Welcome to MilkyWay@home

Posts by Keck_Komputers

1) Message boards : Number crunching : credit comparison to other projects (Message 9064)
Posted 25 Jan 2009 by ProfileKeck_Komputers
Post:
according to this site: http://boincstats.com/stats/project_cpcs.php

Milkyway is still giving about twice as much credit as the average project.
2) Message boards : Number crunching : Vote for the MilkyWay favicon! (Message 8178)
Posted 4 Jan 2009 by ProfileKeck_Komputers
Post:
#7 gets my vote
3) Message boards : MilkyWay@home Science : Tasks with over 400 hours "To completion" finish afer 1 hour (Message 7554)
Posted 9 Dec 2008 by ProfileKeck_Komputers
Post:
What is a few WU's?.
I already run at least 10 WU's in the last week or 2 and this still happens.


With that much of a difference it may take many tasks before it straightens out. The DCF moves in three different modes; if it is too low it immediately jumps up to the new estimate. If it is somewhat high it moves 10% of the difference at a time, usually 10-20 tasks will be enough to fix this amount of difference. If the estimate is very high it moves 1% of the difference at a time, in this case it can take 100s of tasks to get it straightened out. It was done this way prevent some tasks ending quickly compared to other tasks of the same type from changing the DCF too quickly.

The client will eventually straighten out, until then you see it running Milkyway tasks in high priority mode then running other projects for a time until the CPU debt is paid back. Detaching and reattaching should fix it if you do not want to wait.
4) Message boards : MilkyWay@home Science : If this project wont play well with others..... (Message 7475)
Posted 5 Dec 2008 by ProfileKeck_Komputers
Post:
BOINC will handle this automatically by not getting more milkyway work for a little while after it finishes this batch. That way your other projects will get the proper share of your CPU.
5) Message boards : Number crunching : No further Support for Milkyway at this time! (Message 6176)
Posted 15 Nov 2008 by ProfileKeck_Komputers
Post:
It's amazing that everytime a project comes out with an optimized client that DA shows his head and demands project parity. And to really add insult to injury seti is the worse for having gobbs of optimized clients. So when those optimized clients start ramping up the credits seti does their "Credit Adjustment Dance" and then whines to all the other projects. At the rate its going we're going to be doing 100 times more science work for [B/] NO CREDIT[/B].

If you ask me it is amazing that this project has not seen wider and more vocal outcry about the excessive amounts of credit given here. In the best case it is still giving about 4 times the credit it should. From my experience, this project currently has about 10% of the resource share I have for SETI and never had more than 50% of my resource share relative to SETI, yet my RAC is nearly the same on the two projects.
6) Message boards : Number crunching : rate of processing (Message 5018)
Posted 23 Aug 2008 by ProfileKeck_Komputers
Post:
However, there is one other possibility... Can return_results_immediately be triggered by a science application or, put another way, can science applications initiate a communication session on their own?

Yes and No. A project can set a maximum delay before the client will contact it again. Also if the deadline is tight enough you get the same effect.
7) Message boards : Number crunching : When will the deadline be extended? (Message 4639)
Posted 6 Aug 2008 by ProfileKeck_Komputers
Post:
Not to worry. BOINC will give this project extra time so that it can meet the deadline. Then it will avoid getting more work from this project until the extra time is paid back to your other projects.
8) Message boards : Number crunching : Credit Calculations. (Message 4615)
Posted 2 Aug 2008 by ProfileKeck_Komputers
Post:
As for optimized CLIENTS rather than apps, I *think* that's a non-issue. Since (again, I *think*) this method no longer uses a benchmark * time based claim, but rather counting operations (everyone calls it flop counting, but to the best of my knowledge it's actually a combination of both integer and float ops), I don't think there's a change in the client that could affect the claim (again... I'm relatively ignorant of the details on this, so I might be totally wrong). It does require the application to report these amounts to the client (and thus the server) and I'm not sure if they've worked the kinks out of that or not though.

Operation counts can be gamed in a similar way to the benchmark * time method. The primary defense against this in the new automatic credit corrections system is taking the mean of a good sized sample of results. A project would have to have at least 2,000 bad clients or 1/5 of active clients before it would become a serious problem (rough guess, the actual number depends on the exact formula used). Also it won't instantly go bad if one too many clients is off, it just gets a little more off for each one.

I agree this is a great thing for project administrators and BOINC in general. There may be problems in the future adapting this system to account for GPU and other non-CPU resources used. However I don't think they will be as insurmountable as some participants have made them seem.
9) Message boards : Number crunching : Peer Review...Seti vs. Boinc (Message 4599)
Posted 1 Aug 2008 by ProfileKeck_Komputers
Post:
I see Cappy basically has a personal grudge to bare and is not going to discuss the facts - definition of a troll maybe? Hard not be drawn into a major discussion when such falsaties are being spouted.

If basic stock apps do not give the same credit per hour for the same computer across multiple projects then something needs to be fixed. This, IMHO, is what is being attempted. That is all......

Live long and BOINC!


amen
10) Message boards : Number crunching : Credit Calculations. (Message 4468)
Posted 25 Jul 2008 by ProfileKeck_Komputers
Post:
Ok then...why haven't DA, yourself, or any other member of the "credit police" attack projects that are sub-par to the norm?

I have seen countless threads across BOINC projects from the CP screaming "FOUL" your credits are too high!! Not once, have I ever seen a thread from this same group crying "FOUL" your credits are too low!!

This is NOT Cross-project parity...and if your going to see any type of success ALL projects need to be adressed with the same passion.

You are correct and perhaps we should look at this more.

Personally I don't compare the credits much, I see a thread, then I go and check it out, then I start posting about it. In almost all cases that type of thread has pointed out a project that is granting too much credit.

It also seems to be harder to spot a project that grants too low. When you see one that is too high you glance across the chart and see a bunch of 1.xxx's. When they are too low you see 0.8xx, 0.7xx, and 0.9xx, that just doesn't make as much of an impression.
11) Message boards : Number crunching : BOINC 6.2.xx public release (Message 4410)
Posted 23 Jul 2008 by ProfileKeck_Komputers
Post:
Seems like a good idea to me. Most of the project scientists don't have the time to worry about credit rates. Having a client/server version that handles the adjustment for them automatically should help alot in the credit flame wars.
12) Message boards : Number crunching : Lost WU (Message 4231)
Posted 18 Jul 2008 by ProfileKeck_Komputers
Post:
I thought that "day" was a float, so you can put 0.1 for a little over 2 hours.
13) Message boards : Number crunching : program with CUDA (Message 4012)
Posted 7 Jul 2008 by ProfileKeck_Komputers
Post:
I don't think Crunch3r is writing CUDA off too quickly, just trying to inject some realism into the marketing hype.

GPUs sacrifice accuracy for speed. This can be fatal for most DC projects.

The quoted speed increases are normally burst speeds for single precision floats. Almost all projects need sustained double precision operations.

For some projects CUDA and other auxiliary resources will be a big help. Major changes in BOINC's scheduling system will make utilizing these resources easier for the projects. Some will be in the 6.2.x client and the rest in the 6.4.x client.
14) Message boards : Number crunching : No new WU (Message 3799)
Posted 16 Jun 2008 by ProfileKeck_Komputers
Post:
There is nothing more annoying then seeing a quad core machine sitting idle waiting for work...not to mention the thermal variances that it undergoes between running at full load and running at idle...I know there are complexities of the projects computer servers that I don't know about and that their system has to handle the load of many users, but is it really that difficult to supply a steady stream of work...of course barring any major system meltdowns as they happen and are usually unforeseen.

I have caught my computer in idle no less then a dozen times in the last 24 hours as it waits for the project server to send out a new batch of work (20 WU's) that lasts less then 25 mins on this machine.



There is nothing more annoying than to see people complain about this kind of thing when there is a score of worthy projects with regular work that can be used as a backup......the good of the project outweighs the good of the user.

I agree. My two favorite projects rarely have work, but my CPUs never get cold.

edit: The thing that really irks me is when a project sends out "make work". My computers could be running more important work with another project.
15) Message boards : Cafe MilkyWay : ATA thread (Message 3123)
Posted 12 Apr 2008 by ProfileKeck_Komputers
Post:
Hi folks, my name is John, and I have a problem...

I think I may be starting to recover. I heard about a project yesterday and have not signed up for it yet.
16) Message boards : Number crunching : 20 workunit limit (Message 2378)
Posted 19 Mar 2008 by ProfileKeck_Komputers
Post:
BOINC should work fine with this project. The short deadlines will make the client work on this project more in the short term, however the client will eventually decide to run other projects.

Reducing the number of tasks out per CPU will not cause insurmountable problems. I also run BURP there the deadlines are normally 12 hours and you are only allowed 2 tasks per CPU.

Not having a constant supply of work can also be a plus in this situation. It makes the client switch to different projects more often.

It is a disadvantage for participants that do not have a constant connection to the internet though. If they do not connect often enough to return the tasks they may not be able to return and get credit quick enough.
17) Message boards : Number crunching : windows popup errors (Message 2248)
Posted 14 Mar 2008 by ProfileKeck_Komputers
Post:
@banditwolf
What client version are you using? I think this is fixed in the current recommended 5.10.45.

Basically what is happening is that since many tasks have the same deadline the CPU scheduler can not decide which one is most important to run. Also just a few seconds of processing can change which task is most important.
18) Questions and Answers : Unix/Linux : GLIBC version (Message 1397)
Posted 8 Jan 2008 by ProfileKeck_Komputers
Post:
If you copy the old data directory to the new install BOINC will usually keep the same host info. I personally do not worry about new hosts anymore because sometimes you get them no matter what you do.
19) Message boards : Number crunching : Credit Award Error (Message 940)
Posted 6 Dec 2007 by ProfileKeck_Komputers
Post:
That happens from time to time in all projects. Nothing to worry about unless it is a major difference, in that case it is normally a problem with your computer.
20) Message boards : Cafe MilkyWay : First! (Message 45)
Posted 3 Oct 2007 by ProfileKeck_Komputers
Post:
Hi all! The ATA eagle has landed :p



Nice very appropriate.




©2020 Astroinformatics Group