New workunits
Message boards :
News :
New workunits
Message board moderation
Author | Message |
---|---|
Send message Joined: 17 Apr 14 Posts: 5 Credit: 10,440,960 RAC: 0 |
|
Send message Joined: 9 Jun 12 Posts: 584 Credit: 52,667,664 RAC: 0 |
|
Send message Joined: 26 Sep 13 Posts: 33 Credit: 6,103,305 RAC: 359 |
Last modified: 11 Apr 2015, 2:50:50 UTC Thank you for your patient. I did my moving so I really was out of time. I am moved now so I can finally continue with fixing this problem. I bet it feels good to have the move finished. Have we reached the point where the computing power available to Asteroids@Home is capable of completing all the work before the next weekly WU pool? Should I increase my buffer from 2 to 10 days to get me past these periods where my machines are idle on Asteroids@Home? Can I get an asteroid named Marmot? :) |
Send message Joined: 13 Nov 13 Posts: 6 Credit: 7,693,424 RAC: 90 |
|
Send message Joined: 23 Nov 14 Posts: 11 Credit: 15,990,720 RAC: 0 |
|
Send message Joined: 26 Sep 13 Posts: 33 Credit: 6,103,305 RAC: 359 |
of course that you can, you can set it in your boinc client, in computing preferences, there is somethink like additional work buffer or somethink like that, defaultly its set to 3 days.. Sure but there are 5 projects running. If I set the buffer to 10 and there is no work for Asteroids then the other projects fill up the 10 day buffer and when WU for Asteroids finally become available Asteroids will only take a few WU down anyway. If I could adjust the buffer differently for individual projects that would help. Asteroids is my 750/1000 resource share. I'd set the other 4 projects to 2 day buffers and Asteroids to 10 days. |
Send message Joined: 1 Jan 13 Posts: 3 Credit: 2,884,728 RAC: 637 |
|
Send message Joined: 9 Jun 12 Posts: 584 Credit: 52,667,664 RAC: 0 |
|
Send message Joined: 30 May 14 Posts: 6 Credit: 31,978,085 RAC: 0 |
Last modified: 13 Apr 2015, 13:04:45 UTC |
Send message Joined: 28 Sep 13 Posts: 29 Credit: 120,811,959 RAC: 7,404 |
Last modified: 13 Apr 2015, 17:03:18 UTC Since I have about 10 machines (on a good day) running just Asteroids (filling in when I have to with World Grid, Milkyway, etc.), when I moved back to the Asteroids project, I only saw one failed download out of about 200. Thanks for your hard work, Kyong! Everything works and looks great here! When Asteroids goes dry on WU because of your work, moves, etc, I just set my day load to 1, with 0 reserve, and enable the other projects. Then, when you bring us more, I set my day load levels to 4 or 5 total, with reserve, AFTER I shut down the other projects. So it only takes a day for everything to get back to just Asteroids. I don't set day loads to go to 10 just because of unforeseen problems with any one computer going "poof" and disappearing for good - which has happened. |
Send message Joined: 26 Sep 13 Posts: 33 Credit: 6,103,305 RAC: 359 |
Last modified: 13 Apr 2015, 18:03:26 UTC @Kyong ~380 WU received with 5 d/l error. ---- @BilBg
There are a couple apps I d/l from the PortableApps store that have those features and they are task managers. Process Hacker - not obvious and there doesn't seem to be a global option. Choose process then rt-click -> miscellaneous -> reduce working set. DTaskManager - rt click on any process and choose 'trim RAM to all processes'. Thought you might be interested. ------ @DadX If you have the memory, run this project in a VM with a 6-8 day buffer. When you get WUs then rebalance the number of processors given to the Host OS or the VM. Thanks, I'll try and set that up. |
Send message Joined: 4 Apr 15 Posts: 4 Credit: 39,187,015 RAC: 2,653 |
|
Send message Joined: 1 Jan 14 Posts: 302 Credit: 32,671,868 RAC: 0 |
of course that you can, you can set it in your boinc client, in computing preferences, there is somethink like additional work buffer or somethink like that, defaultly its set to 3 days.. If you can make a 3/4 to 1 day cache, that way it checks more often. And no it would let you run Boinc the standard way with a different cache size at each project. Even a gpu is subject to the same cache size restrictions if you run both cpu and gpu units. |
Send message Joined: 19 Nov 14 Posts: 93 Credit: 30,066,240 RAC: 0 |
|
Send message Joined: 23 Nov 14 Posts: 11 Credit: 15,990,720 RAC: 0 |
The BOINC programs appear to Download files to your computer on the basis of: 1. How many days work you set each BOINC project (Max 10 days) 2. The speed of your processing.This reflects on how much work you can get through in the number of days (to Max 10 days) you preset for each project. 3. If you are processing more that One Project, you can set different amounts for each project. 4. Note that BOINC processing will process the Files with the shortest completion date, in that order. 5. Thus the files are normally processed in order of "Required By" date. 6. You are processing 5 Projects. 7. Your PC is approximately the same as mine - I have Intel 17 3.5GHz 4 core (8 engines) and a Video card similar to you, but I have 32GB of RAM. 8. You may need more RAM, that will help with processing speeds, likewise a Solid State Drive will also increase processing speeds. 9. It is possible you have not set the local processing speed - I have mine set at 90% CPU at all times, PC runs all the time, never stops. (That's the best way for long life of computers, as microdamage occurs during warmup-cool down periods, as parts expand/contract at different rates) 10. How BOINC works: The Faster you process, means you get more work in that Project. I have plenty of Work Units ahead for my 3 projects. Lets know your further progress. Cheers from a Far-Away-Land Kerry |
Send message Joined: 1 Jan 14 Posts: 302 Credit: 32,671,868 RAC: 0 |
This can pretty much be negated by changing the setting in the Boinc Manager on how often the 'checkpoints to disk' write to the disk, from the default of 60 seconds to 900 seconds or 15 minutes. With enough ram the disk is not used for the actual crunching, just the pre, post and checkpointing stuff. |
Send message Joined: 11 Jul 13 Posts: 49 Credit: 1,543,411 RAC: 109 |
Is that possible yet? I have set this to 180 seconds for years now because I tried a few years ago and higher values didn't work. Did the BOINC devs fix that issue? Would be great news. |
Send message Joined: 19 Jun 12 Posts: 221 Credit: 623,640 RAC: 0 |
It always was working, explain in more detail what you expect from this setting and what you see. I'm on BOINC 6.10.58 and "Tasks checkpoint to disk at most every XXX seconds" works as expected (and was working in older versions) This don't control how often BOINC write to disk (e.g. in files like stdoutdae.txt, client_state.xml, ...) This only tells/asks apps to postpone write of their checkpoint files till the interval is passed (the app may not obey) BOINC passes the parameter in init_data.xml in <BOINC-Data>\slots\ <checkpoint_period>XXX.000000</checkpoint_period> <disk_interval>XXX.000000</disk_interval> So the effect will be not visible for already running tasks - but if you 'Snooze' / 'UnSnooze' they will be restarted and new value will be in effect. - ALF - "Find out what you don't do well ..... then don't do it!" :) |
Send message Joined: 19 Jun 12 Posts: 221 Credit: 623,640 RAC: 0 |
8. You may need more RAM, that will help with processing speeds Not at all on this project (and many other) Can't you see how only a few MB are used per task? ... likewise a Solid State Drive will also increase processing speeds. No - ALF - "Find out what you don't do well ..... then don't do it!" :) |
Send message Joined: 28 Sep 13 Posts: 29 Credit: 120,811,959 RAC: 7,404 |
|
Message boards :
News :
New workunits