Hi, i've a problem with estimate computing times, last week they started to be incorrect and now all tasks give me 00:-8:-8 to completion; this number changes only when the task is close to be finished and show others incorrect times (e.g. -437582:23:-4). The rest seems to work fine anyway but all tasks run with High Priority tag.

Im running BOINC v5.10.28 under Windows XP SP2 (italian version), CPU=AMD Athlon64x2 @4600+, DirectX v9.0c

Any other have this problem or know how to fix?

Which project(s) is this on?

---

I've 14 projects running: boincsimap, SETI@home, Einstein@home, climateprediction.net, Leiden Classical, lhcathome, malariacontrol.net.beta, predictor@home, TANPAKU, QMC@home, Rosetta@home, Spinhenge@home, uFluids and World Community Grid

Now it occurs with all tasks from all projects

OK, I think you have a corrupt client_state.xml file (and that the backup client_state_prev.xml is also corrupt).

So my advice would be to put all projects to No New Tasks, run their cache dry, report everything, then exit BOINC.
Navigate to your BOINC directory, copy all account_*.xml files to a safe place.
Uninstall BOINC and delete the whole BOINC directory and sub-directories.
(optional: defrag the drive)
(optional: write something large to the drive, then delete it again)
Reinstall BOINC, but don't let it start up.
Navigate back to your BOINC directory and copy the account_*.xml files back.
Restart BOINC and see if that fixed it. (it should)

---

OK, I think you have a corrupt client_state.xml file (and that the backup client_state_prev.xml is also corrupt).

I think there could be some hope of recovery (he's running CPDN so that would be desirable!).

Zorro: send me the client_state.xml, I'll see if there is anything that can be done with it. If I find it's hopeless, you could go on with Ageless's suggestion.

---

Sent.

The only unusual thing i remember was some day ago with a task from SETI@home, at the end i've seen computing error in status column, it reported to project anyway and since then i've noticed this problem but can't be sure they're linked.

Thanx for helping :)

Thanks Nicolas. I overlooked that he was attached to CPDN. Would've been a long wait for that. I hope you can fix it.

And let me know what to look for a next time, please, if you do know how to fix it.

---

Sent.

Hmm I found nothing wrong. As in, the client_state.xml doesn't seem to be corrupt.

Did you ever restart the client or the computer during the last week?

---

I shut down my PC daily (by night), i've put BOINC icon in Windows Autoexe folder so the client start each time i turn on, sometimes i also leave the client running while i use other softwares (i've a dual core CPU); all my drivers are up to date (except audio maybe)

P.S. Yesterday i tried to install back BOINC using repair option, no changes

I'd bet it has something to do with Zorro's Regional Settings (Start / Control Panel / Regional and Language Options / Customize... / Time tab), maybe something in the Time setup is confusing Boinc? IIRC it's output depends on them.

Peter

I'm using default settings for date/time; in the field you posted the only setting filled is the time format that is: H,mm,ss ... nothing more.

I've an AMD processor and i'm using AMD Dual Core Optimizer, i've heard that rarely could cause problems ... maybe BOINC is confused reading the correct CPU clock time. I'm using BOINC default settings for CPU usage, which means 100% both cores.

I add this screenshot so you can see the weird estimate time.
http://digilander.libero.it/zorro2k2/test.jpg

Anyway i'll follow Ageless suggestion, as soon as my PC finish CPDN task i'll try uninstall/clean/install operation. I'll keep you updated ;)

Well, then something related to what Sekerob suggested.
Did you have any time jumps on your computer recently? SixTrack's ETA sems to be 48.5 years ago, is your curent date correct? (Really suggests a broken client_state.xml file, but as Nicolas said...)

Peter

The fact that both processes run in high priority suggests that the system date/time is not correct.

You've beaten me by seconds :-)

Peter

Could repeat the question i put earlier in this thread, but simply ask for a Piece of the message log (select the start up message and use button in left margin to copy to clipboard).

The fact that both processes run in high priority suggests that the system date/time is not correct.

LHC job 17 minutes and 48% leaves it about 20 minutes to go and due on dec.24, yet it runs hi prior
CPDN, 93 hours, 23%.... leaves it about 300 hours and its due nov.2008.

Please also post the timestats section of the client_state.xml like

<time_stats>
<on_frac>0.998574</on_frac>
<connected_frac>1.000000</connected_frac>
<active_frac>0.993950</active_frac>
<cpu_efficiency>0.920385</cpu_efficiency>
<last_update>1197897894.053999</last_update>
</time_stats>

Sorry haven't answered before ... my OS date/time is fine, no jumps or weird happenings. I've upgraded CPU recently, before i had the same processor but 3800+; should be an unuseful info cause BOINC has been installed only after the upgrade.
Here asked info:
<time_stats>
<on_frac>0.741006</on_frac>
<connected_frac>-1.000000</connected_frac>
<active_frac>0.946174</active_frac>
<cpu_efficiency>0.797275</cpu_efficiency>
<last_update>1197900033.218750</last_update>
</time_stats>

Don't you think that BOINC could understimate my CPU clock? I've seen that when a task is close to end (about 10 mins) estimate time show 'normal numbers' ... but it's anyway a time much bigger than real (e.g. real 10 min=25:54:35 ET) then each second show +/- 20 min less ET. Looks like BOINC tagged my PC as sloooow but it isn't.

P.S. Forgot to specify that CPU computing time is working fine, the problem is just with ET

How do your <p_fpops>1376555982.527014</p_fpops> and <p_iops>2802058669.237640</p_iops> values look like?

Peter

<p_fpops>2453584320.323899</p_fpops>
<p_iops>4436056243.410433</p_iops>

Try clicking "Run CPU benchmarks", from the Advanced menu.

---

Further the connect line is obscure, showing a minus value:
<connected_frac>-1.000000</connected_frac>

That has been like that for *many* BOINC versions.

---

Further the connect line is obscure, showing a minus value:
<connected_frac>-1.000000</connected_frac>

That has been like that for *many* BOINC versions.

So, are you 'telling' or is it 'telling'?

It's the same on mine.

Peter

BIOS date/time is the same than OS. The problem is somewhere else.

Here the CPU benchmark result requested by Nicolas:

18/12/2007 15.10.04||Benchmark results:
18/12/2007 15.10.04|| Number of CPUs: 2
18/12/2007 15.10.04|| 2444 floating point MIPS (Whetstone) per CPU
18/12/2007 15.10.04|| 4392 integer MIPS (Dhrystone) per CPU

I'm not sure what message log want to see Sekerob, i guess this:

18/12/2007 14.47.50||Starting BOINC client version 5.10.28 for windows_intelx86
18/12/2007 14.47.50||log flags: task, file_xfer, sched_ops
18/12/2007 14.47.50||Libraries: libcurl/7.17.0 OpenSSL/0.9.8e zlib/1.2.3
18/12/2007 14.47.50||Data directory: C:ProgrammiBOINC
18/12/2007 14.47.51||Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ [x86 Family 15 Model 43 Stepping 1]
18/12/2007 14.47.51||Processor features: fpu tsc pae nx sse sse2 3dnow mmx
18/12/2007 14.47.51||OS: Microsoft Windows XP: Professional Edition, Service Pack 2, (05.01.2600.00)
18/12/2007 14.47.51||Memory: 2.00 GB physical, 3.85 GB virtual
18/12/2007 14.47.51||Disk: 35.32 GB total, 14.83 GB free
18/12/2007 14.47.51||Local time is UTC +1 hours

If you want time_stats_log file, let me know, i'll send you :)

P.S. I've a 56k connection