My concern is that such people wanting CPU work for the new GW run only, may get a surprise to find both FGRPB1 and BRP6-cuda55 tasks (if they have an NVIDIA GPU) which they weren't expecting. If they had deliberately selected only O1AS20-100T, they might be st least mystified and possibly upset by getting all three.
Gary, while I have myself observed that the "Run test applications?" setting (it does not actually specify beta) when activated appears to allow both CPU test applications if one has enabled CPU work at all, regardless of whether one has enabled the specific type, if one has the CPU enable turned off, but the GPU enable turned on, then one gets neither of the test CPU applications, but just gets the CUDA55 Parkes beta work.
I've not tested, but rather suspect, that the flip-flop case is that if one has GPU work disabled, and CPU work enabled, one will get both of the current CPU beta work types, but NOT any GPU work, beta or not.
Why the code is written to honor the CPU and GPU settings, but ignore the specific application enables, when "run test applications?" is enabled is not something I'll guess at. Perhaps even it is somewhat accidental.
... if one has GPU work disabled, and CPU work enabled, one will get both of the current CPU beta work types, but NOT any GPU work, beta or not.
I believe that is correct, as long as using the specific GPU has been disabled. I'm thinking of a somewhat different situation, one that I believe could be a quite common case. A volunteer whose GPU is working at some other project hears about the GW detection and decides to get involved here just for the new GW run. Unless they understand what is going on here, I suspect they will just add the O1AST run and expect to get CPU tasks for their (perhaps) under-engaged CPU cores. They don't get tasks immediately and when they ask they get told they need to turn on test apps. Because they haven't selected a GPU run, they probably wont see the need to turn off their specific GPU type here. People get quite frustrated when unexpected things happen and I'm just looking at how to make things as smooth as possible for any influx of volunteers.
Why the code is written to honor the CPU and GPU settings, but ignore the specific application enables, when "run test applications?" is enabled is not something I'll guess at. Perhaps even it is somewhat accidental.
This has to to with how Boinc and prefs work. When selection a resource to use in prefs, ie GPU or CPU, that gets communicated to Boinc on the participants computer and then Boinc will only ask for work for the selected resources. The next step is how Boinc actually asks for work, it only asks for an amount of work for a particular resource, ie seconds of work, not specifying what applications/science runs that work should be for, that gets decided on the server via prefs and work availability.
Will the GPU version come after the beta phase or?
Certainly not during that "tuning" run.
We noticed problems with the previous Windows application versions and the setup of the workunts, thus the app is still in "Beta" test.
We fixed the workunits and built and published a new Windows app version. This has no AVX support, but at least it should run at all. We'll continue to work on that.
I don't know if this is useful for the project or if it's just my machine having a brain fart.
work unit h1_0021.45_O1C01Cl1In1__O1AS20-100T_21.5Hz_171_3 did this after over 12 hrs of processing.
Quote:
2016-02-12 15:05:13.8710 (9408) [normal]: Finished main analysis.
2016-02-12 15:05:13.8710 (9408) [normal]: Recalculating statistics for the final toplist...
2016-02-12 15:09:32.6940 (9408) [normal]: Finished recalculating toplist statistics.
2016-02-12 15:09:32.6940 (9408) [debug]: Writing output ... toplist2 ... toplist3 ... done.
FPU status flags: PRECISION
2016-02-12 15:09:33.8562 (9408) [normal]: done. calling boinc_finish(0).
15:09:33 (9408): called boinc_finish
This would appear to be one of the three types of issues that Christian lists here. In particular, it's the third in the list about missing result files. Your snippet above shows two result files that are missing (-161). If you look at the various reports following Christian's post you will see that these (and other) errors are happening on other machines and is certainly not an issue with your host.
Unfortunately, until there is more feedback from Christian, this looks like something that may not be solved until Monday. It would probably be useful to Christian for any further reports of issues like this to be added to the above thread in the Problems section rather than the News thread.
Sat Feb 13 02:36:32 2016 | Einstein@Home | Computation for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 finished
Sat Feb 13 02:36:32 2016 | Einstein@Home | Output file h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_1 for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 absent
Sat Feb 13 02:36:32 2016 | Einstein@Home | Output file h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_2 for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 absent
Sat Feb 13 02:36:32 2016 | Einstein@Home | Starting task LATeah0158E_80.0_528_-2e-10_1
Sat Feb 13 02:36:34 2016 | Einstein@Home | Started upload of h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_0
Sat Feb 13 02:36:43 2016 | Einstein@Home | Finished upload of h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_0
Sat Feb 13 02:38:09 2016 | Einstein@Home | Sending scheduler request: To report completed tasks.
Gary Roberts wrote:My concern
)
Gary, while I have myself observed that the "Run test applications?" setting (it does not actually specify beta) when activated appears to allow both CPU test applications if one has enabled CPU work at all, regardless of whether one has enabled the specific type, if one has the CPU enable turned off, but the GPU enable turned on, then one gets neither of the test CPU applications, but just gets the CUDA55 Parkes beta work.
I've not tested, but rather suspect, that the flip-flop case is that if one has GPU work disabled, and CPU work enabled, one will get both of the current CPU beta work types, but NOT any GPU work, beta or not.
Why the code is written to honor the CPU and GPU settings, but ignore the specific application enables, when "run test applications?" is enabled is not something I'll guess at. Perhaps even it is somewhat accidental.
RE: ... if one has GPU work
)
I believe that is correct, as long as using the specific GPU has been disabled. I'm thinking of a somewhat different situation, one that I believe could be a quite common case. A volunteer whose GPU is working at some other project hears about the GW detection and decides to get involved here just for the new GW run. Unless they understand what is going on here, I suspect they will just add the O1AST run and expect to get CPU tasks for their (perhaps) under-engaged CPU cores. They don't get tasks immediately and when they ask they get told they need to turn on test apps. Because they haven't selected a GPU run, they probably wont see the need to turn off their specific GPU type here. People get quite frustrated when unexpected things happen and I'm just looking at how to make things as smooth as possible for any influx of volunteers.
Cheers,
Gary.
RE: Why the code is written
)
This has to to with how Boinc and prefs work. When selection a resource to use in prefs, ie GPU or CPU, that gets communicated to Boinc on the participants computer and then Boinc will only ask for work for the selected resources. The next step is how Boinc actually asks for work, it only asks for an amount of work for a particular resource, ie seconds of work, not specifying what applications/science runs that work should be for, that gets decided on the server via prefs and work availability.
RE: Will the GPU version
)
Certainly not during that "tuning" run.
We noticed problems with the previous Windows application versions and the setup of the workunts, thus the app is still in "Beta" test.
We fixed the workunits and built and published a new Windows app version. This has no AVX support, but at least it should run at all. We'll continue to work on that.
BM
BM
FYI: I had and "01" job
)
FYI:
I had and "01" job with runtime/cputime ~38500
It failed with:
116.......c
.....................................c
................................c
.................................c
..................................c
..............................c
.....................................c
........
2016-02-11 22:19:22.1356 (19492) [normal]: Finished main analysis.
2016-02-11 22:19:22.1356 (19492) [normal]: Recalculating statistics for the final toplist...
2016-02-11 22:21:58.9230 (19492) [normal]: Finished recalculating toplist statistics.
2016-02-11 22:21:58.9230 (19492) [debug]: Writing output ... toplist2 ... toplist3 ... done.
FPU status flags: COND_3 PRECISION
2016-02-11 22:21:59.6677 (19492) [normal]: done. calling boinc_finish(0).
22:21:59 (19492): called boinc_finish
upload failure:
h1_0024.55_O1C01Cl1In1__O1AS20-100T_24.6Hz_171_1_1
-161 (not found)
h1_0024.55_O1C01Cl1In1__O1AS20-100T_24.6Hz_171_1_2
-161 (not found)
]]>
Last GW GPU app version
)
Last GW GPU app version required double precision hardware. What about next?
I don't know if this is
)
I don't know if this is useful for the project or if it's just my machine having a brain fart.
work unit h1_0021.45_O1C01Cl1In1__O1AS20-100T_21.5Hz_171_3 did this after over 12 hrs of processing.
Those processors and
)
Those processors and operating systems keep changing
RE: upload failure:
)
This would appear to be one of the three types of issues that Christian lists here. In particular, it's the third in the list about missing result files. Your snippet above shows two result files that are missing (-161). If you look at the various reports following Christian's post you will see that these (and other) errors are happening on other machines and is certainly not an issue with your host.
Unfortunately, until there is more feedback from Christian, this looks like something that may not be solved until Monday. It would probably be useful to Christian for any further reports of issues like this to be added to the above thread in the Problems section rather than the News thread.
Cheers,
Gary.
Just to pile on.... Sat
)
Just to pile on....
Sat Feb 13 02:36:32 2016 | Einstein@Home | Computation for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 finished
Sat Feb 13 02:36:32 2016 | Einstein@Home | Output file h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_1 for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 absent
Sat Feb 13 02:36:32 2016 | Einstein@Home | Output file h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_2 for task h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4 absent
Sat Feb 13 02:36:32 2016 | Einstein@Home | Starting task LATeah0158E_80.0_528_-2e-10_1
Sat Feb 13 02:36:34 2016 | Einstein@Home | Started upload of h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_0
Sat Feb 13 02:36:43 2016 | Einstein@Home | Finished upload of h1_0094.35_O1C01Cl1In1__O1AS20-100T_94.45Hz_2478_4_0
Sat Feb 13 02:38:09 2016 | Einstein@Home | Sending scheduler request: To report completed tasks.