Message boards : Number crunching : Requesting 0 seconds of work ...
| Author | Message |
|---|---|
|
Hello all, | |
ID: 6675 | Rating: 0 | rate:
 /   Reply
Quote
| |
 Paul D. BuckSend message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level  Scientific publications      | |
|
Change to BOINC Manager version 6.5.0 ... | |
| ID: 6682 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Thanks for your quick help. | |
| ID: 6697 | Rating: 0 | rate:
/ Reply
Quote
| |
|
| |
| ID: 6700 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
|
6.5.0 is also rated as "unstable" though it is the "best" version that has caused the fewest problems for the most people. | |
| ID: 6702 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Rick A. Sponholz   Send message Joined: 20 Jan 09 Posts: 52 Credit: 2,518,707,115 RAC: 0 Level  Scientific publications  | |
|
Thanks for the workaround, but I want to split my GPU between SETI and GPUGRID. Any other suggestions? | |
| ID: 6720 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
Thanks for the workaround, but I want to split my GPU between SETI and GPUGRID. Any other suggestions? Set them to equal share ... or whatever proportion you desire. | |
| ID: 6724 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Rick A. Sponholz Send message Joined: 20 Jan 09 Posts: 52 Credit: 2,518,707,115 RAC: 0 Level Scientific publications | |
|
At least for me, setting them at equal, or even 10 times higher for GPUGRID hasn't worked. | |
| ID: 6729 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
At least for me, setting them at equal, or even 10 times higher for GPUGRID hasn't worked. Ok, it is not clear to me what you are trying to do, and what the problem is that you are seeing. Can we start at the beginning? How much time do you want to do with GPU Grid vs. SaH? What other projects are attached? What else do you see? Lets start there and try again. :) | |
| ID: 6730 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Rick A. Sponholz Send message Joined: 20 Jan 09 Posts: 52 Credit: 2,518,707,115 RAC: 0 Level Scientific publications | |
|
My "normal" settings are | |
| ID: 6736 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
My "normal" settings are Ok, sadly, it is not likely to do what you want with the settings you have. The problem is that you want to stress SaH and not GPU Grid and there is no distinction in the BOINC Managers for CUDA vs CPU work though SaH has kinda a work around. My suggestion is that you pick one system and attach it to GPU Grid, if needs be set up a specific profile for it and set the resource share for GPU Grid to 200, with BOINC Manager 6.5.0 you will reliably fetch work (well, it works for me on 3 systems reliably since december something) ... the other systems set to SaH ... with what ever configuration you want for them as far as CPU vs GPU work ... Until there is a version of BOINC that distinguishes between CUDA and CPU work and fetches reliably you are not going to be able to get there from here ... the main problem is that you want to stress SaH and only do a little on other projects ... | |
| ID: 6741 | Rating: 0 | rate:
/ Reply
Quote
| |
|
@Paul: Thanks Paul for your suggestions to my initial question! | |
| ID: 6742 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Rick A. Sponholz Send message Joined: 20 Jan 09 Posts: 52 Credit: 2,518,707,115 RAC: 0 Level Scientific publications | |
|
Thanks Martin, | |
| ID: 6748 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
@Paul: Thanks Paul for your suggestions to my initial question! The largest difference I have seen is that you don't need a configuration file to get BOINC to properly "see" the GPUs and to use them. With 6.4.5 you seem to need a file to tell BOINC to use the GPUs ... The version deltas are not well maintained so it is hard to say what is in the various versions. Sometimes you can find notes on the BOINC web site's forums as to some of the builds. But the good news is that you are up and running. With the 9800 GT in my system I usually have one running and one or two spares. Real run times are between 12 to 30 hours ... | |
| ID: 6752 | Rating: 0 | rate:
/ Reply
Quote
| |
|
| |
| ID: 6753 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
|
@ms113 | |
| ID: 6771 | Rating: 0 | rate:
/ Reply
Quote
| |
|
@Paul | |
| ID: 6776 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
@Paul I conceeded victory to my failing heath ... I am the proud owner of enough psychological problems to keep 10-15 theripists busy 24/7 ... they however, keep wanting to send me to the neurologists as several of the symptom sets don't fit into their systems, the neurologits send me right back ... after at least 20 drug trials, none of which bothered the more serious symptoms I surrendered on that front too ... :) The side effects were worse than the problems ... Since you asked ... :) My laundry list includes Asperger's Syndrom (autism with high intelligence), suicidal depression, anxiety disorder, panic disorder, possible agoraphobia, possible OCD, possible rapid cycle bipolar (cycles take hours to days vs. days to weeks to months), ... and some other things ... the good news is that the autism sort of cancels the sucidal tendencies in that I have the thoughts all the time and the logical mind just considers it and rejects it as soon as it comes up ... or carefully considers all the aspects in such detail that it winds up being who cares ... But, enough ... mental illness won ... I husband my strength and putter with my computers, on bad days I don't even do that much ... Sorry again, but I really miss the whole foretime ... Aside from the above, the mavens that run BOINC have made it quite clear that they are not at all interested in my input or my work ... If you have the time and want to amuse yourself you can goto almost any of the really old projects and find my "footprints" all over there ... just trace back to the earlier messages and you can see ... But there is also a lot that you cannot see ... unless you look carefully and watch over time. You also have to watch the mailing lists ... and see the responses, or lack thereof to posts ... In many cases, serious people are not listened to, and you can decide for yourself if I should or should not be in that group. But, it is hard to get the attention of people in most projects. GPU Grid is different, FOR THE MOMENT, but so was Rosetta in the beginning ... now you are lucky to see feedback once a month from one of the project principles ... Remember, moderators for the most part are participant volunteers ... not project paid staff ... In my humble opinion, boinc is a serious projekt (isn't it?) This is a classic issue I tried to convince the "powers that be" could be easily fixed and was either ignored or told that it would not happen ... The "best" documentation is: - Unofficial BOINC Wiki (UBW) Old with few updates since I stopped working on it two plus years ago when I took a sabatical from BOINC. - FAQ Service much of the same content here will be found in the UBW but this is still live, though I still question why the author had to keep his content out of either Wiki, his excuse (when he explained it to me) was that Wiki markup was too hard. - Trak Wiki where the "design" documentation, such as it is, is kept. This also contains the rules and procedures to stand up a project. Bug reports are filed here for a more convient way to ignore them in bulk ... - E-mail list sign up here is where you sign up for emails from the general lists ... I watch the "projects" and "Dev" lists and post on one or the other often ... rarely get an answer... :) As an example, the question was raised on GPU detection, I did some research on the three APIs and posted a message with the results ... no acknolegement if what I did was useful, or not, though there is a hint in the latest release notes that indicates that it may have been, though there is no way to be sure ... it could also just be coincidence ... - UCB User Manual which I think is very much lamer than what I did in the UBW because of the insistance of Dr. Anderson that the documentation of BOINC should not include any information about projects... which never made sense to me in that the whole reason for BOINC and the usual interaction of a participant with BOINC is in relationship to projects ... compare and contrast this Wiki with the UBW's treatment and decide which approach is better ... - BOINC message Boards where information about BOINC and only BOINC is posted. This is also where the "release notes" seem to be collected. I tried really hard to find answers to several, sometimes technical, questions about the project and its context (e.g. special applications ...) before asking questions in a forum. Think of BOINC as a game console. The science applications from the project as the game cartridges. This mind model was what I proposed before BOINC Beta ended as it is simple direct and easy to grasp. I was told I did not know what I was doing ... years later we are still trying to explain what a core client is vs BOINC Manager vs. science application. The Core Client (aka the Daemon, there was a time when this was the official name, now lapsed again) is the "engine" that actually does the work of BOINC on your system. It is a framework of routines that allows the projects to concentrate on development of the science application and leaving all of the drudgery of the "housekeeping" to the BOINC Client. This manages the scheduling of the tasks, the running of the science applications, and the communitcations to the servers. The BOINC Manager is just the GUI that allows you to work with the core client, that little engine that is actually doing the work ... The Science Applications are the programs that actually do the science. Each project writes their own and they are the real point of BOINC. They are the things that take up the CPU or GPU time. oh, and the screen savers and graphics are a separtate program provided by the project (or not) ... The change log though the next one will likely be a different sticky later on ... oh, you have to create an account here to make it so it tracks what you have read ... Even to find simple Downloads - I mean I could not find Version 6.5 for linux using google, boinc.berkeley.edu, before j2satx told me (above in this thread). I always just google BOINC and then go from there... but I agree the front page of the BOINC site sucks. But, as I said, criticism and input is only welcome from some. If you are not on that list, well, ... Maybe I am to .. whatever. Nope, this is the classic mistake people make... I don't understand "x" so I must be an idiot ... not in the slightest. I do computers, I used to be a systems engineer specializing in the logical design of relational database systems. A specialty so narrow most people have never heard of it. Perfect for working on BOINC ... when I was in the Navy I was in electronics repair using Automated Test Equipment and as part of my duties I did technical publication review ... perfect for working in BOINC ... in the Navy and in civilian life ti also taught in technical schools and college ... perfect for BOINC ... Anyway, back to the point. I don't know what you do for a living, but I am sure that I would not be good at it. I am good at teaching, technical writing, and figuring out systems and identifying their flaws. But, 99% of mathematics is over my head ... I asked in ABC's NC forum about the explanation of the project and you can see for yourself the same "typical" treatment of someone that does not understand ... I on the other hand think that if there is a failure of communication it is the failure of the person WITH the knowledge when the explanation does not make sense. Interestingly enough this is this quote in the news on ABC: 7 October 2008 Along with this paper which I don't think was on the site at the time I asked the questions. They certainly did not point me there when I asked ... You seem to be very experienced and interested in this Projekt (talking about response and responsiveness). Maybe my typically long and boring dissertation will help ... if not, ask specifically, and I will try again ... I still love to teach though I cannot walk reliably ... and this is the only place (the BOINC Universe) where I can sometimes still manage to find a student that is willing to listen ... :) P.S.: I opened another thread ("CPU Utilization varies ...") .. where I also did not find an answer yet . Maybe you can take a look. Did not see it yet, so I will go look. If I don't know, I usually won't make a comment. But the only stupid question is the one you don't ask. Oh, and sometimes I am too down and only skim and don't speak up ... in those cases you can send me a PM or better still an e-mail ... those I am anal about ... | |
| ID: 6782 | Rating: 0 | rate:
/ Reply
Quote
| |
|
@Paul | |
| ID: 6816 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
|
Well, the good news is that you can print it out and spend some time with it ... | |
| ID: 6823 | Rating: 0 | rate:
/ Reply
Quote
| |
Message boards : Number crunching : Requesting 0 seconds of work ...
