I haven't gotten any of the new WUs yet, still working off an old set - but when I do, I'll try to remember to check this thread, and I'll get a list of the files I have, to paste here. Then we can see what is missing from yours, if anything.
Since I have processed an Albert unit, maybe I can shed a little more light on the subject. I have 3 new "Albert" files in my Einstein folder - 1/ albert_4.37_windows_intelx86 application, 2/albert_4.37_windows_intelx86 PDB File and 3/ config_S4R2a.cfg CFG File. I also still have all the usual einstein 4.79 files. Furthermore, after processing the one Albert unit on Xmas Eve, I am now back to processing Einstein 4.79 units. If you check my Albert result, you will see that it lists 4.37 as the application. All my other results list 4.79 or 0.18 (the Beta app) for the application. My point is that all your results list 4.79 as the app and so does not appear you have had an Albert unit yet. On the other hand, if you have the Albert app files (cited above) in your Einstein folder, I am totally wrong (and it won't be the first time).
I reset the project and it downloaded einstein_4.79_windows_intelx86.exe and einstein_4.79_windows_intelx86.pdb but made no difference and had the same error within 2 minutes of starting the result. I'll just have to wait for more work to be returned and see if it's a bad batch or if it just doesn't like something about my system.
For reference here is the list of files in my project folder.
My fault on assuming you had "Albert" - I saw "All-sky pulsar search" and missed the fact it wasn't "Improved All-sky pulsar search". The fact that this isn't the new app makes it more puzzling, as I've never seen this error before from anyone.
If you look at http://einsteinathome.org/workunit/3073131 you can see that you had the error, but the others in your workgroup did _not_ have the error. (The piece that was missing when Bernd responded.) This means it is not likely to be a problem with the WU file.
The "reset" you did eliminated the l1_0977.5__0977.x_0.1_T05_S4lD series and replaced it with the T06 series, but not a new "main WU file" - and you're still getting the error. (It seems to happen much quicker with this series... you have two from it, 39 seconds and 100 seconds, where the T05 was in the thousands.)
The files you have are the same ones I have, except I have additional, older "Config" files - I assume from previous work. (Well, obviously your WU file is l1xxx where mine is w1xxx.)
If you are not seriously overclocked, and your CPU is running at a reasonable temperature, the only thing _I_ can think of to try would be to make sure your computer is running as close to a "normal" OS as you can get (turn on the installer service, etc.), then detach from Einstein, reboot, and re-attach to Einstein. This will force it to replace the _entire_ WU file, as well as all "accessories". Run one result through and see if it is successful; if it is, then you can start tweaking your system back to how you like it, watching for when or if you start getting errors again.
I've been running this processor overclocked from 1.8Ghz to 2.232 Ghz for 3 years and for some reason it won't recognize the default anymore so returning the CPU back to normal status is no longer possible.
Back when I was running the old Version 3 Seti client it produced temps at 52C and now under BOINC they fluctuate between 36C and 44C depending on the project. The computer itself is located near a window with no case sides and a homemade air tube drawing nice and cold winter air directly to the CPU fan and another fan drawing air away from the chipset and cards. When I touch the heatsink it's cold and doesn't feel like it's even on so I don't believe overheating is a problem. This setup has kept it crunching even with the room temp reaching 45C for several days.
I'll have a closer look into my OS and any malware and script protection that may be active and enable or disable as needed then re-attach to the project and cross my fingers, this is going to take awhile.
I've been running this processor overclocked from 1.8Ghz to 2.232 Ghz for 3 years and for some reason it won't recognize the default anymore so returning the CPU back to normal status is no longer possible.
That doesn't seem "too extreme", but some projects seem to exercise different parts of the CPU than others, and your cooling seems, um, unconventional but adequate. :-) I know some people had to _underclock_ their systems to get reliable results on LHC, even when Prime95 and everything else said they were fine. I'm overclocked on mine (AMD64 3700) from 2.2GHz to 2.53GHz, no problems anywhere. Even if you can't go back to "stock", can you "turn it down a little"? I don't know if it'd help or not...
Pushing my overclocking even one step further will actually slow things down so this one is as far as I can push it but the thought of _underclocking_ gives me shivers. I've throttled it back to 2.07Ghz which I've found gives the best overall efficiency for it so all I need to to is wait for the server to give me a result which it's not doing because it says I've reached my daily quota of one. Time to grab some more caffeine and wait it out. :-)
Since you are sort of grasping at straws now anyway, you might want to see if changing to the Beta application makes any difference. I admit it is a long shot (because the Beta mainly fixes graphics problems) but the Beta also uses the new BOINC library (where Einstein 4.79 uses the library that was in effect when BOINC 4.19 was current). On the other hand, since you are not having problems with the other projects, changing the Einstein software might make some sense. (I used the Beta until "Albert" was announced. It is very stable.)
Few tips besides clocking it down:
1.Rip out every hardware to a minimum and test again.(minimize conflicts)
2.Don't know if Einstein need more memory than your other projects.
If it does,I can imagine it could hit some higher memlocation with an error that the others don't "reach".(But giving the same error each time seems odd.)
Run a memtest86+
3.Prime95 (tests cpu with small FFTs) have a builtin test which reports errors immidiatly when they occur.
This has happend to me not too long ago:
Everything works as usual (including Boinc) but some 3D games just dropped me
back to windows again with no messages and everything worked like before.
No crash,hangups or anything.I ran all kind of tests to nail the problem(including memtest).
Clocked it back to basics,newest drivers and all that I could think of.
Nothing worked. One day I decided to increase my memory.
Guess what happend when I pulled out my old memstick and added new ones...
No more kickbacks... so if you have a chance to replace parts for free..
And let a memtest be on for a longer time,I think I just ran it for an hour.
Here's my theory on what happened. The Windows screensaver is unreliable when overclocked because of the bigger gap with the hardware timing. I experienced this with a different video card than what I'm using now so I chalked it up to one of the hazards of overclocking. Now throw in the ATI card I'm running now which I've never had a problem with and the ATI driver problem other members have experienced and it's starting to become a little clearer. The problem seems to manifest itself when returning to do work so the problem may be when the project is setting everything back up and the video driver needs to communicate with the OS using a piece of driver code that is hyper-sensitive to timing issues which is returning a generic device not found. I'm just glad it's fixed.
Glad to hear it! Just remember that when the project switches over to Albert officially (and retires v4.79), you will need to switch to Albert. (Albert is basically the Beta app with new science algorithms. Or, alternatively, the Beta app is Albert but with v4.79 science algorithms.) Anyway, if you are are Beta user, you won't get Albert automatically. This message on the Message boards : Cafe Einstein : NEW: WINDOWS TEST APPLICATION FOR EINSTEIN@HOME thread explains how to switch.
RE: I haven't gotten any
)
Since I have processed an Albert unit, maybe I can shed a little more light on the subject. I have 3 new "Albert" files in my Einstein folder - 1/ albert_4.37_windows_intelx86 application, 2/albert_4.37_windows_intelx86 PDB File and 3/ config_S4R2a.cfg CFG File. I also still have all the usual einstein 4.79 files. Furthermore, after processing the one Albert unit on Xmas Eve, I am now back to processing Einstein 4.79 units. If you check my Albert result, you will see that it lists 4.37 as the application. All my other results list 4.79 or 0.18 (the Beta app) for the application. My point is that all your results list 4.79 as the app and so does not appear you have had an Albert unit yet. On the other hand, if you have the Albert app files (cited above) in your Einstein folder, I am totally wrong (and it won't be the first time).
I reset the project and it
)
I reset the project and it downloaded einstein_4.79_windows_intelx86.exe and einstein_4.79_windows_intelx86.pdb but made no difference and had the same error within 2 minutes of starting the result. I'll just have to wait for more work to be returned and see if it's a bad batch or if it just doesn't like something about my system.
For reference here is the list of files in my project folder.
einstein_4.79_windows_intelx86.exe
einstein_4.79_windows_intelx86.pdb
Config_L_S4lD
l1_0977.5
earth_05_09
sun_05_09
My fault on assuming you had
)
My fault on assuming you had "Albert" - I saw "All-sky pulsar search" and missed the fact it wasn't "Improved All-sky pulsar search". The fact that this isn't the new app makes it more puzzling, as I've never seen this error before from anyone.
If you look at http://einsteinathome.org/workunit/3073131 you can see that you had the error, but the others in your workgroup did _not_ have the error. (The piece that was missing when Bernd responded.) This means it is not likely to be a problem with the WU file.
The "reset" you did eliminated the l1_0977.5__0977.x_0.1_T05_S4lD series and replaced it with the T06 series, but not a new "main WU file" - and you're still getting the error. (It seems to happen much quicker with this series... you have two from it, 39 seconds and 100 seconds, where the T05 was in the thousands.)
The files you have are the same ones I have, except I have additional, older "Config" files - I assume from previous work. (Well, obviously your WU file is l1xxx where mine is w1xxx.)
If you are not seriously overclocked, and your CPU is running at a reasonable temperature, the only thing _I_ can think of to try would be to make sure your computer is running as close to a "normal" OS as you can get (turn on the installer service, etc.), then detach from Einstein, reboot, and re-attach to Einstein. This will force it to replace the _entire_ WU file, as well as all "accessories". Run one result through and see if it is successful; if it is, then you can start tweaking your system back to how you like it, watching for when or if you start getting errors again.
I've been running this
)
I've been running this processor overclocked from 1.8Ghz to 2.232 Ghz for 3 years and for some reason it won't recognize the default anymore so returning the CPU back to normal status is no longer possible.
Back when I was running the old Version 3 Seti client it produced temps at 52C and now under BOINC they fluctuate between 36C and 44C depending on the project. The computer itself is located near a window with no case sides and a homemade air tube drawing nice and cold winter air directly to the CPU fan and another fan drawing air away from the chipset and cards. When I touch the heatsink it's cold and doesn't feel like it's even on so I don't believe overheating is a problem. This setup has kept it crunching even with the room temp reaching 45C for several days.
I'll have a closer look into my OS and any malware and script protection that may be active and enable or disable as needed then re-attach to the project and cross my fingers, this is going to take awhile.
RE: I've been running this
)
That doesn't seem "too extreme", but some projects seem to exercise different parts of the CPU than others, and your cooling seems, um, unconventional but adequate. :-) I know some people had to _underclock_ their systems to get reliable results on LHC, even when Prime95 and everything else said they were fine. I'm overclocked on mine (AMD64 3700) from 2.2GHz to 2.53GHz, no problems anywhere. Even if you can't go back to "stock", can you "turn it down a little"? I don't know if it'd help or not...
Pushing my overclocking even
)
Pushing my overclocking even one step further will actually slow things down so this one is as far as I can push it but the thought of _underclocking_ gives me shivers. I've throttled it back to 2.07Ghz which I've found gives the best overall efficiency for it so all I need to to is wait for the server to give me a result which it's not doing because it says I've reached my daily quota of one. Time to grab some more caffeine and wait it out. :-)
Since you are sort of
)
Since you are sort of grasping at straws now anyway, you might want to see if changing to the Beta application makes any difference. I admit it is a long shot (because the Beta mainly fixes graphics problems) but the Beta also uses the new BOINC library (where Einstein 4.79 uses the library that was in effect when BOINC 4.19 was current). On the other hand, since you are not having problems with the other projects, changing the Einstein software might make some sense. (I used the Beta until "Albert" was announced. It is very stable.)
I wonder if that device(is
)
I wonder if that device(is not ready) is real?
Few tips besides clocking it down:
1.Rip out every hardware to a minimum and test again.(minimize conflicts)
2.Don't know if Einstein need more memory than your other projects.
If it does,I can imagine it could hit some higher memlocation with an error that the others don't "reach".(But giving the same error each time seems odd.)
Run a memtest86+
3.Prime95 (tests cpu with small FFTs) have a builtin test which reports errors immidiatly when they occur.
This has happend to me not too long ago:
Everything works as usual (including Boinc) but some 3D games just dropped me
back to windows again with no messages and everything worked like before.
No crash,hangups or anything.I ran all kind of tests to nail the problem(including memtest).
Clocked it back to basics,newest drivers and all that I could think of.
Nothing worked. One day I decided to increase my memory.
Guess what happend when I pulled out my old memstick and added new ones...
No more kickbacks... so if you have a chance to replace parts for free..
And let a memtest be on for a longer time,I think I just ran it for an hour.
Hope you find your strange problem.
Strange problem fixed, thanks
)
Strange problem fixed, thanks for all the help. Beta version and de-clocking did the trick. :-)
5.2.13
Fstats.Ha: bytecount 983521 checksum 47225568
Fstats.Hb: bytecount 1076659 checksum 51698355
Here's my theory on what happened. The Windows screensaver is unreliable when overclocked because of the bigger gap with the hardware timing. I experienced this with a different video card than what I'm using now so I chalked it up to one of the hazards of overclocking. Now throw in the ATI card I'm running now which I've never had a problem with and the ATI driver problem other members have experienced and it's starting to become a little clearer. The problem seems to manifest itself when returning to do work so the problem may be when the project is setting everything back up and the video driver needs to communicate with the OS using a piece of driver code that is hyper-sensitive to timing issues which is returning a generic device not found. I'm just glad it's fixed.
Glad to hear it! Just
)
Glad to hear it! Just remember that when the project switches over to Albert officially (and retires v4.79), you will need to switch to Albert. (Albert is basically the Beta app with new science algorithms. Or, alternatively, the Beta app is Albert but with v4.79 science algorithms.) Anyway, if you are are Beta user, you won't get Albert automatically. This message on the Message boards : Cafe Einstein : NEW: WINDOWS TEST APPLICATION FOR EINSTEIN@HOME thread explains how to switch.