Advanced search

Forums : Wish list : Smaller WUs
Message board moderation

To post messages, you must log in.

AuthorMessage
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23052 - Posted: 5 Jun 2022, 1:59:15 UTC

Ones those of us without $5000 computers can actually finish. The ones given get 1 month to finish but wouldn't finish in 3 months.
ID: 23052 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23054 - Posted: 6 Jun 2022, 19:40:55 UTC - in response to Message 23052.  
Last modified: 6 Jun 2022, 19:41:49 UTC

Not sure what you are talking about. Your system should be more than capable of completing the work units well within deadlines. You have NOT described how you are running your system nor your settings. So, we can only assume you are shutting your system off for long periods of time. This is a user problem. Not a computer nor a project problem. I have much weaker systems than yours that complete work units well within deadlines. Try asking for help while providing more details of your situation so that not only us but potentially an admin can better address your issue.
ID: 23054 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23055 - Posted: 8 Jun 2022, 1:01:17 UTC - in response to Message 23054.  

I don't understand a lot of that stuff. I do have my computer sleep when I'm not on to avoid a giant electric bill and suspend when I play games that use a lot of data. Most people only have 1 computer for the things they do online, games, etc. Not too many rich enough to have a dedicated computer just for BOINC. I still have 2 hours left on one due 5/25. I don't allow more than 2 from the same project other than WCD (which has been down some time) due to it slows everything else down. 13 hours for one due 6/20 (but it is usually much longer than it says) and 14 for suspended due 6/21. Have a Milkyway also running but only 2 hours left. When it says 13 hours it does not proceed at nearly real time so it's more like 130 hours.
ID: 23055 · Report as offensive     Reply Quote
.clair.

Send message
Joined: 4 Nov 07
Posts: 636
Credit: 12,528,641
RAC: 8,690
Message 23056 - Posted: 8 Jun 2022, 17:40:08 UTC

One thing you should do is set a very small cache setting in boink manager settings ,
like 0.1 days so that you don't get work you cannot compleat in time .
you already have one marked - Timed out - no response - in your tasks page
ID: 23056 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23057 - Posted: 9 Jun 2022, 15:29:20 UTC - in response to Message 23055.  

I don't understand a lot of that stuff. I do have my computer sleep when I'm not on to avoid a giant electric bill and suspend when I play games that use a lot of data. Most people only have 1 computer for the things they do online, games, etc. Not too many rich enough to have a dedicated computer just for BOINC. I still have 2 hours left on one due 5/25. I don't allow more than 2 from the same project other than WCD (which has been down some time) due to it slows everything else down. 13 hours for one due 6/20 (but it is usually much longer than it says) and 14 for suspended due 6/21. Have a Milkyway also running but only 2 hours left. When it says 13 hours it does not proceed at nearly real time so it's more like 130 hours.


Actually, a TON of people can afford a dedicated computer to run BOINC on. Maybe not in your circle but in general many do. With so much computing hardware in the world today and the amount that is just tossed aside it is very easy to come by in many countries. The cost of electricity will vary quite a bit. Household incomes will as well. So, whether your budget allows for a computer to run 24/7 is the ultimate question. I'm unemployed and still run several servers in my home. I'm blessed with a lower electric rate. At least for now. However, the equation is simple. If your computer is off, it isn't going to process work. The deadlines are way longer than they need to be for the majority that support BOINC projects. It is you that is limiting it and making it not complete on time. You don't need a $5000 computer like you alluded to in your original post. A 10-15 year old PC can still complete these tasks with plenty of time. What needs to be adjusted is how you run them if you wish to support the project. These lengths are not unreasonable for a project.
ID: 23057 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23059 - Posted: 9 Jun 2022, 23:25:16 UTC - in response to Message 23057.  

They are unreasonable. I'm not going to have everything I do slowed down. It's supposed to use unused computer power, not take away from other things. I'm on a fixed income - $1000 a month so a 2nd computer is out of the question. My computer is only 3 years old but a bit of a potato with slowness. This is the only project I have that gets things anywhere near that long, Except Climate Prediction which gives bigger units but gives like a YEAR to finish.
ID: 23059 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23060 - Posted: 9 Jun 2022, 23:28:10 UTC - in response to Message 23056.  

One thing you should do is set a very small cache setting in boink manager settings ,
like 0.1 days so that you don't get work you cannot compleat in time .
you already have one marked - Timed out - no response - in your tasks page


I literally have no idea what that means or how to do it. I could not find any settings that said cache.
ID: 23060 · Report as offensive     Reply Quote
.clair.

Send message
Joined: 4 Nov 07
Posts: 636
Credit: 12,528,641
RAC: 8,690
Message 23061 - Posted: 10 Jun 2022, 0:52:11 UTC
Last modified: 10 Jun 2022, 1:27:07 UTC

Sorry about that , it`s not labelled as "cache" , a store of work , its just what it does ,
In boinc manager `Advanced view` [ I think you are already there ] click on the `options` tab ,
then click `computing preferences` in the drop down box ,
a window will pop up , then click the `computing` tab ,
at the bottom of that window in the `other` section you can select the amount of work to hold ,
it will be seen as "ready to start" in the `Tasks` tab of boinc manager ,
Click in the box`s to change the numbers in them ,
the ones you a looking for are :-
"store at least 0.1 days of work"
"store up to an additional 0.1 days of work"
that adds up to as aprox 5 hours of work , or one work unit per CPU per project which ever the largest
Boinc manager will work that out for itself .
Another thing you can do is set the " use at most 50% of CPU`s" that setting will also reduce the work units that get downloaded .
Then click "save" button when done .
I see you have two cosmo work units "in progress"
There are many other setting in there .
ID: 23061 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23063 - Posted: 10 Jun 2022, 16:30:26 UTC - in response to Message 23059.  

No, you are unreasonable. I understand your issue with limited income. That is a you problem not a project problem. You may just need to support a different project. And I don't know what you are doing that these work units are causing your system to slow down. But learning the settings is the first thing you should be doing. As stated above, you can reduce the number of work units your system is running by reducing the % of CPU's you allow BOINC to utilize. I'm doubting your system really needs all of the CPU cores/threads for general use. If I can game with lesser computers while letting BOINC run, then you should be able to as well. I'm seeing that you have multiple GPU's on that rig but none are all that powerful. You have NOT told us what other projects you are running and if they are utilizing your GPU's. If they are, there is a good possibility the GPU work is slowing your system down and that is easy to solve as well. Being that this computer of yours is only a Quad core with 8 threads, it certainly isn't going to be top notch for anything beyond general web browsing. It should still be capable of completing work if you aren't holding it back. The biggest changes that need to be done is to allow it to run. If you cannot afford to allow it to run, again you should consider a project with shorter work units or longer deadlines to meet YOUR needs.

When you say "It's supposed to use unused computer power, not take away from other things." that is exactly what it is doing. That isn't the problem here. The problem is that you don't have any unused computer power that is reasonable for this project because you shut your systems down or put them to sleep.
ID: 23063 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23064 - Posted: 10 Jun 2022, 17:10:22 UTC - in response to Message 23063.  

Looking at your computer here - http://www.cosmologyathome.org/show_host_detail.php?hostid=393811
It appears that you only have 16GB of RAM total for the system. 6GB of that appears to be reserved for the APU (integrated Graphics in the processor) which is far better performer than your nVidia GT 710 card you also have in the box. That leaves roughly 10GB RAM for the rest of the computer to use for everything it does including BOINC.

Why is this important? Well some GPU projects (many really) will want a full CPU dedicated to each GPU work unit you run. Especially AMD cards. But if you tell BOINC to use 100% of CPU's, then it will try and load up 8 CPU work units plus the 2 GPU's as well. They will then have to wait for CPU resources to open up slowing your work units down as well. Factor in that if you are trying to use your system while all of this is going on, resources will have to constantly switch between priorities. Thus your "other" activities may have to wait on BOINC to finish what it was doing to let go of the resources because your overall system may be over committed.

From those specs, we don't know if your system has a SSD HDD or a traditional mechanical "spinner" HDD. This will make a huge difference in how responsive your system gets. Remember...some project write to the HDD a TON and other do not. This can severely effect the responsiveness of your system. And that can also cause the system to not be able to fully utilize its other resources until the slowest component catches up. But since you didn't know what the term "cache" was really meaning, I have a feeling a lot of this stuff is over your head as well. It isn't as simple as having "free CPU resources". There is a balance to the whole PC. Every project has different needs and will behave different. So, until we have a better idea what you are doing with it....we can only make assumptions.
ID: 23064 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23065 - Posted: 11 Jun 2022, 22:03:18 UTC - in response to Message 23064.  

I was already set on use no more than 40%. Other projects I run (no problem) are Milky Way, Climate Prediction and World Community Grid (which has been on hold over a month.) I just aborted the Cosmology one due 5/25 - it had more than 24 hours run time and still had more than 2 (been saying about 2 hours left for a long time.)
ID: 23065 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23066 - Posted: 12 Jun 2022, 16:24:14 UTC - in response to Message 23065.  
Last modified: 12 Jun 2022, 16:35:37 UTC

Well, Milkyway is multi-threaded and therefore is using multiple CPU's to complete the work units. Therefore you are giving each work unit more resources there than you are here in such a short amount of time.

ClimatePrediction doesn't have consistent work either but yes they have a 1 year deadline even though you will find that they don't really give you a year....

WCG has work units that have smaller runtimes. At least when they have work available. So, you are seeing exactly what I told you needed to be done for your situation.

If you aren't going to leave the computer on for significant amounts of time, then there are just some projects that aren't for you. You really aren't contributing all that much run time in the grand scheme of things. You happen to be one of the few that really isn't worth all that much to a project because you don't return much work. Considering most people return their work units without problems easily within the deadlines it really is a You issue. Thus, you should stick to projects with smaller work units, those with crazy deadlines, or in the case of milkyway multi-threaded work.

Edit: And I'm assuming Milkyway was on CPU only with the above. Realistically, you are probably running that on your GPU's. If both GPU's are running, then you are probably using 2 of the 3 CPU threads you gave BOINC to feed both GPU's. That would then leave Cosmology with one full thread and then maybe a partial thread. Depending on how the client is handling things, the GPU work units could also be pushing those Cosmology work units to the back until they get really close to the deadline. But depending on how it has "learned" your computing behavior, it may be miscalculating how much time it needs. Can you tell us if you are using one or both GPU's? What happens when you don't run GPU's for the full deadline of the Cosmology work units? This may still just be mismanaging your resources.
ID: 23066 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23070 - Posted: 16 Jun 2022, 23:03:17 UTC - in response to Message 23066.  
Last modified: 16 Jun 2022, 23:04:36 UTC

I wouldn't know how to tell. I know there were times there were no WUs for any project except Cosmology. I also reject Milky Way units that use more than 1 CPU.
ID: 23070 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23071 - Posted: 17 Jun 2022, 0:01:40 UTC - in response to Message 23070.  

I wouldn't know how to tell. I know there were times there were no WUs for any project except Cosmology. I also reject Milky Way units that use more than 1 CPU.


Are you saying you don't know how to tell if your GPU's are being used? If so, that is simple. Just look at the work units in the BOINC client (Advanced View) and under the Status category it will state that it is using x CPU and 1 GPU if it is using the GPU. Just like if it is mutli-threaded, it will say using X CPU's.

When you say you reject Milky Way units that use more than one CPU, are you stating you abort them or what? You can actually make an app_config file to limit the number of threads to use per work unit if you need. I may be able to help with that. Let me know.

You can also edit your project preferences at Milkyway (log into your account there) to NOT use the GPU or the CPU if you aren't wanting to use it. The GPU's will use more electricity but you will find they finish work TONS faster. Many don't think running the CPU work units are worth the time and electricity.

As far as not having any work units available, WCG typically has an endless supply. This isn't typical for them. Just be patient with them. Milkyway sometimes runs out but its not often. Cosmology runs out often. Cosmology seems to have infinite work to do. However, it also depends on where you are looking to come to this conclusion. BOINC Manager will try and keep tallies of how much time you give to each project. It will then try and balance that. So, sometimes you will only run one or a few projects and sometimes you will have work from several. In the end, it balances out. So, like right now since WCG is down, it is creating a deficit that is owed to them. When they come back up, you will probably find your client focusing on just them on CPU for a while and then whatever you have going for GPU's.
ID: 23071 · Report as offensive     Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 6 Aug 08
Posts: 26
Credit: 10,247,484
RAC: 614
Message 23072 - Posted: 17 Jun 2022, 0:05:56 UTC - in response to Message 23071.  

Also, since you have 2 GPU's in your system (nVidia card and AMD APU in the processor) you can either crunch work units on both or you can use one for your general use and dedicate the other to crunching. The reason this is to be considered is because if your system is dragging to a crawl doing certain tasks while BOINC is running, it could be due to work being ran on the GPU's while you are using it. What you can do is edit the cc_config file to limit BOINC to only using one GPU of your choice. I can help with that too if you need.
ID: 23072 · Report as offensive     Reply Quote
Jonathan

Send message
Joined: 27 Sep 17
Posts: 190
Credit: 8,321,383
RAC: 0
Message 23073 - Posted: 17 Jun 2022, 0:37:58 UTC - in response to Message 23072.  

Do the legacy applications checkpoint? If so, how often? That could be another problem if your work units here don't make it to a checkpoint and then get stopped. They would have to start over from the beginning or the last good checkpoint.
ID: 23073 · Report as offensive     Reply Quote
MattShizzle

Send message
Joined: 23 Aug 09
Posts: 12
Credit: 273,766
RAC: 0
Message 23074 - Posted: 17 Jun 2022, 2:13:38 UTC - in response to Message 23073.  
Last modified: 17 Jun 2022, 2:14:58 UTC

I really don't know. I have no clue about a lot of this computer stuff (It took over a year with help to learn how to unzip a file.) I was much better with 1980s technology like Commodore 64 and VCRs. I don't know what most of the last 3 posts mean. I'm not a young man. I always aborted the MW ones that used more than one CPU but later found out how to block them.
ID: 23074 · Report as offensive     Reply Quote
mmonnin

Send message
Joined: 29 Dec 16
Posts: 53
Credit: 11,204,246
RAC: 5
Message 23076 - Posted: 19 Jun 2022, 1:43:18 UTC

So you're demanding change on topics you have no clue about. You've admitted multiple times you don't know what you're doing and have not taken initiative to learn about it. In you're other thread you just said wtf is that instead of googling it. So instead of learning and improving yourself and PCs, demand everyone else to change. Na.
ID: 23076 · Report as offensive     Reply Quote
kotenok2000
Avatar

Send message
Joined: 20 May 10
Posts: 7
Credit: 107,301
RAC: 0
Message 23082 - Posted: 23 Jun 2022, 17:34:32 UTC - in response to Message 23074.  

To block multicpu milkyway tasks go to https://milkyway.cs.rpi.edu/milkyway/prefs.php?subset=project, press edit settings and then untick "Milkyway@home N-Body Simulation"
You can also untick "If no work for selected applications is available, accept work from other applications"
ID: 23082 · Report as offensive     Reply Quote

Forums : Wish list : Smaller WUs