Hi, I have recently started (again) crunching data and since this evening I noticed something strange:
The percentage of crunched data is going up in a lineair way, but at he same time the remaining time for the work unit is also going up. Not lineair but then a few seconds up, some seconds down, then a standstil and slowly going up again.
I have tried to stop as many services in the computer as possible, Boinc takes 99% of the processor all the time.
What can cause this effect? I didn't see it before and I have been staring at the remaining time a lot since it used to go down 4-5 seconds at a time, making the WU finish a lot faster than originally stated.

DeMus

Don't worry about it for now. This is BOINC just getting used to what your machine can do in terms of it's performance, and it takes a number WU's getting run through it for the estimates to settle down.

Alinator

Just in case, try:

Shutdown BOINC
Restart BOINC

Some time back there was a failure scenario like this where
units seemed to be making no net gain on the end-time.

Never heard that it was fully understood or fixed, so it might
still be out there. BOINC restart made BOINC understand something
was wrong, and the unit quickly errored out and BOINC went on with life.

Dave

Some time back there was a failure scenario like this where
units seemed to be making no net gain on the end-time.

This doesn’t sound much like that situation, though: DeMus’s tasks are apparently making steady progress according to the percentage indicator. I agree with Alinator that the most likely cause is the BOINC client’s having a short enough ‘track record’ that each new timing estimate, especially those that are slower than average, still makes the RDCF jump.

In general the time-to-completion figures are somewhat prone to getting out of whack, but there’s a self-correcting mechanism that works pretty well, as long as the task-size estimates that come from the project are reasonably accurate—which is usually the case here.

---

This doesn’t sound much like that situation, though: DeMus’s tasks are apparently making steady progress according to the percentage indicator. I agree with Alinator that the most likely cause is the BOINC client’s having a short enough ‘track record’ that each new timing estimate, especially those that are slower than average, still makes the RDCF jump.

In general the time-to-completion figures are somewhat prone to getting out of whack, but there’s a self-correcting mechanism that works pretty well, as long as the task-size estimates that come from the project are reasonably accurate—which is usually the case here.

Could you please translate that into plain English, since it is not my first language. I thought I could manage pretty well, but you lost me. Sorry.
What is: short enough ‘track record’ , what is a RDCF jump? Just so I will understand what you are talking about.
Thanks.

This doesn’t sound much like that situation, though: DeMus’s tasks are apparently making steady progress according to the percentage indicator. I agree with Alinator that the most likely cause is the BOINC client’s having a short enough ‘track record’ that each new timing estimate, especially those that are slower than average, still makes the RDCF jump.

In general the time-to-completion figures are somewhat prone to getting out of whack, but there’s a self-correcting mechanism that works pretty well, as long as the task-size estimates that come from the project are reasonably accurate—which is usually the case here.

Could you please translate that into plain English, since it is not my first language. I thought I could manage pretty well, but you lost me. Sorry.
What is: short enough ‘track record’ , what is a RDCF jump? Just so I will understand what you are talking about.
Thanks.

The Result Duration Correction Factor (RDCF) is reduced slowly and exponentially if the results are processed faster than the original projected completion time, but increased immediately if the processing time is longer than the predicted time.

It takes 20 to 25 results to be completed quicker than predicted for the RDCF to stabalise.

And after the RDCF has stabalised there are some angle_ranges (AR) results where the actual processing time is slightly longer than predicted. This causes the RDCF to vary slightly, and if plotted would look like a sawtooth pattern.

What is: short enough ‘track record’ , what is a RDCF jump?

Sorry for the jargon. What I meant was that BOINC has not yet collected enough statistics on the host’s performance to make accurate predictions. When there aren’t many data points in a sample, the ‘weight’ of each added point is comparatively high; the larger the sample gets, the smaller the influence of each new datum.

A host’s RDCF in a given project can be found near the bottom of its Computer Summary page; BOINC multiplies this number by the project’s duration estimate when calculating how long each task is expected to take. By design, longer-than-expected tasks increase the RDCF more than shorter-than-expected tasks decrease it.

I've noticed "time to completion" increasing on some Einstein units. Probably the initial estimate was too low or else inhomogeneities in the workunit caused the ups and downs in "time to completion". The Einsteins all got done OK and I saw no problem. I haven't seen rising "time to completions" with Seti units unless work froze. Resetting Boinc at the lower right icon, using a mouse rightclick, often rectifies the problem. If not, I have aborted the unit but, of course, received no credit for the lost time.

---

If you are not useing the newest verision, you might download it.

I've noticed "time to completion" increasing on some Einstein units. Probably the initial estimate was too low or else inhomogeneities in the workunit caused the ups and downs in "time to completion". The Einsteins all got done OK and I saw no problem. I haven't seen rising "time to completions" with Seti units unless work froze. Resetting Boinc at the lower right icon, using a mouse rightclick, often rectifies the problem. If not, I have aborted the unit but, of course, received no credit for the lost time.

The variation in time to completion for Einstein units was plotted by archae86 and Richard Haselgrove in Einstein - S5R3 thread.

Hi everyone,
The problem that I am having is that work units are getting to last way to long to complete, if they complete at all. I have a AMD 64 X2 - 3800, running at 2ghz, with 4 gigs of ram. I run only Seti with Boinc 5.10.30. I was using crunch3r's newest mods including both the Seti and Boinc mods, but stopped using them in Dec when I started having problems. I removed the mods, as well as completely re-installing Boinc. I was successful at running 40 or so work units, since then, that were smaller, but now I am getting longer work units that went from lasting 4 hours to ones that last 22 hours plus, if they complete at all. I am now aborting work units that start to rise in remaining time, after about 4 hours of running time. In all I have aborted around 30-40 work units, because they just don't complete. I have Boinc installed in the root directory of "C" drive as per instrutions to get Bonic to work on Windows Vista. I am debating deleting Boinc from my computer and stopping working for Seti because of these problems. The work units should complete in less than 12 hours on my machine, if not I will just not run Seti anymore.

1/14/2008 1:25:49 PM||Starting BOINC client version 5.10.30 for windows_intelx86
1/14/2008 1:25:49 PM||log flags: task, file_xfer, sched_ops
1/14/2008 1:25:49 PM||Libraries: libcurl/7.17.1 OpenSSL/0.9.8e zlib/1.2.3
1/14/2008 1:25:49 PM||Data directory: C:\\Seti\\Boinc
1/14/2008 1:25:49 PM||Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ [x86 Family 15 Model 75 Stepping 2]
1/14/2008 1:25:49 PM||Processor features: fpu tsc pae nx sse sse2 pni 3dnow mmx
1/14/2008 1:25:49 PM||OS: Microsoft Windows Vista: Home Edition, Service Pack 1, v.668, (06.00.6001.00)
1/14/2008 1:25:49 PM||Memory: 3.50 GB physical, 7.16 GB virtual
1/14/2008 1:25:49 PM||Disk: 298.04 GB total, 248.35 GB free
1/14/2008 1:25:49 PM||Local time is UTC -8 hours (Pacific time- USA)
1/14/2008 1:25:49 PM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: "deleted for this message"; location: home; project prefs: default
1/14/2008 1:25:49 PM||General prefs: from SETI@home (last modified 27-Nov-2007 22:35:44)
1/14/2008 1:25:49 PM||Host location: home
1/14/2008 1:25:49 PM||General prefs: no separate prefs for home; using your defaults
1/14/2008 1:25:49 PM||Reading preferences override file
1/14/2008 1:25:49 PM||Preferences limit memory usage when active to 1790.79MB
1/14/2008 1:25:49 PM||Preferences limit memory usage when idle to 3223.41MB
1/14/2008 1:25:49 PM||Preferences limit disk usage to 4.66GB

I had Seti (Bonic),set to run at 50% of processor time, but I now have it set to run 75% of the time. I see no need to run it at 100% of the time with a 2gig processor, however I could be wrong. Still it seems to be taking longer and longer to complete work units, if they complete at all. I did not have any problem doing the 40 or so work units, that I have completed since I re-installed Boinc in December. Time remaining has changed form 4 hours per work unit, to a bottom of 10 hours and a top of 22 hours. Is this normal for a machine as fast as mine? If not, how do I fix it? Oh, the first thing I do is try to restart Boinc, and this has had no effect. I then restart the computer, also to no effect.

Catman

---

The Catman is Back

Cat, don't stop being part of the SAH family. Plug in Vista in the search box and it will give you the information you need to fix your problem.

---

Pluto will always be a planet to me.

Seti Ambassador
Not to late to order an Anni Shirt

I had Seti (Bonic),set to run at 50% of processor time, but I now have it set to run 75% of the time. I see no need to run it at 100% of the time with a 2gig processor, however I could be wrong. Still it seems to be taking longer and longer to complete work units, if they complete at all. I did not have any problem doing the 40 or so work units, that I have completed since I re-installed Boinc in December. Time remaining has changed form 4 hours per work unit, to a bottom of 10 hours and a top of 22 hours. Is this normal for a machine as fast as mine? If not, how do I fix it? Oh, the first thing I do is try to restart Boinc, and this has had no effect. I then restart the computer, also to no effect.

Catman

At first You have hidden Your computer, so I cannot see Your Benchmarks. This would help me to to help You :)

Next, I know what it is like when English is not Your first language, so I will try to make it plain. (Sorry to all You native Folks :) for the longer post)

Now:

Try Task Manager, Processes, sort on CPU Utilization and watch.

Does Each SETI task get 37% CPU time? (You restricted BOINC to 75%)

SETI Tasks are low priority giwing way to all Your normal programs, so I do not think it is neccessary to restrict BOINC. Try setting it at 100%.

Each SETI Task should now get almost 50% CPU Power, but if You have some other task eating up Your CPU time, You will see the percentage for SETI go way down.

Your benchmarks are run at normal priority and gets the assigned amount of CPU Power. This should get almost all CPU Power minus the Windows core utilities. Say 98%. This may tell SETI that this next task should take say 3 Hours, which is set as time remaining. Now Your computer has some other task running that leaves only 20% for SETI. Time still passes by, but SETI needs way more time to crunsh, therefore Time Remaining goes up.

Try running SETI for 7-8 hours, and tell me how much CPU Power SETI is getting in Task Manager.

Best of luck :)

ChrisD

---

Hi Everyone,
Here are my CPU Benchmarks at 75%:
1/15/2008 10:23:44 PM||Running CPU benchmarks
1/15/2008 10:24:16 PM||Benchmark results:
1/15/2008 10:24:16 PM|| Number of CPUs: 2
1/15/2008 10:24:16 PM|| 1994 floating point MIPS (Whetstone) per CPU
1/15/2008 10:24:16 PM|| 3682 integer MIPS (Dhrystone) per CPU
1/15/2008 10:24:17 PM||Resuming computation

Here are the results from Task Manager:
Boinc Client: 3,716k
Boinc manager for Windows: 3,328k
setiathome_enhanced: 74,204k
setiathome_enhanced: 74,528k

Now I set Bonic to run at 100%
CPU Benchmarks:
1/15/2008 10:28:10 PM||Running CPU benchmarks
1/15/2008 10:28:10 PM||Suspending computation - running CPU benchmarks
1/15/2008 10:28:41 PM||Benchmark results:
1/15/2008 10:28:41 PM|| Number of CPUs: 2
1/15/2008 10:28:41 PM|| 1972 floating point MIPS (Whetstone) per CPU
1/15/2008 10:28:41 PM|| 3674 integer MIPS (Dhrystone) per CPU
1/15/2008 10:28:43 PM||Resuming computation

Here are the results from task manager:
Boinc Client: 3,716k
Boinc manager for Windows: 3,352k
steiathome_enhanced: 40,964k
setiathome_enhanced: 40,808k
under the cpu tab it says: 48 and 48

how do I un-hide my computer?
oh by the way my Computer and Boinc never gets shut down, but it does get re-started every several days or so, to clean out the garbage. Boinc is set to run always.

I have Boinc installed in the C:\\seti\\boinc folder, not in program files, as suggested in these forums.

Thanks for the help, Catman

---

The Catman is Back

The control for hiding is in the default section of the Resource Share and Graphics Preferences page.

Alinator

Hi Everyone,
This is what the Resource Share and Graphics Preferences page says:

Resource share
If you participate in multiple BOINC projects, this is the proportion of your resources used by SETI@home 100

Is it OK for SETI@home and your team (if any) to email you?
Emails will be sent from setiadm@ssl.berkeley.edu; make sure your spam filter accepts this address. yes

Should SETI@home show your computers on its web site? yes
*** I take this to mean that the computers are not hidden, I think that I did remove this info from my earlier post though.***

Default computer location home

Graphics preferences
Select Custom to edit individual parameters SETI@home classic

Color preferences:
Select Custom to edit individual parameters Water World

URL of background image
This image (JPEG, Targa, BMP or PPM format) will be displayed in the background. none
URL of logo image

This image (JPEG, Targa, BMP or PPM format) will be displayed in the lower corner. none

Oh I forgot to add this before, the user account control is OFF OFF OFF, as I hate popups of any kind.

Here are my computers:

4106670 dabigkittie2 154.40 8,816 AuthenticAMD
AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ [x86 Family 15 Model 75 Stepping 2] Microsoft Windows Vista
Home Edition, Service Pack 1, v.668, (06.00.6001.00) 73 16 Jan 2008 6:31:31 UTC

4032922 DABIGKITTIE 0.68 117 GenuineIntel
x86 Family 6 Model 8 Stepping 3 798MHz [x86 Family 6 Model 8 Stepping 3] Microsoft Windows XP
Professional Edition, Service Pack 2, (05.01.2600.00) 1 29 Dec 2007 5:34:21 UTC

I also installed Crunch3r's 2.4v mods for both Boinc and Seti on my Dell 800 XPS Dimension, 512 ram, to try and help work unit competion time. At the moment the 800 takes about a week to complete 1 work unit, at 100% CPU utilization, so I stopped running the computer. I have all user programs stopped on that computer. Seti was my only reason to run the computer.

I did have problems running the Crunch3r's mods on this computer, the AMD one, after a while, in December. I can't remember what the problem was at this time, but I uninstalled Boinc with the mods and reinstalled Boinc clean, and have done at least 40-50 successful work units until I got into the "28's, then the time increase problem started, I ended up aborting all of the "28's. Now I get, what seems to be, every other one having to be aborted.

Thanks for the help, Catman

---

The Catman is Back

I did have problems running the Crunch3r's mods on this computer, the AMD one, after a while, in December. I can't remember what the problem was at this time, but I uninstalled Boinc with the mods and reinstalled Boinc clean, and have done at least 40-50 successful work units until I got into the "28's, then the time increase problem started, I ended up aborting all of the "28's. Now I get, what seems to be, every other one having to be aborted.

Thanks for the help, Catman

Hi Catman,

Your phase "all the 28's" caught my eye.

I've noticed the WUs labelled 22fe07ah run far longer than the 3hrs originally estimated. I have one at 20hrs currently, but am encouraged that two actually finished this evening.
204492441 and 204492399

Hang in there, OK?

Andrew

---

AndrewM

Hi everyone,
The problem that I am having is that work units are getting to last way to long to complete, if they complete at all.
...
Catman

It looks like your system is sometimes losing the checkpoint file content so restarts have to start from the beginning of a WU again. That's been a problem which affects some systems more than others, usually setting "Leave applications in memory while suspended?" to 'yes' in Computing preferences is a help. You have over 3GB memory so it should have almost no impact.

                                                               Joe

Hi Everyone,
Thanks for the reply's. Yes, you may be right about the checkpoint file content. It has seemed that the work units were restarting every few hours or so. Boinc seems to be working ok at the moment.

If things start to go funky again, then I will set the "Leave applications in memory while suspended?" to 'yes', but for the moment I will leave things as they are.

I'm not sure why things might get suspended, as I have Boinc set to run always, but I will try this fix, if things get bad again.

---

The Catman is Back