Message boards : Number crunching : Computer become unresponsive
Author | Message |
---|---|
TestPilot Send message Joined: 23 Sep 05 Posts: 30 Credit: 419,033 RAC: 0 |
Suddenly. Sometimes I had to wait few seconds before I could see mouse pointer moving. And sure, if I quite or snooze BOINC, everything immediately back to normal. Only Rosetta installed under BOINC. Current tasks: https://boinc.bakerlab.org/rosetta/result.php?resultid=367823012 and https://boinc.bakerlab.org/rosetta/result.php?resultid=367829870 I could probably abort those tasks and see what happen... TestPilot, AKA Administrator |
EdMulock Send message Joined: 14 Mar 06 Posts: 30 Credit: 2,347,485 RAC: 0 |
I had to reset the project. Ran out of memory. Using no CPU. |
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,135,082 RAC: 4,703 |
I had to reset the project. Ran out of memory. Using no CPU. There are reports of Rosie units taking up to 1.5gig of memory per unit, only one of your pc's seems to have 2gig of memory with most having 256meg of memory! I think you are simply running out of memory. |
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,135,082 RAC: 4,703 |
Suddenly. Sometimes I had to wait few seconds before I could see mouse pointer moving. Sadly I think you too are running into memory problems and the fact that Boinc does not like to give up crunching as easily as it used to. The machine you pointed to has 2gig of ram but is a dual core pc, so if you are running 2 Rosie units and they are each taking 1.5gig of ram you are into the hard drive as spare ram and this makes the pc slower. I have 2 dual core pc's running Rosie and am thinking of upping the ram in one of them to 4gig to let it breathe again, the 2nd pc already has 4gig of ram in it. |
joseps Send message Joined: 25 Jun 06 Posts: 72 Credit: 8,173,820 RAC: 0 |
Suddenly. Sometimes I had to wait few seconds before I could see mouse pointer moving. I experienced your problem on my 2 computers, I was able to solve it by editing the individual computer preference on usage as follows: 100% of processors and, 70% of cpu time. As a bonus, it lowers the core temperatures while using the Intel supplied stock cpu cooler (HSF). Now average core temp is 55*C. joseps I turned off my 5computers when I went on vacation. When I return today, I can not upload work. Need work units to run computers. joseps |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
Until the issue with extreme memory usage is addressed, there are a few things one can do to help their machine run smoother. All will cut the amount of credit potential, but help your machine environment. 1) reduce the number of CPUs BOINC uses. If you have a 2CPU machine and only allow BOINC to run on 50% of the CPUs, it will only start one task rather then 2, and thus the memory footprint for Rosetta will be half of what it would be running 2 at a time. 2) reduce the amount of memory BOINC is allowed to use when the machine is in use. It may still be sluggish when you first return to using the machine, but should then help it devote more of it's resources to your current applications. 3) attach to additional BOINC projects. Review the memory requirements of other projects and attach to some that interest you and have lower memory requirements. If you give the projects equal resource shares, then in general BOINC will run one Rosetta task and one task for the other project. Thus running a potentially high memory task with a task that uses less memory. This is a way to keep all of the cores busy, yet still limit the number of simultaneous Rosetta tasks, as in #1. Rosetta Moderator: Mod.Sense |
nusbaumc Send message Joined: 4 Jun 10 Posts: 11 Credit: 3,747,410 RAC: 0 |
Rosetta uses gobs of memory - 5-8GB on my i7 is pretty normal. Unless you have tons of RAM you'll probably see a big performance hit from it unless you modify your BOINC settings. Or you could install more RAM if your motherboard supports it. I have 12GB installed on my main desktop and the only time I have to stop BOINC is when I'm gaming. |
Chilean Send message Joined: 16 Oct 05 Posts: 711 Credit: 26,694,507 RAC: 0 |
|
Greg_BE Send message Joined: 30 May 06 Posts: 5691 Credit: 5,859,226 RAC: 0 |
take a look at the 2.15 thread. seems they discovered or are responding to the large memory usage in 2.15. mentioned something about a revert. |
VrskoD Send message Joined: 5 Jan 09 Posts: 5 Credit: 325,841 RAC: 0 |
Suddenly. Sometimes I had to wait few seconds before I could see mouse pointer moving. Hi, I experienced similar problems. On all PC's run Malariacontrol and Rosetta projects simultaneously. Malaria with no problems, Roseta ? Working on 2 Hewlett-Packard PC's each with 2 core Intel pentium4 processors and 2GB of RAM, Win Vista-32bit - both begun to be slow and later completely frozen. I tried to change settings, but finaly I had to restart PC's and abort Roseta. Onlz abort of Roseta returned to normal. This is the list of working units I had to abort. 12.10.2010 10:57:42 rosetta@home task lr5_combined_torsion_it01_run01_A_rlbd_2iiy_SAVE_ALL_OUT_IGNORE_THE_REST_DECOY_18669_2851_0 12.10.2010 10:57:42 rosetta@home task mem_widd_run01_B_2rlf_SAVE_ALL_OUT_IGNORE_THE_REST_22156_46669_0 12.10.2010 10:57:42 rosetta@home task oct_refactor_MFR_ccr19_BOINC_abrelax.score12.fastrelax.v4_SAVE_ALL_OUT_22356_602_0 12.10.2010 10:57:42 rosetta@home task 2occ_DISULF_NOMATCH__BOINC_abrelax.score12.fastrelax.v4_SAVE_ALL_OUT_22351_741_0 12.10.2010 10:57:42 rosetta@home task mem_abinitio_bench_run01_B_BRD4_SAVE_ALL_OUT_IGNORE_THE_REST_22295_21069_0 12.10.2010 10:57:42 rosetta@home task mem_abinitio_bench_run01_B_RHOD_SAVE_ALL_OUT_IGNORE_THE_REST_22295_23905_0 What is worse, similar situation I experienced with my "best" PC - 4 core AthlonII-X4, 8GB of RAM, Win7-64bit. Malariacontrol worked for many days with no problem. Short after instalation of Roseta (1 core Roseta, 3 core Malaria) computer begun to freeze temporarily. I tried to change settings to very conservative values with no effect. Finaly it crushed definitively - blue screen with information something like "It is impossible to start Windows, call your PC provider...". I tried to repair the instalation of Win7 - no effect. I tried to start Linux Ubuntu from CD - no problems. I had to change some memory settings (down frequency, 2T...)and reinstall Win7. It seems to work - but complete reinstalation of the system needs some time and I was not able to finish it until now. I am affraid, I have to leave Roseta a work on Malariacontrol only. This PC is new, not overlocked, BIOS fail safe settings, memory and HD tests normal. It seems there are some problems with RAM, but - who knows. Anybody have some suggestions? |
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,135,082 RAC: 4,703 |
Suddenly. Sometimes I had to wait few seconds before I could see mouse pointer moving. Maybe...first you cannot set Boinc to do this "(1 core Roseta, 3 core Malaria)", Boinc does not work that way. What you can do is set Boinc to 75% one project and 25% for another project and hope it works out that way. But Boinc cannot be fine tuned to work like that. Second both Malaria AND Rosetta are experiencing some memory issues right now with their workunits. Rosetta units can take up to 1.5gb each and the Malaria ones can take up to 512mb each, both projects are working on this. So if you are running 2 Malaria units on your dual core you only have 1mb of ram left for other stuff INCLUDING Rosetta which at 1.5mb for each unit means you are short and using the hard drive as an active memory device, resulting in a very slow pc! On your quad core with 8 meg of ram, you are running 4 Malaria units using 2 gb of ram, leaving 6gb free, now you add in 4 Rosetta units and you have 1.5gb of ram usage for each unit and voila no memory left, meaning again the hard drive is being used as an active memory device making for a very slow pc! Remember all this memory is the total amount of memory and we have not even talked about the OS or whatever else runs on those pc's. |
TestPilot Send message Joined: 23 Sep 05 Posts: 30 Credit: 419,033 RAC: 0 |
I made BOINC project settings somewhere on this web site set to use maximum 50% of RAM and max 50% of swap file. That stalled Rosetta computations for a wile, but I did not mind it. If my computer (2 gig RAM dual core) is not powerful enough to run Rosetta - so be it. And few days ago computations resumed to my surprise. I'm crunching for over five years, and will be around for years to come. But I'm not a fanatic - making sure that computer performance is not affected (or barely affected by BOINC) is much more important for me. But I don't really mind it to crunch in the background if it can! |
VrskoD Send message Joined: 5 Jan 09 Posts: 5 Credit: 325,841 RAC: 0 |
Second both Malaria AND Rosetta are experiencing some memory issues right now with their workunits. Rosetta units can take up to 1.5gb each and the Malaria ones can take up to 512mb each, both projects are working on this. So if you are running 2 Malaria units on your dual core you only have 1mb of ram left for other stuff INCLUDING Rosetta which at 1.5mb for each unit means you are short and using the hard drive as an active memory device, resulting in a very slow pc! On your quad core with 8 meg of ram, you are running 4 Malaria units using 2 gb of ram, leaving 6gb free, now you add in 4 Rosetta units and you have 1.5gb of ram usage for each unit and voila no memory left, meaning again the hard drive is being used as an active memory device making for a very slow pc! Remember all this memory is the total amount of memory and we have not even talked about the OS or whatever else runs on those pc's.[/quote] Hi, Problem seems to be solved. I had to cease my activities on Roseta for some time and I spent this time for checking of my Athlon X4 / 8 GB RAM PC and found out - hard disk (older IDE Western Digital one from previous computer) presented on most of tests huge amount of unreadable segments. So I backed up important files, bought and installed new SATAII disk, reinstalled Win7/64 and everything seems to be OK. I think, the problem was in communication between program and the hard disk which was not in the best conditions. System stopped and /or crushed when it tried to read from corrupted segments? By the way, I am not a PC expert - I only have electronics and later computers like a hobby for more than 40 years of my live - but I think, that 2,9 GHz, 4 core Athlon with 8 GigaBytes of RAM must be enaugh for system Win7/64, some small programs, and one Rosetta a 3 Malaria concomitant computations. Thanks to all. |
mikey Send message Joined: 5 Jan 06 Posts: 1895 Credit: 9,135,082 RAC: 4,703 |
I think the new Sata disk is MUCH faster than the old IDE drive and that is helping too. So not only were the bad sectors slowing you down but the drive speed itself was slower. Now if you really want to speed up the system try an SSD drive. That stands for Solid State Drive meaning no moving parts and VERY fast, it also means VERY expensive!! The prices ARE coming down though and as the drives get bigger the costs should come down even more, in the future it may be the way to go for the next step in the speed race. |
Message boards :
Number crunching :
Computer become unresponsive
©2024 University of Washington
https://www.bakerlab.org