 2.4 and 2.6. Your others seem fine. The reason it was believed to by your host only was:

1)because you're the only one out of a few hundred thousand reporting this error, if it were boinc/seti related then usually we'd see more reports of this.
2) It's only your hosts returning errors where everyone else crunching the same wu are doing so successfully.

However, now we see two of your host (and only your hosts) returning computing errors. It's possible both puters are failing at the same time, but not as likely as if you were only having trouble with one. So, we need to figure out why it's just your host/s.

What boinc client, and which Seti application are you using on each of those? Is it possible that the application you've installed (if optimized) isn't suited to "celeron" math instructions?

However, now we see two of your host (and only your hosts) returning computing errors. It's possible both puters are failing at the same time, but not as likely as if you were only having trouble with one. So, we need to figure out why it's just your hosts.

What boinc client, and which Seti application are you using on each of those? Is it possible that the application you've installed (if optimized) isn't suited to "celeron" math instructions?

Sorry for the delay. Because I don't want to waste your time, I wanted to check everything again and make sure I'm not giving you bad information. I ran (again) the memtest program and the prime program without problems on BOTH machines -- after a false start due to idiot user error (me).

I am running 5.2.13 -- which I added to the Mail Server Client when I installed the UPS version. Oddly that's (apparently) when it started failing. So it seems to me that 5.2.13 isn't working right with the Celeron chip. I don't understand, though, why I'd be the only one. Further notice that a number of results actually COMPLETE (status DONE) and then STILL fail. This is what perplexes me.

But Mail Server (MS) runs XP Home and UPS System (UPS) runs XP Pro. The machines are only on the network to the router so they can talk to the outside world and use the printer but are not networked with each other in any way.

I check my results from time to time and never had this problem with the MS client before. But when I added the UPS machine in, I also upgraded to 5.2.13 on the MS client and didn't check results since that time. (W've been busy at work).

I tried to look to see if there were any glitches reported with Celerons and 5.2.13 but didn't see anything, so your advice would be appreciated.

I was going to de-install and re-install but inuitively this just doesn't make sense. Two machines don't both go bad on the same day. The UPS unit is almost new (4 mo) and the MS is very old. The memory is checking out fine. They don't even use the same OS so no matter what I do, I end up pointing back to the actual client.

Looking under messages on the MS machine there are NO ERRROS being reported in the log. The UPS does report computational errors and indicate the result is aborted.

Do you think a downgrade is in order? My fear is at some point the older versions are cancelled out. Then, I'll end up switching to the new version and have the same problem.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

Are you using any optimized seti applications?

Are you using any optimized seti applications?

Nope. Just the one download and that's it. I don't even know what your question means. I'm not an advanced user. But I did get permission from work to install this client as screen saver on all our systems.

---

Visit my website where you can see my Seti page, my blog, and anything else of interest. Write me at darsys at aol dot com if you want. My blog is at http;//www.darsys.net/quagmire.html

I did get permission from work to install this client as screen saver on all our systems.

- exit code -1073741819 (0xc0000005) is concurrent with problems with the videocard or videochip on the motherboard.

Is it OpenGL capable? Are the drivers up to date? DirectX installed?
Have you tried running it without the screensaver, to see if the errors then go away?

---