Strange behaviour with rosetta mini 3.54

Message boards : Number crunching : Strange behaviour with rosetta mini 3.54

To post messages, you must log in.

AuthorMessage
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78240 - Posted: 31 May 2015, 10:02:43 UTC

Hi there

I just installed BOINC on one of my old laptops.

My setting is using 70% of CPU time.

I'm run two main projects SETI and Rosetta.

I found that SETI task works normal, but for rosetta mini, it keep crunching but only use about 20% CPU time, and seems the tasks are endless.

I'm crunching Rosetta on several laptops, and this never happened before.

Here is the screenshot



Could you please help?

Thanks

ID: 78240 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1829
Credit: 116,393,926
RAC: 71,810
Message 78241 - Posted: 31 May 2015, 10:18:00 UTC

I'm not sure I can help with why it is only using 20%, other than to say Rosetta work units tend to be disk-limited when starting, so it might change once the initialisation has completed and it is running properly (i.e. CPU limited).

Are you running @ 70% to keep the temperature down? If so, you might find running on one core @ 100% is more effective because you'll only have one task in RAM and it will get all of the L2 cache then.

D
ID: 78241 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78242 - Posted: 31 May 2015, 11:31:19 UTC - in response to Message 78241.  

I'm not sure I can help with why it is only using 20%, other than to say Rosetta work units tend to be disk-limited when starting, so it might change once the initialisation has completed and it is running properly (i.e. CPU limited).

Are you running @ 70% to keep the temperature down? If so, you might find running on one core @ 100% is more effective because you'll only have one task in RAM and it will get all of the L2 cache then.

D


Hi D
Thank you for the reply.

If you look the screenshot carefully, you will notice that the remaining time is missing.

Effectiveness is less important here, the issue is all the mini tasks become endless.

ID: 78242 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
aguiar@carrier.com.br

Send message
Joined: 19 Feb 06
Posts: 6
Credit: 367,089
RAC: 0
Message 78243 - Posted: 31 May 2015, 21:12:16 UTC

Hi, all!

Here I have a WU running for 31:57 hours at 27.154%. Also, no remaining time is shown. Mini 3.54, Windows.

Valter Aguiar.
denise.valter@cmg.com.br
ID: 78243 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78244 - Posted: 1 Jun 2015, 0:11:23 UTC

If you select one of the Rosetta tasks and click properties on the left, what comes up under CPU time at last checkpoint, CPU time & Elapsed time? I'm guessing there's very little CPU time at all, which indicates a problem somewhere else
ID: 78244 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78246 - Posted: 1 Jun 2015, 13:23:34 UTC - in response to Message 78244.  

If you select one of the Rosetta tasks and click properties on the left, what comes up under CPU time at last checkpoint, CPU time & Elapsed time? I'm guessing there's very little CPU time at all, which indicates a problem somewhere else


Hi Sid

Thank you for the reply.

I'm aware there are is a problem, since SETI works fine, so the problem should be rosetta mini.

ID: 78246 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78249 - Posted: 2 Jun 2015, 0:06:46 UTC - in response to Message 78246.  

If you select one of the Rosetta tasks and click properties on the left, what comes up under CPU time at last checkpoint, CPU time & Elapsed time? I'm guessing there's very little CPU time at all, which indicates a problem somewhere else

Hi Sid

Thank you for the reply.

I'm aware there are is a problem, since SETI works fine, so the problem should be rosetta mini.

Yes, I know that too. I'm asking for more specific details of the problem so we know where you should look next. An example would be:

CPU time at last checkpoint 4:52:23
CPU time 4:52:27
Elapsed time 5:08:56

This would be quite normal, but if CPU time lagged way behind elapsed time then it wouldn't matter how long since the task started, it wouldn't be processing much in that time. If CPU time was close to Elapsed time, but there was no checkpoint, that would indicate something different again. Which is it? Or would you like us to guess until someone guesses right?

Also, you should always have your CPU set to run at 100% CPU time. 70% actually means "run at 100% for 70% of the time, then run at 0% for 30% of the time". What it doesn't mean is "run at 70% for 100% of the time". Irrespective of that 100% setting, priority is set so low on Rosetta tasks that even if you just move your mouse or type, CPU is handed back to prioritise those things.
ID: 78249 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78251 - Posted: 2 Jun 2015, 14:33:32 UTC - in response to Message 78249.  

If you select one of the Rosetta tasks and click properties on the left, what comes up under CPU time at last checkpoint, CPU time & Elapsed time? I'm guessing there's very little CPU time at all, which indicates a problem somewhere else

Hi Sid

Thank you for the reply.

I'm aware there are is a problem, since SETI works fine, so the problem should be rosetta mini.

Yes, I know that too. I'm asking for more specific details of the problem so we know where you should look next. An example would be:

CPU time at last checkpoint 4:52:23
CPU time 4:52:27
Elapsed time 5:08:56

This would be quite normal, but if CPU time lagged way behind elapsed time then it wouldn't matter how long since the task started, it wouldn't be processing much in that time. If CPU time was close to Elapsed time, but there was no checkpoint, that would indicate something different again. Which is it? Or would you like us to guess until someone guesses right?

Also, you should always have your CPU set to run at 100% CPU time. 70% actually means "run at 100% for 70% of the time, then run at 0% for 30% of the time". What it doesn't mean is "run at 70% for 100% of the time". Irrespective of that 100% setting, priority is set so low on Rosetta tasks that even if you just move your mouse or type, CPU is handed back to prioritise those things.


Hi Sid
Thank you for the reply.

Let's focus on the endless time issue, not the 70% or 100%.

I reset the project, so I'm unable to copy the detail time, I still remember that CPU time lagged far way behind elapsed time.

I initial question is that WHY my CPU was not processing much in that time.

I just realized that there's a Q&A forum here, I will ask for help there.

Cheers

ID: 78251 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78254 - Posted: 3 Jun 2015, 3:32:37 UTC - in response to Message 78251.  

Hi Sid
Thank you for the reply.

Let's focus on the endless time issue, not the 70% or 100%.

I reset the project, so I'm unable to copy the detail time, I still remember that CPU time lagged far way behind elapsed time.

I initial question is that WHY my CPU was not processing much in that time.

This is harder work than it needs to be if you could just answer the question so that someone can help you.

The reason why is that you're constrained for either RAM or disk space. Not just the total you have but the amount you've allocated to Boinc (under ToolsComputing preferences).

You seem to have 3 machines. A single core with just 1Gb RAM, and 2 dual-core machines with 4Gb & 2Gb RAM. You don't say which is being affected, but that's a red flag for a start as each task uses 4-500Mb RAM and that has to fit into what you've allocated to Boinc. If you haven't allocated (or don't have) enough it's not surprising they don't run at all.

ID: 78254 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78255 - Posted: 3 Jun 2015, 13:10:44 UTC - in response to Message 78254.  

Hi Sid
Thank you for the reply.

Let's focus on the endless time issue, not the 70% or 100%.

I reset the project, so I'm unable to copy the detail time, I still remember that CPU time lagged far way behind elapsed time.

I initial question is that WHY my CPU was not processing much in that time.

This is harder work than it needs to be if you could just answer the question so that someone can help you.

The reason why is that you're constrained for either RAM or disk space. Not just the total you have but the amount you've allocated to Boinc (under ToolsComputing preferences).

You seem to have 3 machines. A single core with just 1Gb RAM, and 2 dual-core machines with 4Gb & 2Gb RAM. You don't say which is being affected, but that's a red flag for a start as each task uses 4-500Mb RAM and that has to fit into what you've allocated to Boinc. If you haven't allocated (or don't have) enough it's not surprising they don't run at all.


Hi Sid

Thank you for the help.

This issue happened on the 4GB machine.

https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=2302829

I didn't restrict any RAM nor disk space on the tasks. There is no any error information in the log.

Maybe try the debug flag?

Cheers
ID: 78255 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78259 - Posted: 4 Jun 2015, 0:43:19 UTC - in response to Message 78255.  

Hi Sid
Thank you for the reply.

Let's focus on the endless time issue, not the 70% or 100%.

I reset the project, so I'm unable to copy the detail time, I still remember that CPU time lagged far way behind elapsed time.

I initial question is that WHY my CPU was not processing much in that time.

This is harder work than it needs to be if you could just answer the question so that someone can help you.

The reason why is that you're constrained for either RAM or disk space. Not just the total you have but the amount you've allocated to Boinc (under ToolsComputing preferences).

You seem to have 3 machines. A single core with just 1Gb RAM, and 2 dual-core machines with 4Gb & 2Gb RAM. You don't say which is being affected, but that's a red flag for a start as each task uses 4-500Mb RAM and that has to fit into what you've allocated to Boinc. If you haven't allocated (or don't have) enough it's not surprising they don't run at all.

Hi Sid

Thank you for the help.

This issue happened on the 4GB machine.

https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=2302829

I didn't restrict any RAM nor disk space on the tasks. There is no any error information in the log.

Maybe try the debug flag?

Cheers

Urgh. The least likely one to give a problem...

There's always some restriction, though the defaults are usually still more than enough. Just to cover the basics:

The Disk tab in Boinc will show how much disk space you have allocated. Just confirm you have more than enough allocated to Boinc (5Gb+ ought to be plenty for a 2-core machine)

Under ToolsComputing Preferences, go to the Disk & Memory usage tab
In the Memory Usage section, I use 60% for when the computer is in use and 90% when the computer's not in use. Always ensure that "Leave applications in memory while suspended" is ticked. Boinc gives funny problems when it isn't (even if you'd prefer not to tick it).

If those are ok, I'm only left with that "use at most xx% of processorsCPU time" setting under the Processor Usage tab, which I recommend setting to 100%, but leave that until last if you insist on having it at 70% - though, as I said before, it's of no benefit to set it lower than 100%.

I notice you've started to get some tasks completing in the last day or so. Do tasks seem to be behaving themselves better already? Or not?

I have no idea about what the debug flag will tell you tbh. I'm not techie enough to say, though someone else here might be able to help better.
ID: 78259 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78264 - Posted: 4 Jun 2015, 14:23:18 UTC - in response to Message 78259.  

Hi Sid
Thank you for the reply.

Let's focus on the endless time issue, not the 70% or 100%.

I reset the project, so I'm unable to copy the detail time, I still remember that CPU time lagged far way behind elapsed time.

I initial question is that WHY my CPU was not processing much in that time.

This is harder work than it needs to be if you could just answer the question so that someone can help you.

The reason why is that you're constrained for either RAM or disk space. Not just the total you have but the amount you've allocated to Boinc (under ToolsComputing preferences).

You seem to have 3 machines. A single core with just 1Gb RAM, and 2 dual-core machines with 4Gb & 2Gb RAM. You don't say which is being affected, but that's a red flag for a start as each task uses 4-500Mb RAM and that has to fit into what you've allocated to Boinc. If you haven't allocated (or don't have) enough it's not surprising they don't run at all.

Hi Sid

Thank you for the help.

This issue happened on the 4GB machine.

https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=2302829

I didn't restrict any RAM nor disk space on the tasks. There is no any error information in the log.

Maybe try the debug flag?

Cheers

Urgh. The least likely one to give a problem...

There's always some restriction, though the defaults are usually still more than enough. Just to cover the basics:

The Disk tab in Boinc will show how much disk space you have allocated. Just confirm you have more than enough allocated to Boinc (5Gb+ ought to be plenty for a 2-core machine)

Under ToolsComputing Preferences, go to the Disk & Memory usage tab
In the Memory Usage section, I use 60% for when the computer is in use and 90% when the computer's not in use. Always ensure that "Leave applications in memory while suspended" is ticked. Boinc gives funny problems when it isn't (even if you'd prefer not to tick it).

If those are ok, I'm only left with that "use at most xx% of processorsCPU time" setting under the Processor Usage tab, which I recommend setting to 100%, but leave that until last if you insist on having it at 70% - though, as I said before, it's of no benefit to set it lower than 100%.

I notice you've started to get some tasks completing in the last day or so. Do tasks seem to be behaving themselves better already? Or not?

I have no idea about what the debug flag will tell you tbh. I'm not techie enough to say, though someone else here might be able to help better.


Hi Sid

Thank you for the instructions.
Here are the numbers.

Disk Tab, free available to BOINC 102.54GB

In the Memory Usage section, I use 50% for when the computer is in use and 90% when the computer's not in use. Currently I did not use this laptop for anything but BOINC, not environment friendly though.

I didn't tick "Leave applications in memory while suspended", and I ticked it now.

Yes, I also noticed that there're some tasks finished without any issues, since there's no log, I'm unable to tell the differences between the tasks.

Currently, there's another endless task, I checked the properties, and here is the time:

CPU time at last checkpoint 01:37:23
CPU time 01:38:09
Elapsed time 25:37:33

I ticked "Leave applications in memory while suspended" now, Let's keep our fingers crossed.

Cheers
ID: 78264 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78280 - Posted: 6 Jun 2015, 10:06:11 UTC - in response to Message 78264.  
Last modified: 6 Jun 2015, 10:12:30 UTC

Hi Sid

Thank you for the instructions.
Here are the numbers.

Disk Tab, free available to BOINC 102.54GB

In the Memory Usage section, I use 50% for when the computer is in use and 90% when the computer's not in use. Currently I did not use this laptop for anything but BOINC, not environment friendly though.

I didn't tick "Leave applications in memory while suspended", and I ticked it now.

Yes, I also noticed that there're some tasks finished without any issues, since there's no log, I'm unable to tell the differences between the tasks.

Currently, there's another endless task, I checked the properties, and here is the time:

CPU time at last checkpoint 01:37:23
CPU time 01:38:09
Elapsed time 25:37:33

I ticked "Leave applications in memory while suspended" now, Let's keep our fingers crossed.

Cheers

One other check. In ToolsComputing Preferences, on the processor usage tab:

Only after computer has been idle for x.xx minutes
While processor usage is less than x per cent (0 means no restriction)

Both values should be set at zero. Especially if your laptop isn't used for anything else.

I'm currently typing from my own laptop, which is very similar to yours - 4Gb RAM, Windows 7 x64 - running Rosetta, WCG and Malaria while browsing, occasional youtube use, social media - routine stuff, no slowdown in use at any time. Your settings ought to run fine if they're set the same as I have here. The difference you show between CPU time and Elapsed time indicates Rosetta is (nearly) being prevented from running AT ALL!

I have a current task showing:
CPU time at last checkpoint 02:33:34
CPU time 02:35:10
Elapsed time 02:43:18

Yours shouldn't be needing 25hrs to allow you 1 hour less CPU time, so just bump up your "use at most xx% when computer is in use" to 60% under Memory usage as well. It can't do any harm if I don't notice a problem here in general use.

Edit: The disk space you allocate is massive. I'm beginning to wonder if there's any left for normal windows operations! (I'm sure there is, but it makes me wonder). Take this setting down by 50Gb just to be on the safe side. I only use 10Gb total. Maybe 20Gb total is ok as a compromise figure.
ID: 78280 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78298 - Posted: 13 Jun 2015, 2:29:31 UTC - in response to Message 78280.  

Hi Sid

Thank you for the instructions.
Here are the numbers.

Disk Tab, free available to BOINC 102.54GB

In the Memory Usage section, I use 50% for when the computer is in use and 90% when the computer's not in use. Currently I did not use this laptop for anything but BOINC, not environment friendly though.

I didn't tick "Leave applications in memory while suspended", and I ticked it now.

Yes, I also noticed that there're some tasks finished without any issues, since there's no log, I'm unable to tell the differences between the tasks.

Currently, there's another endless task, I checked the properties, and here is the time:

CPU time at last checkpoint 01:37:23
CPU time 01:38:09
Elapsed time 25:37:33

I ticked "Leave applications in memory while suspended" now, Let's keep our fingers crossed.

Cheers

One other check. In ToolsComputing Preferences, on the processor usage tab:

Only after computer has been idle for x.xx minutes
While processor usage is less than x per cent (0 means no restriction)

Both values should be set at zero. Especially if your laptop isn't used for anything else.

I'm currently typing from my own laptop, which is very similar to yours - 4Gb RAM, Windows 7 x64 - running Rosetta, WCG and Malaria while browsing, occasional youtube use, social media - routine stuff, no slowdown in use at any time. Your settings ought to run fine if they're set the same as I have here. The difference you show between CPU time and Elapsed time indicates Rosetta is (nearly) being prevented from running AT ALL!

I have a current task showing:
CPU time at last checkpoint 02:33:34
CPU time 02:35:10
Elapsed time 02:43:18

Yours shouldn't be needing 25hrs to allow you 1 hour less CPU time, so just bump up your "use at most xx% when computer is in use" to 60% under Memory usage as well. It can't do any harm if I don't notice a problem here in general use.

Edit: The disk space you allocate is massive. I'm beginning to wonder if there's any left for normal windows operations! (I'm sure there is, but it makes me wonder). Take this setting down by 50Gb just to be on the safe side. I only use 10Gb total. Maybe 20Gb total is ok as a compromise figure.


Hi Sid

Thank you for the help.

My setting is computing when computer is in use. And I changed the While processor usage is less than x per cent to Zero.

My C disk still have 100GB space, so the space is not a concern.

Let's see if the situation will be improved.

Cheers

ID: 78298 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
droople
Avatar

Send message
Joined: 19 Aug 08
Posts: 18
Credit: 3,332,765
RAC: 64
Message 78331 - Posted: 22 Jun 2015, 2:00:34 UTC

bump
ID: 78331 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Timo
Avatar

Send message
Joined: 9 Jan 12
Posts: 185
Credit: 45,644,940
RAC: 157
Message 78332 - Posted: 22 Jun 2015, 2:34:42 UTC

Just reading over this thread. Usually if rosetta (or any BOINC app for that matter) is being throttled due to not enough disk or not enough ram, it will say so

ie. it will say something like "waiting for memory" like this:


... So, given that you're not seeing that, I doubt it's an issue with being memory/disk constrained. Also, given that the work units themselves are being successfully completed by other user's computers once they fail to complete on your PC tells me that it's not an issue with the WU but definitely with something on your PC.

So, a couple odd things that come to mind that may the issue
1) Do you have any antivirus that may be blocking the process from actually crunching? To test, turn off any antivirus program and see if progress starts to move forward.
2) Is the computer clock / date / time correct? This sounds really odd, but I've worked technical support long enough that I've seen my share of strange behavior that turned out to be due to a bad CMOS battery causing the Time/Date to run off (ie. if your computer thinks its the year 1981, it could cause some issues) - to check this, double click your clock on the bottom right corner of the screen and ensure that your timezone is set correctly for your location and that the date/time is otherwise sound.

Good luck!
ID: 78332 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2003
Credit: 38,699,332
RAC: 20,277
Message 78333 - Posted: 22 Jun 2015, 3:22:31 UTC - in response to Message 78331.  
Last modified: 22 Jun 2015, 3:33:50 UTC

bump

Is it reporting the same as before? No improvement at all? I notice you had to abort one task that had been running for a very long time, then another went through (apparently) normally. But just one successful task in 9 days is dreadful.

I confess, I'm out of ideas, especially as your laptop is so similar to mine... sorry :(
ID: 78333 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 78337 - Posted: 23 Jun 2015, 2:13:51 UTC

I changed the While processor usage is less than x per cent to Zero.


So this means that you only want BOINC tasks running if your CPU is absolutely doing nothing else. Any AV or operating system operation kicking in will exceed zero and BOINC will suspend the tasks.

Also, are you looking directly at the settings on the laptop? Or are you looking at the default configuration on the R@h website?
Rosetta Moderator: Mod.Sense
ID: 78337 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Timo
Avatar

Send message
Joined: 9 Jan 12
Posts: 185
Credit: 45,644,940
RAC: 157
Message 78339 - Posted: 23 Jun 2015, 4:31:30 UTC - in response to Message 78337.  

I changed the While processor usage is less than x per cent to Zero.


So this means that you only want BOINC tasks running if your CPU is absolutely doing nothing else. Any AV or operating system operation kicking in will exceed zero and BOINC will suspend the tasks.


... Actually '0' should imply no restrictions:




Also, if it were an issue with insufficient memory/processor settings within BOINC itself, the tasks would actually say 'Waiting for memory' or 'Waiting to run' etc. If they just say 'Running' but are not running, then I'm far more likely to suspect a third party application blocking the tasks (Antivirus) or OS misconfiguration (wrong timezone/system time/bad CMOS battery, etc.)

:)
ID: 78339 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 78341 - Posted: 23 Jun 2015, 22:54:25 UTC

Whoops, yep, I stand corrected.

Is there any difference in BOINC version between the various systems you have? The BOINC Manager actually controls when a given task gets CPU. Some older BOINC versions had more problems with this.
Rosetta Moderator: Mod.Sense
ID: 78341 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Strange behaviour with rosetta mini 3.54



©2024 University of Washington
https://www.bakerlab.org