1)
Message boards :
Number crunching :
Collatz and MW togetther
(Message 36568)
Posted 16 Feb 2010 by j2satx
Post:
Just for the "halibut", try reducing your share for collatz and MW to about 1/10th of your CPU projects.
You can use smaller numbers, rather than thousands to get the same percentages.
|
2)
Message boards :
Number crunching :
Collatz and MW togetther
(Message 36561)
Posted 15 Feb 2010 by j2satx
Post:
I have Intel quad and ATI 4850
W7 RC 64 with 9-11 video driver
BOINC 6.10.18
PrimeGrid share 50 (project selected runs on CPU only)
Collatz share 5 (GPU only selected on project)
MW share 5 (GPU only selected on project)
cache 2.25 (usually 0, but participating in PG challenge)
When MW completes WU, a new WU downloads if cache is deficient.
The cache has WUs from all three projects...400 PG, 23 MW and 129 Collatz (PG is limited by project to 400).
I did not start from scratch.....I was running PG and MW, then added Collatz to see how it reacted after reading this thread. Collatz downloaded WUs and it all seems to be working automagically.
What else can we compare to help make sense of the issue?
|
3)
Message boards :
Number crunching :
more wus generated: nm_test1
(Message 6405)
Posted 22 Nov 2008 by j2satx
Post:
So I set my prefs to allow test apps, remove app_xml file, reset project and reboot/restart BOINC.
I am still getting 1.22 WU's.. is this correct ?
We're still generating workunits for 1.22, so you should get both. I only generated 100 new workunits with nm_test1, because i didnt want to flood the system with bad workunits (if they didn't work).
Since I haven't heard any complaints about the new batch, I'm going to create a bunch more and see how these do. They'll be under nm_test2.
test2 WUs look like they are running fine on Windows XP.....all are pending so far, so I can't tell how much they pay.
|
4)
Message boards :
Number crunching :
@MW staff: credit limit and credit level
(Message 6305)
Posted 19 Nov 2008 by j2satx
Post:
We're also have a strategy in the works to check results (without redundancy), which will flag suspicious clients for further examination. With this if we see any apps generating bad results we'll stop awarding them credit.
Better to stop giving them WUs.
|
5)
Message boards :
Number crunching :
Credit Calculations.
(Message 4419)
Posted 23 Jul 2008 by j2satx
Post:
I'm not sure what happen in Nov 2006 besides one of my hosts crashed and then the problem bagan. All my old projects had there CPIDs stolen away in this transition. :D What I can't figure is how can a host have different CPIDs (4 in most of them) and control/change the Project CPIDs when they connect. I thought the Projects Servers controled the hosts. The one good thing is I got down from 10 CPIDs to 4. I even detached from BAM to see if CPID Problem would clear up but it didn't
When BAM retires a project they detach all the hosts from the project and it keeps detaching them for an eternity. An example is ABC@Home Beta. It only has work as needed (Nov 2007 was last WU I finished) & it wasn't reachable for a while so I guess that is why they retired it. I attach to it ever once in a while to see if it up and as soon as my host syncs BAM it detached. If It does give out any work, I'll have to Detach from BAM to run it.
Hmmm...
Now that is interesting. I seem to remember loosing a couple of host CPID's on BOINCStats around then, but everything resync'ed on the same Account CPID.
The only side effect was a couple of hosts lost their graphical data for everything from before the event.
Let me check my logs and get back to you on this.
Alinator
I tried everything to solve the problem. Including begging the Dev (which is a low priorty issue). I would like to see if they could straighten out the problem, but I think I going to detach 15 hosts from all the projects, delete/clear out all BOINC. Then go into one node and change all the CPID to my main one (or suspose to be my main one) then run it until I get all the projects to Sync with my host. Then put the others back online again.
This is what works for me. I always use one project on every computer. I always connect the computer to that project first. I then shut down BOINC, restart and connect to the other projects. I do not have any problems with CPIDs.
|
6)
Message boards :
Number crunching :
Os-X Time comparisons PPC vs. Intel
(Message 3430)
Posted 30 Apr 2008 by j2satx
Post:
A couple of questions (very simple). Firstly, are the optimized apps officially endorsed and/or provided by the MilkyWay administration? Secondly, will the project administration notify all project participants that optimized clients are available for use, along with instructions on how to install them? Then there's the question of whether optimized apps give some platforms (ie operating systems and CPUs) advantages over others based on particular optimizations that are being done. Like are the optimizations being done to fully take advantage of each platform's capabilities.
Basically this is what I see it comes down to. Those using non-optimized apps will get less RAC (recent average credit) because the workunits will take longer to complete, while the ones using optimized apps will end up with much more RAC while crunching less. Am I missing something or oversimplifying here?
[EDITED]
The optimized app are not endorsed by the University. However, this project is open source so we will not restrict people from doing this. Also, the administration says we can't "notify" people, but if Cruncher wants to make a thread to list optimized apps, I will be happy to make it a sticky.
If you allow optimized apps, who is checking to see that the results are the same as results from the "official" app?
That's exactly the problem we have. We are working on a way to verify while still allowing optimized apps.
Don't you think the verification process should come before allowing optimized apps?
The project needs to be science, not magic.
Well we technically can't disallow optimized apps. In theory, anyone can make an .exe file and install it on their computer and send results in to any BOINC project. I believe this is how it works, and until BOINC makes a more secure environment for executables, we are at a loss. If my understanding is wrong, please correct me.
I can't correct you because I don't know the rules on "open" source. I do know that I stop participating in projects that don't "endorse" the optimized apps that they allow as providing correct and accurate results.
I didn't know that the science apps had to be "open" source....I thought it was just the BOINC provided software that was "open" source.
You're correct. Not all science apps have to be open source. However, since we're a university, we're all about sharing our ideas and bettering the software as a whole. So, if there's a way to make something better, even if we don't do it, we still like to see it done.
I'm all for making it better also, as long as it is still accurate and creates results the project will verify as acceptable. Faster inefficiencies and incorrect results are not better science.
|
7)
Message boards :
Number crunching :
Os-X Time comparisons PPC vs. Intel
(Message 3426)
Posted 30 Apr 2008 by j2satx
Post:
A couple of questions (very simple). Firstly, are the optimized apps officially endorsed and/or provided by the MilkyWay administration? Secondly, will the project administration notify all project participants that optimized clients are available for use, along with instructions on how to install them? Then there's the question of whether optimized apps give some platforms (ie operating systems and CPUs) advantages over others based on particular optimizations that are being done. Like are the optimizations being done to fully take advantage of each platform's capabilities.
Basically this is what I see it comes down to. Those using non-optimized apps will get less RAC (recent average credit) because the workunits will take longer to complete, while the ones using optimized apps will end up with much more RAC while crunching less. Am I missing something or oversimplifying here?
[EDITED]
The optimized app are not endorsed by the University. However, this project is open source so we will not restrict people from doing this. Also, the administration says we can't "notify" people, but if Cruncher wants to make a thread to list optimized apps, I will be happy to make it a sticky.
If you allow optimized apps, who is checking to see that the results are the same as results from the "official" app?
That's exactly the problem we have. We are working on a way to verify while still allowing optimized apps.
Don't you think the verification process should come before allowing optimized apps?
The project needs to be science, not magic.
Well we technically can't disallow optimized apps. In theory, anyone can make an .exe file and install it on their computer and send results in to any BOINC project. I believe this is how it works, and until BOINC makes a more secure environment for executables, we are at a loss. If my understanding is wrong, please correct me.
I can't correct you because I don't know the rules on "open" source. I do know that I stop participating in projects that don't "endorse" the optimized apps that they allow as providing correct and accurate results.
I didn't know that the science apps had to be "open" source....I thought it was just the BOINC provided software that was "open" source.
|
8)
Message boards :
Number crunching :
Os-X Time comparisons PPC vs. Intel
(Message 3424)
Posted 30 Apr 2008 by j2satx
Post:
A couple of questions (very simple). Firstly, are the optimized apps officially endorsed and/or provided by the MilkyWay administration? Secondly, will the project administration notify all project participants that optimized clients are available for use, along with instructions on how to install them? Then there's the question of whether optimized apps give some platforms (ie operating systems and CPUs) advantages over others based on particular optimizations that are being done. Like are the optimizations being done to fully take advantage of each platform's capabilities.
Basically this is what I see it comes down to. Those using non-optimized apps will get less RAC (recent average credit) because the workunits will take longer to complete, while the ones using optimized apps will end up with much more RAC while crunching less. Am I missing something or oversimplifying here?
[EDITED]
The optimized app are not endorsed by the University. However, this project is open source so we will not restrict people from doing this. Also, the administration says we can't "notify" people, but if Cruncher wants to make a thread to list optimized apps, I will be happy to make it a sticky.
If you allow optimized apps, who is checking to see that the results are the same as results from the "official" app?
That's exactly the problem we have. We are working on a way to verify while still allowing optimized apps.
Don't you think the verification process should come before allowing optimized apps?
The project needs to be science, not magic.
|
9)
Message boards :
Number crunching :
Os-X Time comparisons PPC vs. Intel
(Message 3421)
Posted 30 Apr 2008 by j2satx
Post:
A couple of questions (very simple). Firstly, are the optimized apps officially endorsed and/or provided by the MilkyWay administration? Secondly, will the project administration notify all project participants that optimized clients are available for use, along with instructions on how to install them? Then there's the question of whether optimized apps give some platforms (ie operating systems and CPUs) advantages over others based on particular optimizations that are being done. Like are the optimizations being done to fully take advantage of each platform's capabilities.
Basically this is what I see it comes down to. Those using non-optimized apps will get less RAC (recent average credit) because the workunits will take longer to complete, while the ones using optimized apps will end up with much more RAC while crunching less. Am I missing something or oversimplifying here?
[EDITED]
The optimized app are not endorsed by the University. However, this project is open source so we will not restrict people from doing this. Also, the administration says we can't "notify" people, but if Cruncher wants to make a thread to list optimized apps, I will be happy to make it a sticky.
If you allow optimized apps, who is checking to see that the results are the same as results from the "official" app?
|
10)
Message boards :
Number crunching :
Poll: How much credit do you think is fair?
(Message 2713)
Posted 24 Mar 2008 by j2satx
Post:
Poll: How much credit do you think is fair?
Post how much credit you think is 'fair' per WU for this project.
No discussion, just post a number! :)
Al.
Credits should be what the computer asks for with a cap.....4 or whatever.
|
11)
Message boards :
Number crunching :
updated granted credit
(Message 2711)
Posted 24 Mar 2008 by j2satx
Post:
I agree that granted credit was too high before (although I enjoyed it) and is too low now. Some of my 64 bit Linux boxes are now getting less than claimed, while they get more than claimed on other projects with 64bit applications.
It will certainly put MilkyWay lower on my list of priorities as there are quite a few projects which will give more credit for the same amount of crunching, bigger caches and longer deadlines.
Sure, you can't buy anything with credits, but with so many projects to choose from, credits do mean something to people who are in it for the competition.
You don't need to be the highest "paying" project, but don't strive to be among the lowest.
Credits should be what the computer asks for with a cap.....4 or whatever.
|
12)
Message boards :
Number crunching :
Smooth sailing-Quiet board
(Message 2415)
Posted 19 Mar 2008 by j2satx
Post:
Good call, I hadn't noticed that.
Yeah, we're trying to fix the assimilator problem. It seems to be working, however it's either reprocessing already processed wus or not deleting them. Either way it's a problem so we're purging the db. After that done, you guys should be receiving work like you were. I'm also in the process of upgrading the server software, however I'm having trouble compiling because it wants to link with some non existant graphics libraries. We'll figure it out though.
Real crunchers don't need no stinkin graphics....... :)
|
13)
Message boards :
Number crunching :
Smooth sailing-Quiet board
(Message 2375)
Posted 19 Mar 2008 by j2satx
Post:
I thought it was a rumor that people ran this project :)
haha thanks guys. that's nice. i just took care of the assimilator backup and the server is generating new work again. it just freezes sometimes due to the older version of boinc we're using. so it's needs some assistance occasionally. this usually requires a kill of the boinc software and/or a reboot of the system and is a little bit time consuming. anywho, all should be good as of now. :)
If you know the problem is the older version of BOINC..........well then...
|