Message boards :
Number crunching :
Cons Vs. Pros - BOINC Manager(s)
Message board moderation
Author | Message |
---|---|
SilentObserver64 Send message Joined: 21 Sep 05 Posts: 139 Credit: 680,037 RAC: 0 |
What would be the cons and pros of having one project running per BOINC Manager vs. running all projects from one BOINC Manager? Will the memory, disk space, and system resources be too great a loss, to be efficient, if you ran each project seperately (Lets just start with 2 projects for this question/theory)? If you were to run 2 Managers (assuming it would not potentially affect you much), would the versions have to be different, or could they be the same? If you ran 2 Managers, would you have to create different directories for them (I assume that's how you do it), and in doing so, would this affect the allowed amount of cache, and other settings, from the projects, that you are allowed to have? (Just asking a few of many questions, to give an idea of what I am looking for.) http://www.goodsearch.com/nonprofit/university-of-california-setihome.aspx |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
Boinc Manager is just the GUI that connects to the Boinc client, what you are really talking about is running multiple Boinc clients. Claggy |
Francis Noel Send message Joined: 30 Aug 05 Posts: 452 Credit: 142,832,523 RAC: 94 |
One major Con of having multiple clients is that one would not know what the other(s) is/are doing. Each would assume to have the machine to itself and thus projects would compete for ressources. This would be very, very inefficient. Having one client allows for sharing, management and scheduling. As an example imagine running BOINC with a few projects in it and then adding some other DC infrastructure like Folding@Home on the host. BOth the BOINC projects and F@H would compete for the same CPU. mambo |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I run two instances of BOINC on 1 machines without to much of an issue. In my case it is Seti@home & Seti@home Beta. I did this to partition the resource use as I wanted instead of how BOINC was managing things on its own for the two projects. One instance I limited to 3 CPUs and the other I limited to 1 CPU. So the one project always gets full use of 1 processor in the machine. If you have 2 projects you wanted to evenly allocate resources between I think this is a better solution. Otherwise BOINC uses a debt management system. So you may only work on one projects for a few days, or weeks, before it runs any work fro the other project. In the grand scheme of things this is fine, but it can be hard to look at and come to terms with. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
SilentObserver64 Send message Joined: 21 Sep 05 Posts: 139 Credit: 680,037 RAC: 0 |
That's kinda what I was looking for HAL. I am currently running only those same two projects on this computer, and was looking to do something similar. Thanks for the heads up. http://www.goodsearch.com/nonprofit/university-of-california-setihome.aspx |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
That's kinda what I was looking for HAL. I am currently running only those same two projects on this computer, and was looking to do something similar. Thanks for the heads up. The only thing that I didn't get at first when setting it up was that you must tell the 2nd instance what port to run on or you won't be able to connect to the 2nd instance with a manager. As I don't use the BOINC installer it was easy for me to make a 2nd folder with BOINC in it and set for another project. If you are using the installer I'm not sure how it will handle a 2nd instance. As the machine I do this on is headless and I don't run BM my start up for it looks something like this. Doing this from memory I might have a few things wonky, but with that and reading the manual you should be able to get it all working as you want. D:\BOINC\boinc.exe --allow_multiple_clients --gui_rpc_port 31417 --detach_console D:\BOINC2\boinc.exe --allow_multiple_clients --gui_rpc_port 31418 --detach_console This setting is not intended for limiting CPU function, but I have found it to work better than expressing a % of processor to use. D:\BOINC\cc_config.xml uses <ncpus>3</ncpus> for S@H D:\BOINC2\cc_config.xml uses <ncpus>1</ncpus> for S@H Beta All of this kind of configuration would defiantly fall in the "advanced user" section I would think. So use at your own risk and such SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
One disadvantage I can see is while you could tell each project they had say 75% and 25% percent (assuming 4 or 8 core CPU here) that should SETI run out of work it will sit idle if you run two instances. If you run a single instance it would use the free CPU on the other project. Overall the debt method works out in the long run. There will be periods where it favors one project and then has to repay the debt later. I find running multiple projects on one instance they usually share happily and run the appropriate number of cores for each project (not withstanding outages). The current alpha test BOINC (6.13.x) is moving away from the debt idea, however it's very early days yet and not stable at the moment. BOINC blog |
Wembley Send message Joined: 16 Sep 09 Posts: 429 Credit: 1,844,293 RAC: 0 |
Would it be feasible to run one BOINC for CPU work and another for GPU work? This might kill 2 birds with one stone: 1) the problem of DCF bouncing all the time 2) the problem of CPU starving while the GPU is always fed |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Would it be feasible to run one BOINC for CPU work and another for GPU work? I was going to do that at one point, but then I switched to only requesting CPU tasks and rescheduling then to the GPU. With my slower GPU I move about 100 tasks over a week. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.