Windows S5R4 App 6.10 available for Beta Test

MarkJ
MarkJ
Joined: 28 Feb 08
Posts: 437
Credit: 137651148
RAC: 17144

RE: @Archae86, Richard

Message 89191 in response to message 89182

Quote:
@Archae86, Richard Haselgrove, and/or any other CUDA-capable people: If you're reading this and you participate with SETI and the CUDA app, can any of you test for this phenomenon with your systems? My poor (and old) 6800GT is not CUDA-compliant...

I tried it while running GPUGRID. BOINC 6.5.0 as reported above. Have given up with the Seti-cuda app until they fix it.

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: RE: @Archae86,

Message 89192 in response to message 89191

Quote:
Quote:
@Archae86, Richard Haselgrove, and/or any other CUDA-capable people: If you're reading this and you participate with SETI and the CUDA app, can any of you test for this phenomenon with your systems? My poor (and old) 6800GT is not CUDA-compliant...

I tried it while running GPUGRID. BOINC 6.5.0 as reported above. Have given up with the Seti-cuda app until they fix it.

OK. Not exactly speaking for Bernd, but since he made the app official even after seeing this discussion, I would tend to think that means that this particular crash is not common. I'd then have to agree with Bikeman, that Sabroe's issue is most likely from overclocking and/or overheating...

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 31201934
RAC: 346

FYI: I had been running v6.10

FYI: I had been running v6.10 as a Beta app but, when it was made official, I deleted my app_info files and restarted BOINC on both my hosts. (I did not have any WU's in progress or in cache at the time.) After that, both hosts downloaded WU's and both WU's failed immediately:

http://einsteinathome.org/task/114104311
http://einsteinathome.org/task/114285523

This message appears to be the reason:
1/12/2009 3:41:43 PM|Einstein@Home|[error] Signature verification failed for einstein_S5R4_6.10_windows_intelx86_2.exe

And, a "Reset project" seems to have fixed the problem.

rroonnaalldd
rroonnaalldd
Joined: 12 Dec 05
Posts: 116
Credit: 537221
RAC: 0

Had the same problem on one

Had the same problem on one host.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5846
Credit: 109976003603
RAC: 29515666

When you delete the

When you delete the app_info.xml file you should also delete all the 6.10 beta test apps as well. I'm not fully sure but I don't think BOINC likes to find the app already there when you restart without an app_info.xml for the first time. If you delete all the app files, BOINC will download what it needs and all will be well.

I believe that BOINC probably doesn't keep checksums in the state file for apps under the control of app_info.xml. They are certainly needed (and checked) for stock apps that are downloaded. So when there is no longer an app_info.xml, BOINC will calculate the checksums for the 6.10 app files it finds and then compare those values with the non-existent values in the state file. Of course this is always going to fail.

I haven't checked this out but something like this is what I think is happening. A possible way to check would be to edit the state file and insert all the missing checksums for the various app files listed there before you restart BOINC after deleting the app_info.xml file. Of course, it's much simpler to just delete all the beta app files and let BOINC download afresh.

Cheers,
Gary.

rbpeake
rbpeake
Joined: 18 Jan 05
Posts: 266
Credit: 979369747
RAC: 639533

RE: When you delete the

Message 89196 in response to message 89195

Quote:

When you delete the app_info.xml file you should also delete all the 6.10 beta test apps as well. I'm not fully sure but I don't think BOINC likes to find the app already there when you restart without an app_info.xml for the first time. If you delete all the app files, BOINC will download what it needs and all will be well.

I believe that BOINC probably doesn't keep checksums in the state file for apps under the control of app_info.xml. They are certainly needed (and checked) for stock apps that are downloaded. So when there is no longer an app_info.xml, BOINC will calculate the checksums for the 6.10 app files it finds and then compare those values with the non-existent values in the state file. Of course this is always going to fail.

I haven't checked this out but something like this is what I think is happening. A possible way to check would be to edit the state file and insert all the missing checksums for the various app files listed there before you restart BOINC after deleting the app_info.xml file. Of course, it's much simpler to just delete all the beta app files and let BOINC download afresh.


A less elegant way that I find works is to detach from EAH entirely, and then attach to it again as a new project. All the files will be new at that point. Admittedly this is using a hammer on a fly, but it is quick and it works! ;)

Stick
Stick
Joined: 24 Feb 05
Posts: 790
Credit: 31201934
RAC: 346

RE: A less elegant way that

Message 89197 in response to message 89196

Quote:
A less elegant way that I find works is to detach from EAH entirely, and then attach to it again as a new project. All the files will be new at that point. Admittedly this is using a hammer on a fly, but it is quick and it works! ;)

The "Reset project" button (as mentioned in my original message) is, in my opinion, the easiest fix for the problem.

samuel7
samuel7
Joined: 16 Feb 05
Posts: 34
Credit: 1579363
RAC: 0

RE: When you delete the

Message 89198 in response to message 89195

Quote:

When you delete the app_info.xml file you should also delete all the 6.10 beta test apps as well. I'm not fully sure but I don't think BOINC likes to find the app already there when you restart without an app_info.xml for the first time. If you delete all the app files, BOINC will download what it needs and all will be well.

I believe that BOINC probably doesn't keep checksums in the state file for apps under the control of app_info.xml. They are certainly needed (and checked) for stock apps that are downloaded. So when there is no longer an app_info.xml, BOINC will calculate the checksums for the 6.10 app files it finds and then compare those values with the non-existent values in the state file. Of course this is always going to fail.

I haven't checked this out but something like this is what I think is happening. A possible way to check would be to edit the state file and insert all the missing checksums for the various app files listed there before you restart BOINC after deleting the app_info.xml file. Of course, it's much simpler to just delete all the beta app files and let BOINC download afresh.

I switched to the stock 6.10 on my Intel laptop and the Quad by just moving the app_info to a different folder. All executables were left as they were. BOINC 6.4.5 on both machines.
A view of the client_state before:

    einstein_S5R4
    einstein_S5R4

einstein_S5R4_6.10_windows_intelx86.exe
0.000000
0.000000
1


...and after:

    einstein_S5R4
    Hierarchical all-sky pulsar search

einstein_S5R4_6.10_windows_intelx86.exe
0.000000
0.000000
1



681b8f733a3ef1ee6742ce64274bde43d47d198bd39469aa9738c475cf38b2be
63cc6d1c2d312958f767bcfa1e6348582942ba73c8298a81ed1f925c70b0f8f9
ff3a0cd464bd89c8487c307d925b8dbf44b83431222ef79c30753906dd599c23
7af3a453c532ad8d09d4d0217757450cd887ed86fe2dd2a297d1845196da02b2
.

http://einstein.aei.mpg.de/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.astro.gla.ac.uk/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.phys.uwm.edu/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.ligo.caltech.edu/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.aei.mpg.de/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.astro.gla.ac.uk/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.phys.uwm.edu/download/einstein_S5R4_6.10_windows_intelx86.exe
http://einstein.ligo.caltech.edu/download/einstein_S5R4_6.10_windows_intelx86.exe

Tasks have validated since the change. So it can work if you want to spare bandwidth.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 689321500
RAC: 217635

RE: Tasks have validated

Message 89199 in response to message 89198

Quote:

Tasks have validated since the change. So it can work if you want to spare bandwidth.

I guess the trick might be that there were two different versions of the 6.10 app, the now official one would then be the one that got uploaded later to the server. If you have that one, it's signature will match the expected one. The original 6.10 beta app had a different checksum and will not match.

CU
Bikeman

samuel7
samuel7
Joined: 16 Feb 05
Posts: 34
Credit: 1579363
RAC: 0

RE: RE: Tasks have

Message 89200 in response to message 89199

Quote:
Quote:

Tasks have validated since the change. So it can work if you want to spare bandwidth.

I guess the trick might be that there were two different versions of the 6.10 app, the now official one would then be the one that got uploaded later to the server. If you have that one, it's signature will match the expected one. The original 6.10 beta app had a different checksum and will not match.

CU
Bikeman

Yes, that's probably it. The point is then that as long you have the exact same files, the checksums are verified automatically upon the first scheduler request after removing app_info.xml.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.