Posts by Neil

1) Message boards : Number crunching : Getting WU's (Message 64827)
Posted 5 Jan 2010 by Profile Neil
Post:
The story from Dave and Ramona Girard demonstrates that when there is a Project outage -- but no information on the homepage or Technical News webpage is provided acknowledging that outage -- how much time and trouble some people will uselessly spend trying to get their Rosettas to work.

Of course, these man-hours are multiplied for each concerned user whom mostly duplicates the efforts of their fellow users, all trying their best to get their computers working again for the benefit of the Project.

It demonstrates that when there is a project outage, it affects not only the Project, but also the users.

The Girard’s story is the clearest example of why the organizers of the project, with all these users trying to do their best for the project, might provide the consideration of taking a few minutes to put a message on the Rosetta homepage and/or the Technical News webpage when there are interruptions on the Project's end.

These two locations are where I go first to confirm if there is a problem on the Project's end. It is my first step in troubleshooting when my Rosetta isn't working.
2) Message boards : Number crunching : Not getting any new tasks (Message 64768)
Posted 4 Jan 2010 by Profile Neil
Post:
Sorry, a few posts ago, on Jan 3, I wrote:

... maybe distribution of tasks will resume Jan 2?


I meant Jan 4.

I imagined toward the end of the day, though. Glad to hear some work units are coming back, already.

Someone in California's working Sunday night.

Woops, I downloaded two Tasks at 2100 EST; thanks.

Now, I have to figure out how I'm going to split my CPUs between my traditional Rosetta@home and Malariacontrol.net, which I just attached to a few days ago and consider worthy, also.
3) Message boards : Number crunching : Not getting any new tasks (Message 64758)
Posted 3 Jan 2010 by Profile Neil
Post:
The problems aren't yours, Hedera.

Many of us are surprised that such an important Project should go down because of the holidays, if that is in fact what the problem is. You can read about us arguing in the dark on thread http://boinc.bakerlab.org/rosetta/forum_thread.php?id=5200

If this is entirely a vacation phenomenon, maybe distribution of tasks will resume Jan 2?

Hope springs eternal.

-Neil-
4) Message boards : Number crunching : Scheduling request completed: got 0 new tasks (Message 64707)
Posted 1 Jan 2010 by Profile Neil
Post:
I have six Q9550 systems...


Wow, you have six and I don't have any. My next processor will be able to handle SSSE4.1, especially if Boinc can take advantage of that. In the meantime:

My 1.8 GHz Celeron 430 and 1.6 GHz Core2Duo T5450 usually do 20 work units per day, combined. They did 3 on December 29th, 5 on the 30th, 0 on the 31st, 0 so far with six hours remaining for New Years UTC.

(Attached both to Malariacontrol.net on Dec 30th, because the apartment was starting to get cold.)

And for those who missed Scene 1, the week began with no credits being granted for a few days while work units came and went, otherwise normally. There was never any notice provided about that problem either, for instance something along the line of, "Sorry our Credit Granting Computer is down -- the problem is on our end..."

Instead, nothing.

When there's a glitch, it's not only felt by the organizers. It's also felt by the participants.

Back when I was a kid walking uphill in both directions and using Vacuum Tubes, I participated in Seti@Home The Original Show. One day, someone drove a backhoe through a fiber-optic cable and stopped production for a week. In a day or so, a notice went up on their Technical Notices webpage to let us know what was going on. We were assured a couple of times as the destruction was being repaired.

And if someone did something stupid like spill coffee into the SETI server, they would provide a notice that said something like, "We hate to say it, but Bob spilled his coffee into the server. We expect to get back into the swing in two days..."

It didn't matter what the problem was. We appreciated being told, and all was forgiven -- just because we were recognized as being a significant part of the loop.
5) Message boards : Number crunching : Scheduling request completed: got 0 new tasks (Message 64701)
Posted 1 Jan 2010 by Profile Neil
Post:
Nothing to get your knickers in a twist about.


That's an antagonistic and unnecessary comment. I don't believe you've seen my knickers up close enough to determine the extent of their twist, so you're just making an Assumption, which is generally not helpful.

There is a connection between the organizers and the volunteers that is being disregarded, and I don't think it's being too assertive to say so directly.

The first line on the Rosetta webpage begins, "Rosetta@home needs your help..." and the fact that we're all lined up to help but given nothing to do is like a practical joke, or maybe a bit like letting your parents worry because you never called to tell them you're OK but going to be late. It's a matter of responsibility and consideration.

The day before I wrote my comment, I attached my Boinc to Malariacontrol.net. I appreciate Mikey filling us in on what Malariacontrol is concerned with, so I'm more comfortable with the choice I made.

My own insight is that it's Christmas and New Year and everyone on the project is having a well-deserved break and worrying about more important things, like their respective families.


If that were the issue, I wouldn't criticize. But how can we know if that's the issue when no one took 15 minutes sometime during the last week to let us know?

6) Message boards : Number crunching : Scheduling request completed: got 0 new tasks (Message 64682)
Posted 31 Dec 2009 by Profile Neil
Post:
Lots of general advice among people of good will, but if you look on the Rosetta homepage, there's no communication from people who really know what the problem is. So we don't really know what the problem is. When the lack of communication gets bad enough, can we be blamed for imagining the Project may never come back?

You might suggest the organizers are working at generating more Work Units for us, and time spent explaining the lack of work will only lengthen the time it takes before work units are finally made available.

In response to that suggestion, I would like to make two points.

1) If Rosetta is going to dry up for a while, it would be good opportunity to stretch our Boincs, and see how the GUI will look if we attach to other projects. But if Rosetta is where our loyalties are, we might think that it's not worth experimenting with other projects if Rosetta is about to come back on line.

I.e., we measure our productivity and usefulness by our Average Credits, and all things being equal, we may not be interested in bothering to accrue 25 Average Credits in another Project. And, if we've been crunching here for years, why should we imagine the project would be down for such an extended period?

Well, the trouble has been going on for at least a week, apparently beginning with no credits being granted for a few days. This down-time is time that we might have used to significantly help other Projects, except that no one has motivated us to do so by being forthcoming about how long this stoppage is going to last. So, aside from our User-disappointment, other Projects got cheated out of potential unused clock-cycles because of our managers' lack of communications.

2) A lot of the User-participation in this project happens because the users believe they have a stake in the project, whether it's because we like to have an excuse to play with our computers, or because we really believe we're helping Medical Science. Occasionally, for whatever reason, some people actually pay hundreds of dollars on extra hardware just to increase their Rosetta work-output. And then, there's their labor.

The point being, by neglecting to communicate with the users about why work is not being distributed (i.e., on Rosetta's homepage or in this forum), the people who are running the show are disrespecting the intentions of the users who have continually demonstrated their interest by participating. You are implying that we are simply a cyber-cloud not worthy of involvement in the goings-on, regardless of the fact that some of us have been involved for years. "Going Skiing" is a euphemism for "If you don't care about our participation in your project, then why should I? Bye-bye." This is a worst-case, but not unexpected example of what lack-of-communication engenders.

--

Therefore, since the lack of work-units has been disturbing and is starting to cause loss of participation, I don't buy the argument that taking out 15 minutes to compose a forum-thread to address the users' concerns about what's going on would have been time poorly spent.
7) Message boards : Number crunching : No new work (Message 64619)
Posted 29 Dec 2009 by Profile Neil
Post:
I just downloaded 4 units a few minutes ago


Wow, four.

I got one (1) task. It started 1 hr 15 minutes ago. Followed by a dozen requests for more, but alas, nada.

3 1/2 hours until my one (1) task finishes. I hope I get some more by then. My proteins are starting to tense up.
8) Message boards : Number crunching : SERVER PROBLEMS - 2. (Message 64597)
Posted 25 Dec 2009 by Profile Neil
Post:
Matterhorn


:)

The Pending work units are starting to get processed.

One by one, slowly at first...

And the red lights in the Server Status are turning green.

For the first time since 1999 when I got web-active and started working on Distributed Computing Projects, I upgraded my system to a 2nd computer and a router.

Finally.

I was watching my Users Average Credit climb from its normal plateau of 250 to an estimated re-plateau of 700 to 750, with the additional computer.

So, it got bumpy around 500 for a couple of days, when the servers glitched. A little Matterhorn Interruptus, but I'll get up to 700.

As Celery says, thanks Server Guys.
9) Message boards : Number crunching : SERVER PROBLEMS - 2. (Message 64592)
Posted 24 Dec 2009 by Profile Neil
Post:
<iam=blue>

Pending... pending...

I knew it. This is going to make a big valley in my Statistics Graph.

Yes, uploading and downloading continues. Just no credit granted.

At least, that's the worst of my problems.

<iam=still blue>
10) Message boards : Number crunching : Credit pending... pending... (Message 64591)
Posted 24 Dec 2009 by Profile Neil
Post:
Thanks, Mikey. Good link.
11) Message boards : Number crunching : Credit pending... pending... (Message 64589)
Posted 24 Dec 2009 by Profile Neil
Post:
I got about 8 work units with "Credit Pending" starting 23 Dec 2009 20:42:12 UTC

Surprised no new Forum thread mentioning it.

Interestingly, the last unit that went through Claimed 37.12, and was Granted 86.42. Usually, my work units get an extra 10%, not over 100%. It's:
http://boinc.bakerlab.org/rosetta/result.php?resultid=306805563
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=279748276

On the Server Status page,
http://boinc.bakerlab.org/rosetta/rah_status.php,
some stuff shows "Not Running," but it's difficult to correlate that chart with delays in credits being granted.

Thanks / Seasonings Sneezings.
12) Message boards : Number crunching : Problems with web site (Message 62811)
Posted 6 Aug 2009 by Profile Neil
Post:
Temporarily failed upload

8/6/2009 7:12:30 PM|rosetta@home|Started upload of lr5_seq_score12_ss5.0_rlbd_2cbm_IGNORE_THE_REST_DECOY_14613_1313_1_0
8/6/2009 7:13:21 PM||Project communication failed: attempting access to reference site
8/6/2009 7:13:21 PM|rosetta@home|Temporarily failed upload of lr5_seq_score12_ss5.0_rlbd_2cbm_IGNORE_THE_REST_DECOY_14613_1313_1_0: connect() failed
8/6/2009 7:13:21 PM|rosetta@home|Backing off 3 hr 55 min 53 sec on upload of lr5_seq_score12_ss5.0_rlbd_2cbm_IGNORE_THE_REST_DECOY_14613_1313_1_0
8/6/2009 7:13:22 PM||Internet access OK - project servers may be temporarily down.

The suspense builds. First upload failed today about 5:20 PM EST.

.
13) Questions and Answers : Windows : Server error (Message 61395)
Posted 26 May 2009 by Profile Neil
Post:
I'm the 2nd person experiencing this problem.

Yup; and I'm not uploading.
14) Message boards : Number crunching : Team Points STOLEN!!!!! Over 6 Million Points! (Message 54627)
Posted 23 Jul 2008 by Profile Neil
Post:

My team's founder stopped crunching last year. So, properly paranoid after reading this thread, I clicked on Request Foundership, to help avoid having our team stolen.

I hope the present founder becomes informed of my request and declines to allow the transfer. But I'm not hopeful.

Otherwise, if I get the Foundership, I know I'll be there to intercept and scrutinize future Foundership Requests -- especially if performed by someone who just joined the team five minutes ago.

Certainly, if Foundership is requested by an interested, at-least-occasionally-contributing member, I will be delighted to hand it over.

OK, I've also sent a PM to the Founder, with an explanation of why I did a Request Foundership, including a link to this thread.

But I wonder: Has anything changed with the Request Foundership procedure, yet? As far as I can tell, I'm going to be offered the Foundership in 60 days. Wasn't that too easy?

Thanks,
-Neil-
15) Questions and Answers : Windows : Cpu usage oscillates between 100% and 0% (Message 52488)
Posted 15 Apr 2008 by Profile Neil
Post:
Go to the "Advanced View", to the Tasks tab, what status is shown for the task that is not running?

I'm guessing perhaps "waiting for memory"? If so, check the % of memory you allow BOINC to use. This is configured in your General Preferences. If you alter the setting, you must then update to the project for the change to take effect.



Sorry, WJHalverson asked a 2nd question, but should have done so by starting a new thread, not in the middle of someone else's Q&A.

And then, Mod.Sense provided an Answer, but it's not clear which question he was answering.

Most importantly, it's not clear whether Pappanz, who began the thread, got a sufficient answer to his question.

I have tried the "Use At Most XXX Percentage CPU Time" setting. It is a good option. It does work by running Boinc for 100% for a while, and then taking a rest for a while. For instance, if you are using a Laptop computer that gets too hot while running 100% continuously, then you can reduce the heat by oscillating between 100% and 0%, with your chosen "duty factor."

Another concern might be that your computer will slow down during those instantaneous times when BOINC is running at 100%. Theoretically, this should not be a problem since BOINC runs at "Idle Priority," and at the microsecond that any other program wants some CPU power, BOINC will graciously step aside and allow anything else to take as much CPU as it wants. At Idle Priority, BOINC will only use clock cycles that no other program wanted.

An exception might be if there's another program that also runs at Idle Priority. In that case, BOINC and that program will compete. Again, if a normal program comes along, both of those Idle programs will step aside.

Whenever my computer seemed to be slowing down, I have tried Suspending Boinc, but it never helped. Boinc was never the problem. The Idle setting seems to be fine.

In my limited experience, the only possible issue is that BOINC can be set to keep your microprocessor running at 100%, which might cause a heat problem. Computers that are more modern than mine (...) have the capability to monitor and display the temperature inside the microprocessor chip. 80 degrees C is considered OK for electronics, but I think it's too hot for continuous operation. (I.e., if your microprocessor is running at 80 degrees, you might want to tell BOINC to use less CPU Percent power.) There's plenty on the subject of CPU temperature if you google.

Heat monitoring software, one CPU heat sink is better than another, water cooling, or just throttling back on your BOINC settings, etc etc.

If the CPU spends some time at 0%, then having it pop up to 100% for a short while is not a problem, heatwise, if that's a concern. In my case, my old microprocessor runs 100%, continuously. It does not run fast enough to generate much heat. I don't know if that's a good thing...

-N-
16) Message boards : Number crunching : Rosetta work units freeze up on Mac OS 10.5.2 dual core machine (Message 52486)
Posted 15 Apr 2008 by Profile Neil
Post:
- stopping and restarting the BOINC manager seems to have worked just fine.


Me too. I'm on Win XP. "Waiting to run" and "Waiting for memory" are two new somethings on my computer. They suddenly started popping up over the last day or two.

My main concern is that "stopping and restarting the BOINC manager" to give a kick to the stalled Work Unit is going to cause work to be lost. If it's been a while since a "Checkpoint" has been saved (or created, or whatever Checkpoints do...), then it could be hours of "CPU Time / Progress" that gets thrown away with capricious stops-and-restarts.

The name of one Work Unit sitting there with "Waiting to run" begins with "FRA_t038..."

-----

Scranton, Pennsylvania, where it's gray and dreary, even on the sunniest of days
17) Questions and Answers : Preferences : 100% CPU Usage (Message 51536)
Posted 21 Feb 2008 by Profile Neil
Post:
I see that Darkbydesign has given up on Boinc. I just wanted to provide an update on Boinc's CPU Usage adjustment, if that might convince him to start running Rosetta, again.

Darkbydesign asked:


> Just installed BOINC manager. And connected it to Rosetta@Home.
The thing is, it fires the PC up to 100% CPU usage, until I suspend it. The fans in the Computer is going haywire, and I'm afraid my CPU will be damaged if I leave it like this.

> Any way to fix this?

It must have been soon after you asked when Boinc implemented the ability for the user to set the percentage of CPU that is used.

The CPU usage setting is made in Boinc Preferences (by "percentage"). But in my experience, the setting is not very accurate. You can use Task Manager to monitor and confirm the percentage of CPU actually being used. Fiddle with the Percentage setting until you get the actual percent usage that you want.

In my experience, when you set Boinc's CPU usage for a limited percentage, the actual usage peaks at 100%, and then rests for a period of time, theoretically averaging out to the percentage that you chose. My problem with Window’s Task Manager is that it provides only an accurate readout of instantaneous CPU usage. But for Average CPU usage -- which is a better indicator of heat generation -- the green moving-graph in Task Manager (“Performance”) only provides a clue to Average usage. The quick high peaks and long low valleys make a mathematically-accurate measure of Average usage difficult to determine.

If you're interested in getting a numerical readout of Average CPU usage (over the range of a few seconds to a full minute or so), I recommend TaskInfo (http://www.iarsn.com/taskinfo.html)

TaskInfo costs $35. I am a satisfied customer, and have gotten good support a few times over the last 7 years. There were plenty of free updates. During that time, I think I had to send another $10 once to continue getting more years of updates.

TaskInfo has lots of juicy functions. It can display the instantaneous CPU usage for each running process, and the average usage over time for each one. You can adjust how many seconds elapse between display refreshes (every 1/2 second to 20 seconds), and how many refreshes are included in the Average reading (2 refreshes to 20 refreshes, or Infinite refreshes).

I've set my refreshes to once every two seconds. That's kind of slow, but that means TaskInfo will take less CPU usage itself, leaving more clock-cycles available for Rosetta. My "Average" readout is set to be calculated using the last 10 refreshes. This arrangement running on my advanced 1.4 GHz Celeron with massive 256 kB L2 cache shows that TaskInfo uses less than 3% (average) CPU usage, and Rosetta uses a bit more than 90% CPU usage (with Boinc Preferences set to 100%).

Since I don't understand most of the information provided by TaskInfo, I could not begin to tell you everything else that it can display. It displays system info, ram and swap usage, and file and threads info. The list of running processes can highlight a process in one color (such as green) when it opens, and can highlight a process in another color (such as red) as it closes. If TaskInfo is set to launch automatically when Windows starts up, the green and red processes blinking on and off during boot-up is psychedelic, not to mention educational.

In the Process list, each process gets its own little moving graph of CPU usage, so if Rosetta takes a dip for a few seconds, you can check the other little graphs to see which process used the clock cycles.

You can change individual process priority on the fly, and save process priorities so TaskInfo will automatically adjust them to your preference when those processes open. For instance, I like to set my media-players to High priority so movies and music will play smoothly, and everything else can take a back seat.

A bunch of other moving graphs are available. I like moving graphs.

TaskInfo offers an animated system-tray-icon that is a tiny, moving graph of CPU usage. The icon also includes separate bar-graphs of ram and Swap usage. In other words, there’s a lot of information packed into that tiny sys-tray icon.

For the graphs of total CPU usage (in the main window and the animated sys-tray icon), you can set the CPU-usage-graph to ignore Low priority processes. For instance, lets say that Rosetta is set to use 100% CPU usage at Low priority. Normally, CPU usage will always be at 100%, and the moving graph will always display 100% -- whether it's being used by Rosetta or any other process with higher priority. That would not make for a useful graph. But if the graph is set to "Ignore Low priority processes" (such as Rosetta), it will normally display only a few-percent usage. If another process starts using a lot of CPU power, it will show up easily on the sys-tray icon graph. When it's finished, the graph will drop back down, but you can be sure that Rosetta is using all those unused clock cycles, again.

If you want to see what the jump in CPU usage was all about, you can check out the main TaskInfo window, where you could see that Rosetta is presently using (for instance) 90% of the CPU. Since the moving graphs show about a minute of past information, you might see that a few seconds earlier, Rosetta’s CPU usage dropped while the antivirus software automatically updated.

There's a tool that recovers ram.

And there's lots of configuration options to display prettily the stuff you understand and to hide the stuff you don't.

TaskInfo makes a worthwhile combination with a distributed computing client.

In summary, if you stopped running Boinc because it was burning up your computer, then I'm glad to inform you that Boinc now includes effective functionality for limiting CPU usage as necessary for your laptop's comfort. Use Boinc Preferences and your computer's heat output and fan speed as a guide for adjusting Boinc CPU usage. If you set Boinc Preferences so that your computer crunches more slowly, then join a project such as Rosetta that has smaller Work Units that complete more quickly.


Just passing through.

Best luck,
-Neil-
18) Message boards : Number crunching : User Average climbing back up after outage (Message 46456)
Posted 17 Sep 2007 by Profile Neil
Post:
I migrated over to... the Human Proteome thing


Sounds beneficial.

If you get a few more uP cores up and running, and I bet you'll devote one of them to Rosetta again, someday soon.

The work-units here are so small and quick, even a Celeron 1.4 GHz with massive 256 MB L2 cache can contribute.
19) Message boards : Number crunching : User Average climbing back up after outage (Message 46445)
Posted 17 Sep 2007 by Profile Neil
Post:
User Average climbing back up after outage

20) Message boards : Number crunching : Poor Credits for w/u twice the size???? (Message 41972)
Posted 8 Jun 2007 by Profile Neil
Post:


So what y'all are saying is that my computer slows down -- relative to other people's computers -- when I'm crunching the big WUs.

But it performs "more like the average computer" with smaller WUs.

------

Since my Y axis starts in the 90s, my graph accentuates the phenomenon. Actually, the variance is only a few percent.

------

Intel is reportedly preparing another round of processor price cuts for July 22, along with introducing new processor models. Volume pricing on the Core 2 Quad Q6600 will drop significantly to $266 per unit. Note, retail prices can take a few days to adjust once new bulk prices are implemented.


I'm working on it...


Next 20



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