Windows S5R3 App 4.13 available for Beta Test
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4332
Credit: 252135464
RAC: 33484
19 Oct 2007 13:04:21 UTC
Topic 193242
(moderation:
A new Windows App is available from our Beta Test Page.
This App fixes the problem that actually was in the command-line parsing of the 4.11 Beta App (exit status 99, "Ending frequency not contained in SFT (sequence)").
In addition it forces immediate syncing of the checkpoint file (thanks Bikeman!), which should make the checkpointing even more reliable.
A new Linux App with these two changes over the 4.12 Beta is currently being compiled.
BM
BM
Windows S5R3 App 4.13 available for Beta Test
)
This one died about 3 minutes after restarting BOINC after changing from 4.07 (I installed it as the beta).
10/19/2007 10:12:17 PM|Einstein@Home|[cpu_sched] Starting h1_0225.00_S5R2__11_S5R3a_1(resume)
10/19/2007 10:12:19 PM|Einstein@Home|Restarting task h1_0225.00_S5R2__11_S5R3a_1 using einstein_S5R3 version 407
...
10/19/2007 10:15:33 PM|Einstein@Home|Computation for task h1_0225.00_S5R2__11_S5R3a_1 finished
10/19/2007 10:15:33 PM|Einstein@Home|Output file h1_0225.00_S5R2__11_S5R3a_1_0 for task h1_0225.00_S5R2__11_S5R3a_1 absent
It *might* have been my fault. I forgot to stop the service before pasting in the new app and xml file. But it did run for about 3 minutes before it died after I stopped/started the service.
I'll go ahead and start a fresh result.
BTW: It's BOINC 5.10.24 as a service, XP Pro, P4 @ 2.8, 512 MB RAM.
[edit]
And another one bites the dust at 00:03:07. So just about the same point as the last one.
Guess I'll go back to the stock app.
Kathryn :o)
Einstein@Home Moderator
Thanks for the report. It
)
Thanks for the report. It seems that actually the syncing of the checkpoint file doesn't work as expected.
I'll disable this feature for now (on Windows).
Sorry for this, I'm a bit in a hurry today as I'll be out of town for a while and wanted to get this fix out of the door.
BM
BM
RE: I'll disable this
)
Done. The App in the current archive (md5sum 86afe181984a94cfc2e99a6c48c2d26a, old was 70f33a3fe18eee94d6a8d4b1bd2f5c59) has this disabled until I have time for more investigation.
If you were fast enough to get the old one, please download a fresh copy and install the file again, now.
BM
BM
RE: If you were fast
)
He's not kidding--with the originally posted files all-inprocess restarts errored, all previously in-queue results errored, fresh result downloads errored, and fresh-fresh results after a project rest errored. Typically after indicated progress reached about .185%, about a minute on my Core 2 machines.
Currently I'm past 1% indicated completion at 6 minutes, so the "even newer" ap does resolve this.
It may be a coincidence, but I was watching on one machine, and my firewall reported to me an internet access attempt when two instances were still running. The moment I hit the accept button, both errored out.
Seems to be working ok for me
)
Seems to be working ok for me now.
Kathryn :o)
Einstein@Home Moderator
Two results completed, both
)
Two results completed, both were fresh starts from new downloads after installation of the "newer new" 4.13.
One was run on a Core 2 Duo under windows XP. The other result in the quorum has not yet been sent, so no telling when validation information might come in.
The second ran on a Core 2 Quad, and also has the other quorum result as yet unsent.
As my installation method and the bug meant I trashed a lot of results, these are not contiguous with ones with known execution time, but my first guess is that they are very like 4.07.
I do have one in-process result which started processing on 4.11, but the host is a Banias and it is only a quarter done. So far it seems to be running peacefully, but bear in mind the download page warning against this practice.
This one worked just fine.
)
This one worked just fine. Started under 4.11, finished under 4.13.
This one also finished with
)
This one also finished with no errors. Started under 4.11, finished under 4.13.
Anybody had success (validate
)
Anybody had success (validate state "valid" or at least "initial") with 4.13 on a workunit where a 4.07 App failed with a "validate error"?
BM
BM
Ok, I have convinced myself
)
Ok, I have convinced myself that the problem of the "validation errors" of the 4.07 App actually is what I thought and that the 4.13 App would cure it. I'm about to advance the 4.13 to the official Windows App.
A new App (4.15) is running internal tests and about to be published, current Beta testers might want to stay on the anonymous platform path for a few more days.
BM
BM