I don't see this problem in the FAQs
4 Sep 2010 2:32:03 UTC
Topic 195310
(moderation:
If there is a FAQ which addresses this problem please let me know.
Background: I use linux. I started in seti in 2000 and einstein in 2005. I had no problems with either of them.
In 2008 I dropped out of both. I rejoined seti late last year and no problems. I rejoined einstein last week and I have a problem. Obviously it has the same computing preferences as seti but just to be sure I "faked" a preference change via seti and saw the "update" in the stdout log. So I expect einstein to work as unobtrusively as seti.
Problem is it does not. When a seti work unit is running I never notice any change in computer performance. When an einstein WU runs the computer is bogged down as in incredibly slow. This is not a priority as in a day or so this single core will be online mainly to do BOINC stuff but the problem may migrate to the new four core. (I have another four core which drives my TV and renders animations and does BOINC work but not having used it as my primary day to day computer I do not know if it is affected by einstein in this same way. So I may still need a solution to this problem.)
I run it as "nice boinc" just to be sure. The top reported usage is around 90% for seti and around 35% for einstein. That does not make sense.
Any suggestions?
I don't see this problem in the FAQs
)
How much RAM do you have? Einstein uses more RAM than my other 5 BOINC projects. Since I have a SuSE 11.1 Linux pae this is not a problem for me as I have 5 GB RAM. Einstein and 4 other projects run on a single core (I have 2), AQUA uses both cores of my Opteron 1210.
Tullio
BTW: I apparently was root
)
BTW: I apparently was root when I attached to einstein for some reason. I do know better that to be root. Otherwise I am Matt_Giwer.
RE: How much RAM do you
)
This machine has only 512M but with 32 bit linux, Redhat but the kernel is the thing so I don't see the distro would matter. This is an AMD about 3-4 years old among their earliest 64 bit chips -- though loaded with 32 bit for the limited memory.
Question, there are three projects and I checked all three. If there is a difference in memory usage do you know which is least? Or more simply, which is the original project as I had no problems with that one.
I have 2 projects running
)
I have 2 projects running now. SETI@home takes 62 MB RAM, QMC@home (a very reliable project, always giving work) 14 MB. Einstein takes 258 MB, QuantumFIRE 10 MB, AQUA 40 MB, climateprediction.net 106 MB with very long deadlines. LHC@home is not giving me any work. I think that 512 MB RAM is too little for Einstein.
Tullio
RE: BTW: I apparently was
)
You can change the user name in your account (other account info). That shouldn't even disturb the statistics sites, as the primary key is the email address.
I'm running Einstein with 512 MB (494.42 MB because of graphics chip) on single-core windows XP without problems.
If the quad also gets sluggish, you could try to reduce the number of cores used by BOINC:
BTW, the only reported task of the single core ended with "process got signal 11" (see this FAQ entry)
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
Hi! Linux is quite good at
)
Hi!
Linux is quite good at running applications in low priority mode (used by BOINC apps, called "niceness") so you would not feel a slowing down of non-BOINC applications.
Either the box is "swapping" (constantly writing to disk because it has to temporaily write stuff to disk to free some RAM) or there is another CPU intensive application running on the single-core because the task that ended with a signal 11 error consumed ca twice as much wall-clock time than CPU time.
using "top" in a console window while the Einstein@Home app is running will tell you much about this: near the top of the display, you'll see how much of the swap space is being used at the current time (also look at the "HD-is-used" light on the PC to check for disk activity. top will also tell you which other applications are getting CPU time at the moment.
HBE
I have been running Einstein
)
I have been running Einstein and SETI on a PII CPU with RAM starting from 128 MB and reaching a maximum of 384 MB. Linux, of course. Then I bought a SUN WS with an Opteron 1210 with 512 MB RAM. I gradually brought it to 5 GB (it could go to 8 GB maximum). Most of the time more RAM is used as a disk cache, so there is very little disk I/O. DDR2 RAM used to be cheap (i paid 52 euros for two 2 GB modules) but it is always a good investment.
Tullio
RE: RE: BTW: I apparently
)
Thank you and done!
That is close enough to what is giving me the problem. If a suggestion works or I figure it out I will post.
RE: ... That is close
)
OK I got the problem under control by detaching the problem machine from einstein.
I look at a machine where einstein works fine and see it is using 6.6% of memory which has 4GB of RAM. I look at the problem machine running seti and see it is using 5.9% of 512MB of RAM. Simple extrapolation says einstein would have only worked invisibly if it had 8x6.6% or 52.8% of 512 MB. I'd say that is the obvious suspect. My usually running apps and OS overhead only run about 30% of RAM so tight but still not an obvious disaster.
That doesn't explain why the same RAM running XP has no problems