Apparently the Linux SSE(2) Apps were built from a wrong code version that has a bug in the checkpointing. Windows SSE(2) and the generic (non-SSE) Linux App are not affected. For the time being I 'deprecated' the Linux SSE(2) Apps, meaning that Linux clients should process workunits they get now with the generic App. Processing will be slower, but checkpointing should work. I'll build and publish new Linux Apps on Tuesday.
[edit] I don't think tasks done with the wrong checkpointing will validate. I'd recommend to abort them [/edit]
I have aborted my unit and downloaded the non optimized app. Now it is at 14.613% after 4:39:53 hours.
Tullio
Yup, the unoptiomized app will take between 2x and 3x as much time, Akos et al did make a big difference :-). It's only for a short time until Bernd will compile the new app, probably tomorrow.
Apparently the Linux SSE(2) Apps were built from a wrong code version that has a bug in the checkpointing. Windows SSE(2) and the generic (non-SSE) Linux App are not affected. For the time being I 'deprecated' the Linux SSE(2) Apps, meaning that Linux clients should process workunits they get now with the generic App. Processing will be slower, but checkpointing should work. I'll build and publish new Linux Apps on Tuesday.
[edit] I don't think tasks done with the wrong checkpointing will validate. I'd recommend to abort them [/edit]
Sorry for the inconvenience.
BM
I stop my computer this night and all the day for private causes.
When I restart it I can see the apps are restart normaly too.
No lost data
If it's a ABP2 task: it was not affected by the problem in the first place.
If it's a S5GC1 task: maybe your BOINC client has already downloaded the "unoptimized" version of the app, whcih also is ok.
CU
HB
It's a S5G1 task
I have noticed that even if I have an "unoptimised" app, when I stop and restart my computer, the estimated time and the final calcul time increase.
I have a ABP2 task too, and not this type of problem
S5GC1 Apps for SSE and SSE2 are shipped now. To add some confusion (well, actually to avoid trouble for the BOINC Clients) the new SSE2 App carries version number 1.03, although it has been built from the same code as the 1.02 Apps in use.
S5GC1 Apps for SSE and SSE2 are shipped now. To add some confusion (well, actually to avoid trouble for the BOINC Clients) the new SSE2 App carries version number 1.03, although it has been built from the same code as the 1.02 Apps in use.
BM
It seems ok with #1 1.03 (S5GCESSE2) version now
Thanks for the the developers job
I have left applications in
)
I have left applications in memory while suspended and it is progressing OK on my SuSE Linux 11.1 , BOINC 6.10.44.
Tullio
Thanks for reporting
)
Thanks for reporting this.
Apparently the Linux SSE(2) Apps were built from a wrong code version that has a bug in the checkpointing. Windows SSE(2) and the generic (non-SSE) Linux App are not affected. For the time being I 'deprecated' the Linux SSE(2) Apps, meaning that Linux clients should process workunits they get now with the generic App. Processing will be slower, but checkpointing should work. I'll build and publish new Linux Apps on Tuesday.
[edit] I don't think tasks done with the wrong checkpointing will validate. I'd recommend to abort them [/edit]
Sorry for the inconvenience.
BM
BM
RE: [edit] I don't think
)
At least this one was valid http://einsteinathome.org/task/174185558
Two others are waiting for wingman.
errare humanum est. ;)
cu,
Michael
I have aborted my unit and
)
I have aborted my unit and downloaded the non optimized app. Now it is at 14.613% after 4:39:53 hours.
Tullio
RE: I have aborted my unit
)
Yup, the unoptiomized app will take between 2x and 3x as much time, Akos et al did make a big difference :-). It's only for a short time until Bernd will compile the new app, probably tomorrow.
CU
HB
RE: Thanks for reporting
)
I stop my computer this night and all the day for private causes.
When I restart it I can see the apps are restart normaly too.
No lost data
End of the problems?
just a poet
RE: End of the
)
If it's a ABP2 task: it was not affected by the problem in the first place.
If it's a S5GC1 task: maybe your BOINC client has already downloaded the "unoptimized" version of the app, whcih also is ok.
CU
HB
RE: RE: End of the
)
It's a S5G1 task
I have noticed that even if I have an "unoptimised" app, when I stop and restart my computer, the estimated time and the final calcul time increase.
I have a ABP2 task too, and not this type of problem
just a poet
S5GC1 Apps for SSE and SSE2
)
S5GC1 Apps for SSE and SSE2 are shipped now. To add some confusion (well, actually to avoid trouble for the BOINC Clients) the new SSE2 App carries version number 1.03, although it has been built from the same code as the 1.02 Apps in use.
BM
BM
RE: S5GC1 Apps for SSE and
)
It seems ok with #1 1.03 (S5GCESSE2) version now
Thanks for the the developers job
regards
just a poet