Linux users - POLL - credits and distros

Hi all :D

Wich distro are you using, and how many credits per hour do you usually claim for?

When possible, calculate a mean for claimed credits per hour when your system is used ONLY by the boinc client and by your distro's services, background processes (i.e. as much as possible when the system is not used by you). It's important to specify if there are no WUs at all on your system being crunched with the whole system dedicated to boinc. Anyway do your best for the results of your means being as "clean" as possible.

Should be nice also the ouput from

echo $[`ps -e|wc -l` - 4]

which, for information, counts the number of row in the output of `ps -e`, which lists all processes loaded on your system. Take the number, subtract 4 because of the title, the command itself, one added by the misterious output from ps and interaction with wc, one is added when I use the output from wc as a number... The last one is veeery misterious for me, can you help? :)

First I write (or copy) a description of the host

Genuine Intel(R) CPU T2500 @ 2.00GHz
1 GB ram
Kubuntu Dapper 6.06
Linux 2.6.15.7 - compiled from sources
Credits per hour: 9.67
Running processes: 115

To calculate the claimed credits per hour, for example you can take the first 3 results from the host (if they represent well the generic result)

(sum_of_claimed_credits / sum_of_cpu_times) * 3600




Maybe you don't have time for playing with numbers and processes :)
If you have some, please do these simple steps so we can see if there's a distro not so good with boinc. Maybe FC?

Is there difference of credits even between distributions? :o
____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

> Well daniele, I have some data for you (I am not sure what you have against Fedora Core but I like it and it works for me).

I am running Fedora Core 3 on 2 computers that only do Boinc and nothing else.

AMD Opteron 275 (dual core dual processor)
Using your little programme shows 127 processes running
Last 5 results claimed mean (57.84,57.77,57.86,58.62,58.06) = 58.195
claimed average = 58.03
Last 5 results granted mean (23.09,29.58,23.98,58.62,23.05) = 40.835
granted average = 31.66
Last 5 results total cpu time (seconds) = 51,244.04
total cpu time (hours) = 14.24
total cpu time (hours) mean/average = 2.855/2.85
Last 5 results claimed cobblestone/hour (mean) 58.195/2.855 = 20.384
claimed cobblestone/hour (average) 58.03/2.85 = 20.36
Last 5 results granted cobblestone/hour (mean) 40.835/2.855 = 14.303
granted cobblestone/hour (average) 31.66/2.85 = 11.11

AMD Opteron 285 (dual core dual processor)
120 processes running
Last 5 results claimed mean (57.89,57.80,57.90,57.83,57.83) = 57.85
claimed average = 57.85
Last 5 results granted mean (23.29,23.92,31.60,23.05,23.13) = 27.325
granted average = 24.998
Last 5 results total cpu time (seconds) = 43,202.30
total cpu time (hours) = 12.00
total cpu time (hours) mean/average = 2.40/2.40
Last 5 results claimed cobblestone/hour (mean) 57.85/2.40 = 24.104
claimed cobblestone/hour (average) 57.85/2.40 = 24.10
Last 5 results granted cobblestone/hour (mean) 27.325/2.40 = 11.385
granted cobblestone/hour (average) 24.998/2.40 = 10.42

daniele the claims made by Linux machines on average are close to half those of the Windows machines. Yes I am using an optimized client as it has always worked and is a hangover from when I set the computers up. On most projects it makes no difference at all due to the new methods used for granting cobblestones as your claim is irrelavent, they will give you what they have deemed you deserve. I am loath to update the client as for one thing I am not sure how to do it on a Linux machine and for another I see no need when it is working fine.
Boinc benchmarking is very poor and usually favours Intels and Windows as that is what the programme was set up on originally (or so I believe).
Linux machines have always underclaimed due to this Boinc benchmarking issue (also applys to Apple Mac's), and it has caused a lot of heated 'discussions' on many project forums and also has lost a lot of users, often to other projects if they are chasing credits, but sometimes they just drop out and don't bother participating anymore because of the bickering.
I don't want to see that happen here at Docking, even though I am not getting anywhere near the credit of some other testers I am still crunching and waiting as I know it will be fixed by the time it goes Beta. I can get a lot more credit on some other projects if I wanted to.
I think you will find that all Linux distros will be claiming about the same and any of them should be ok with any project (I am not sure about the distro you helped Doug with though as I had not heard of it before). I am doing 8 projects on these 2 Linux computers of mine and they all work fine (except Predictor as it is off line).
There are only 2 projects that I participate in that still use Boinc benchmarking, QMC and Docking. QMC will be changing this month and as you can see from my results above my optimized client makes no difference here at Docking. 10 and 11 c/h is what I get at Seti and LHC.
I am against blantantly modifying the client (I personally would not know how to do it), to post scores that in no way could the processor be capable of living up the Dhrystone/Whetstone claim (such as over 31,000 dhrystone million operations per second for an Pentium M processor). The client I use still puts me behind Windows machines but at least I am getting a fairer result, credit/cobblestone wise.
Forgive me for rambling I have avoided nearly all other credit debates but I need to have a say so I am saying it here.
Sorry if this not the sort of information you required.
I suppose this will start people off now I have said my bit, oh well that is what a forum is for is it not?
____________

Dr David Anderson is working on linux benchmark system lest linux should get awarded far lower credits, as WCG mentioned here . Optimised client for linux then will no longer make sense.
____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

Forgive me for rambling I have avoided nearly all other credit debates but I need to have a say so I am saying it here.
Sorry if this not the sort of information you required.
I suppose this will start people off now I have said my bit, oh well that is what a forum is for is it not?

Forgive me for rambling I have avoided nearly all other credit debates but I need to have a say so I am saying it here.
Sorry if this not the sort of information you required.
I suppose this will start people off now I have said my bit, oh well that is what a forum is for is it not?

> @ suguruhirahara, this I know and I said as much in my previous post.

> @ daniele, you asked what distro people are using and what c/h they're claiming, this I gave you in detail (my computers are not hidden, saved anyone having to look), in the way I thought you wanted.
As to how you can compare to me, well, it is all a bit academic really.
I noticed from my 2 computers that they do not claim the same and are not granted the same amounts, due to different processor speeds and what group of people I do workunits with.
A lot of the computers I have looked at have the same specs but not the same benchmark scores, some are above and some are below and a few are way below the amounts similar processors benchmark at.
So even with no optimization you results are not going to be even.
My claims, as against my granted, are wide apart due to the majority of people having so called 'standard' clients, this gives me a 'standard' granted credit on nearly all my workunits. So if you say I get twice as much as you then look at the other computers as I get an average (i.e. the middle value of 3 results) and I think only once has it been my claim (which means some one out there claims more than I do).
I will be getting rid of the optimized client later this month when I get more time and work out how to upgrade Linux software.
Optimized clients are really no longer relevant to Boinc anyway, and I would like to get the better functionality of the new Manager and Client as well.

So daniele, my data is there, it shouldn't be much different to others, but be aware that a slower computer will be claiming less and granted less per hour due to not doing as many workunits as a faster one in the same day (my 275 Opteron, 2.2 GHz, takes about 1,600 seconds longer to process a Docking WU than my 285 Opteron, 2.6 GHz), hence the faster computer will make more credit claims in a day than a slower one.

I still like Fedora Core though, you are using Kubuntu? I have heard some good things about it, particually the latest version.
I will try and forget the credit thing now and get back to a more lighter note in my postings, my rant is about done.

On a lighter note don't you think it is funny that nearly 100,000 people have signed up and added 186,000 computers to fight the global warming issue?
That number of computers uses a lot of electricy, creates a lot of heat, needing airconditioning to keep cool, more electricy, so the electic companies burn more coal to supply the energy and create more greenhouse gases which we then need more computers to compute what is happening which uses more electricity, etc, etc, etc. And there is another 2 of these projects as well.
I am rambling again, I know, I will stop now.
____________

As to how you can compare to me, well, it is all a bit academic really.

...

be aware that a slower computer will be claiming less and granted less per hour due to not doing as many workunits as a faster one in the same day (my 275 Opteron, 2.2 GHz, takes about 1,600 seconds longer to process a Docking WU than my 285 Opteron, 2.6 GHz), hence the faster computer will make more credit claims in a day than a slower one.

On a lighter note don't you think it is funny that nearly 100,000 people have signed up and added 186,000 computers to fight the global warming issue?
That number of computers uses a lot of electricy, creates a lot of heat, needing airconditioning to keep cool, more electricy, so the electic companies burn more coal to supply the energy and create more greenhouse gases which we then need more computers to compute what is happening which uses more electricity, etc, etc, etc.

On a lighter note don't you think it is funny that nearly 100,000 people have signed up and added 186,000 computers to fight the global warming issue?
That number of computers uses a lot of electricy, creates a lot of heat, needing airconditioning to keep cool, more electricy, so the electic companies burn more coal to supply the energy and create more greenhouse gases which we then need more computers to compute what is happening which uses more electricity, etc, etc, etc.

You are right, this is a big problem. Anyway we don't have another mean if we want to calculate something. We live in continuous contradiction and we perservere with it. Probably we like our contradictions, or we are our contradictions... This is sure academic :)

But the fact needs deep concerning. How percentage of the all PCs over the world is participatihttp://docking.utep.edu/forum_reply.php?thread=93&post=1337#inputng in the distributed computing? How many hours in a day do they crunch? How does electricity changes by crunching workunits? and so on...

But the fact needs deep concerning. How percentage of the all PCs over the world is participatihttp://docking.utep.edu/forum_reply.php?thread=93&post=1337#inputng in the distributed computing? How many hours in a day do they crunch? How does electricity changes by crunching workunits? and so on...

Since this thread was born mainly to know each other, can I ask you about your sign? Why "audentes fortuna iuuat"?
Is it a mistake or you meant something particular?
BTW, what does your wonderful name mean?

the phrase is latin and frankly saying I'm not really sure what it can mean particular thing... seek google for it. Because my avatar is a picture of goddess, then I just wanted to have additionaly what relates with the avatar:)

by the way, anyone tried crunching d@h on vine linux or turbo linux? This ditribution is rather popular in Japan.

Maybe I should change the name to this thread and you could move it in the "cafe" section?

Maybe I should change the name to this thread and you could move it in the "cafe" section?

But need I make a new thread for several posts...? Since they're not causing a big confusion I'll keep this thread as it is:)

Okay, let me go back to the the topic!
I'm testing crunching on linux(debian) via VMWare, and

Fri 10 Nov 2006 12:48:39 AM PST||Benchmark results:
Fri 10 Nov 2006 12:48:39 AM PST|| Number of CPUs: 1
Fri 10 Nov 2006 12:48:39 AM PST|| 605 floating point MIPS (Whetstone) per CPU
Fri 10 Nov 2006 12:48:39 AM PST|| 1720 integer MIPS (Dhrystone) per CPU
Fri 10 Nov 2006 12:48:39 AM PST||Finished CPU benchmarks
Fri 10 Nov 2006 12:48:40 AM PST||Resuming computation
Fri 10 Nov 2006 12:48:40 AM PST||Rescheduling CPU: Resuming computation

Hello, WOW

Conan, I think all you said in your first Post was phenominally intelligent,
well thought out, and it caused no such contrversy, you were concerned about.
Indeed, this thread, started by another intelligent Human being, Dannielle.
Conan, the Distro I run is called PClinuxOS "Big Daddy", they made MiniMe also.This 2.4 Gig machine beat out
much better ones, and was credited the same as Windows. It is Seti, our Team
is getting 3rd place back.
http://setiathome.berkeley.edu/workunit.php?wuid=98824252
I am also using a O.C., that could make my P.C. 20,000 seconds faster also.
Funny thing though, I run Boinc 2 different ways.Original way my Bench,s are
a "bit" better:Measured floating point speed 598.05 million ops/sec
Measured integer speed 1626.38 million ops/sec
When I run Docking the second command line way, these are my Bench,s
Measured floating point speed 580.23 million ops/sec
Measured integer speed 1644.35 million ops/sec
"Happiest crunching to all"
Doug Worrall

Hello, all:)

I'm continuing testing of Vine Linux through VMWare, and I have a question around linux itself.
Can linux deal with dual core CPU like PentiumD? Boinc doesn't recognise it as two CPUs. How should I configure the OS?

Thanks,
suguruhirahara
____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

Hello, all:)

I'm continuing testing of Vine Linux through VMWare, and I have a question around linux itself.
Can linux deal with dual core CPU like PentiumD? Boinc doesn't recognise it as two CPUs. How should I configure the OS?

Hello, all:)

I'm continuing testing of Vine Linux through VMWare, and I have a question around linux itself.
Can linux deal with dual core CPU like PentiumD? Boinc doesn't recognise it as two CPUs. How should I configure the OS?

Thanks,
suguruhirahara

Since on this thread things around linux are discussed, I moved this thread to cafe.

I was trying to install kernel-smp, but I failed to configure lilo in right way, so I'm installing vine linux again:( Fortunately results completed had been uploaded and reported!
____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

hi all:)

I was almost fed up with linux, but I finally found I needed kernel-2.6 to deal with dual core. Because Vinelinux is used as stable systems eg for school/univ, it's so later than other distributions to implement new features. I'm now downloading the iso image and I'll upgrade the system so that I can crunch 2 results at the same time on the system.

thanks for reading,
suguruhirahara
____________

I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions.

I havent got enough results in yet to call an average creit claim.

Computer 1 is a Celery 2Ghz system
Kubuntu 6.06
Boinc 5.4.9 Standard client (25% share but getting close to 50% atm due to RCN out of work)

Claims approx 31credits a WU time is 7hr 20mins (only done 2 WU's)
Recieved around 16 credits for the one thats reached quorum
One client claimed as low as 10 credits for such a long WU

Computer 2 is a Celery 700Mhz
Kubuntu 6.06
Boinc 5.4.9 has a 33% share but RCN out of work so 100%

Claims 51.34 credits a WU time is 17hrs 23mins (only one complete so far)

This is kinda sad as my 2Ghz gets 12.60 credits per Einstein short WU and generally knocks em over in and hour and a half. Its amazing what you can get on a project with fixed credits and the palying field is even with windoze.
____________

My puter is an Athlon XP 2200+, running Suse 9.2 and BOINC 5.4.9

My claims are about 5.7 c/h, My grants differ between 4.6 and 7.8 c/h.

My claims and grants at other Projects are (only recent ones, all c/h):

Serverside granting:

Rosetta |  5.2 |  7.7 - 10.1

Ralph   |  5.2 |  7.0 - 11.1

Seti    |  6.6 |  6.6

Einstein| 15.5 | 15.5

CPDN    | 10.1 | 10.1

SAP     |  8.7 |  8.7



Bechmark granting:

Malaria |  5.2 |  5.2 -  9.5

WCG     |  5.3 |  3.6 -  5.9

Simap   |  5.3 | 10.0 - 15.0



Older stuff:

LHC     |  5.2 |  5.2 - 10.1

Pirates |  5.8 |  5.8

Here's my Linux host:

AMD Athlon(tm) XP 2600+
256Mb
Measured floating point speed - 992.2 million ops/sec
Measured integer speed - 1736.4 million ops/sec

Linux 2.6.17-10-386
Ubuntu 6.10
Running processes: 91
Boinc 5.4.9

Approx. 6.07 credits p/h
____________