Linux Users: Sieze the Power!

Open your BOINC/client_state.xml file.
Look for the line that starts with <p_fpops>, about 10 or so lines from the top.

It will look something like <p_fpops>1241479847.30971</p_fpops> (EEEEEwwwwwwww!)

Make it look more like <p_fpops>12414798479.000000</p_fpops> ( Swweeeeeeeet! )

Then watch your credit claims look more like the credit claims for Windoze crunchers. WTF, are we second class? Are we gonna take it? Noooo sirreee!

Look at my claim after I fixed things up as above. But I got beat by the quorum.

If all of us Linuxians do it we'll beat the quorum and get points like we deserve instead of the crumbs we're getting now. And if they don't like it then.... do we really care?

until your benchmarks run again.. then it will be back to normal!

And if they don't like it then.... do we really care?

until your benchmarks run again.. then it will be back to normal!

until your benchmarks run again.. then it will be back to normal!

And if they don't like it then.... do we really care?

Yes, I care.
I'm a Linux cruncher, and I won't do this kind of cheating.
Adjust the benchmark to the win level (is it * 1.5 or * 2 ??) may be OK in most cases, but I don't care that much about it, that's what quorums are for.

Imho credit inflation is cheating.

What if Windows users change that as well?

What if Windows users change that as well?

I want to make my point clear:
Linux is a bit small changed by BOINC, something about 25-40% or so. AFAIK same goes for PPC.

The so-called "optimized" clients go far too far. To get to the ordinary windoze level is OK, everything exceeding this is plain cheating (like the use of 5.5 and such).
The tenfold increase of credits is plain cheating as well. There is no justification for this in the real world.

The tenfold increase of credits is plain cheating as well. There is no justification for this in the real world.

The tenfold increase of credits is plain cheating as well. There is no justification for this in the real world.

It's not as plain as you think.

Read my post. Who said anything about a tenfold increase of credits? I increased my FPOPS in BOINC/client_state.xml by tenfold. That doesn't translate into 10 times the credits.

Need proof? Then click on my name in the left column to see my Account Data page then click on Computers to see what I've been claiming. In a few days, when my pending results validate and I get the credits, you'll likely see I get less than Windows, even with a tenfold increase in FPOPS.

For example, here is a recent result which I crunched with the inflated FPOPS, my computer is 217. See? My claims are just a wee bit higher than Windows machines. Yes, I will be tweaking my FPOPS down a wee bit, probably around 9 times higher than normal, which will probably put my claims almost exactly on par with Windows. Then, if all/most the Linux users here will do the same then us Linux crunchers will get our due.
result that

Hello,
Have been following this thread, I donnot understand most about flops and such, but I do know the Unfairness in credit systems, assides from 2 experiments that Benchs really dont matter.If your P.C. crunches faster than a Widoz Machine, you get the same credit as evryone.I beat out Windoz Machines "Most" of the time, asides from Dual cores.
BUT, I have seen rediculous benchs on Windoz machines and I know they manipulated the processs.That is why there is a process in place to keep things on a level playing field.But most things in life are not fair.Using O.Clients ,if they are legal to use I will, but to manipulate the benchs, may be a hard argument
for both sides.Hopefully once Docking is out of Alpha, they may adopt the timed w/u where just the time spent on a paticular w/u , everyone gets the SAME credits.
Now that I have finally got the ulmit - s unlimited figured out I am being told there is no work Sat Sep 23 06:20:26 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sat Sep 23 06:20:43 2006|Docking@Home|Sending scheduler request to http://docking.utep.edu/docking_cgi/cgi
Sat Sep 23 06:20:43 2006|Docking@Home|Reason: To fetch work
Sat Sep 23 06:20:43 2006|Docking@Home|Requesting 8640 seconds of new work
Sat Sep 23 06:20:48 2006|Docking@Home|Scheduler request succeeded
Sat Sep 23 06:20:48 2006|Docking@Home|Message from server: No work sent
Sat Sep 23 06:20:48 2006|Docking@Home|Message from server: (there was work but it was committed to other platforms)"What Windoz?"
Sat Sep 23 06:20:48 2006|Docking@Home|No work from project

Have been trying all day, are others receiving work still?There is a good argument here for the Linux and Windoz Bench,s.Am sure the staff and Scientists
agree?

Sincerely

Doug Worrall

...Now that I have finally got the ulmit - s unlimited figured out I am being told there is no work Sat Sep 23 06:20:26 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sat Sep 23 06:20:43 2006|Docking@Home|Sending scheduler request to http://docking.utep.edu/docking_cgi/cgi
Sat Sep 23 06:20:43 2006|Docking@Home|Reason: To fetch work
Sat Sep 23 06:20:43 2006|Docking@Home|Requesting 8640 seconds of new work
Sat Sep 23 06:20:48 2006|Docking@Home|Scheduler request succeeded
Sat Sep 23 06:20:48 2006|Docking@Home|Message from server: No work sent
Sat Sep 23 06:20:48 2006|Docking@Home|Message from server: (there was work but it was committed to other platforms)"What Windoz?"
Sat Sep 23 06:20:48 2006|Docking@Home|No work from project

Have been trying all day, are others receiving work still?There is a good argument here for the Linux and Windoz Bench,s.Am sure the staff and Scientists
agree?

Sincerely

Doug Worrall

Same here:

Still quite a lot imho. I've always had the impression of about not even double the credits for my machine in it's old win times compared to now. But I'm not Tony (mmciastro), and I don't have the urge to do sample that much benchmarks for a valid factor. But you ahve to compare of course apples and apples, not oranges, so stock client with stock client, don't use the "opt." client as a benchmark.

I can't see the names of your puters, that's restricted for ones own machines.

I assume you're talking about

#217 (linux)
Measured floating point speed 12414.8 million ops/sec
Measured integer speed 2165.86 million ops/sec

and #254 (windoze)

Measured floating point speed 2261.14 million ops/sec
Measured integer speed 4164.25 million ops/sec

Both are CPU type AuthenticAMD AMD Athlon(tm) 64 Processor 3300+

claimed credit = ([whetstone]+[dhrystone]) * wu_cpu_time_in_sec / 1728000

If you add the numbers for the benchs you'll get 14680 for the lin machine, and 6425 for the win. That'll lead to more than double the credit claim according to the calculation for the lin one.

But, what I see as well is the shorter crunching times of the Linux WUs compared to the Windows puter.

This is a question to the admins:
Is the application for linux and windows the same or do the windows users more of whatever?
Is the app perhaps more suited for Lin than Win?
How does this reflect on the scientific output (however it will be measured)?
Is there a possibility to measure the scientific output at the servers?

1) The source code for windows, linux and mac is exactly the same for all platforms. But of course source code has to be compiled in a binary so that it can be run on those platforms. And that's where the difference might be as optimizations, etc will be different for windows, linux and mac. I'm not a compiler expert myself, but that's where the difference might come from.
2) This app (charmm) was originally written for a Cray supercomputer :-) It has been in development since the early 70's and is written in Fortran77. The windows port was only recently made. Therefore it is probably better suited to Lin/Mac than Win. And what we see is that it runs a whole lot faster on unix than it does on windows.
3) There will actually be differences in the output when run on different platforms. That's why we use homogeneous redundancy in boinc which makes sure that workunit replicas all are send to the same platform (or win, or lin, or mac). That way we do not compromise our scientific results and make sure people actually get credit for their results.
4) I'm not sure what you mean with this. Can you rephrase this question? I suspect your question might have been answered by 3 above...

The whole benchmark problem we are experiencing now is actually because we are using Boinc's standard integer and floating point benchmarks. These benchmarks of course are not Charmm specific, but very general purpose benchmarks. So whatever the boinc client measures is probably way off of what the app actually is using. That's why we will develop our own benchmarks in the future (Boinc has an API to tie your own benchmark in the boinc client) so that we are not dependent on the standard benchmarks anymore. This also means of course that optimized clients will get exactly the same flops and ints numbers as standard clients, which is what we all want in the end: fairness and credit based on realistic numbers.

Hope that helps!
Andre

But, what I see as well is the shorter crunching times of the Linux WUs compared to the Windows puter.

This is a question to the admins:
1) Is the application for linux and windows the same or do the windows users more of whatever?
2) Is the app perhaps more suited for Lin than Win?
3) How does this reflect on the scientific output (however it will be measured)?
4) Is there a possibility to measure the scientific output at the servers?

Andre,

Thank you for the information regarding the BOINC API provision for using your own benchmarks. It sounds like in the future the project's Charmm application will run its own more accurate benchmarks and BOINC will use those rather than its own inaccurate benchmarks. That's good news!! Thanks for being aware of the benchmark problem and thanks for taking time late Saturday evening to post.

Saenger,

You are correct in saying my adjustment of the FPOPS on the Linux machine will lead to more than double the credit claim. In fact it leads to about 4.5 times the credit claim which is almost perfect in this case. Review the history for both machines. The Windoze machine claims about 39, the Linux machine was claiming about 9 but now it claims about 40, a factor of about 4.5.

I will try a multiplier of 9.8 and see if I can get the Linux machine to claim 39 instead of 40. It appears the 10 multiplier is slightly too high, 9.8 or 9.9 might be better.

@Andre:
Thanky you very much for this informative answer.
My question 4) was made in regard of the different possibilities of fair credit granting in distributed computing:
a) measure the possibilities of a puter and the donated time (current method)
b) measure the used operations (Flop-counter) like in Seti
c) measure the delivered scientific results (like Rosetta, CPDN, Folding)

a) is imho quite open to client side meddling, either to get a fair playing field, like Dagorath intended, or for plain cheating. But if it's doen properly, it has the plus that windows users would not be "punished" for the not suited implementation of the code to their machines (lin benchs have to improve nevertheles ;)

b) has the advantage to measure work done on a science independent level, and can possibly implemented on various projects. But it may take some ressources away from the real crunching to bean counting.

c) will grant for what's important for the project. The downside is, that some puter architectures will be "punished" for the not optimized code. This happens atm @Rosetta with PPC, here it would obviouly affect windows crunchers. The upside is, such effects are spotted, as otherwise this effect will go unobserved by the crunchers.

@Dagorath:
In regard of c) your factor seems to be right, in regard of a) mine. Both are fair measurements imho, in an ideal world both should be the same, but they are obviously not.
I personly tend to solution c) as well, but it's up to the project team to decide. I can live with both.

Dagorith,
When I read the Boinc Shell. Cannot find that line 10, and cannot see
</p_fpops> to find that line, any better ideas?
Doug

Dagorith,
When I read the Boinc Shell. Cannot find that line 10, and cannot see
</p_fpops> to find that line, any better ideas?
Doug

It's more complicated than I first realised and a bit of a nuisance so I have decided to simply not crunch projects like D@H that do not reward Linux systems fairly. Yah, I know it's only alpha and all that but my list of minimal expectations for alpha projects now includes fair credits for Linux. I don't give a darn about the credits but so many other fellow crunchers do, it needs to be fixed. I think the only way to force a proper fix is to adopt the position I have adopted which may in turn force the projects and BOINC devs to pull their heads out their butts and do the right thing.

lol. For them there must be other things to do before they go into the credits for Linux specifically. moreover, they surely promised to work on the point in future. What you're doing isn't admirable.

lol. For them there must be other things to do before they go into the credits for Linux specifically. moreover, they surely promised to work on the point in future. What you're doing isn't admirable.

I don't crunch to earn your admiration. I don't owe D@H anything, they are in MY debt not the other way around. Their failure to issue credits fairly is not admirable. For me there are other things to do before I support projects that refuse to tackle issues that have existed for as long as the Linux credits problem has existed. A promise to work on it in the future does not tackle the issue. Other projects have made real efforts and have provided real solutions, I will support them instead.

.

Call it what you want"its unfair" Life is unfair, but that does not give me the right to change a Benchmark.

Call it what you want"its unfair" Life is unfair, but that does not give me the right to change a Benchmark.

I have the right to change whatever I please on my computer as long as it does not violate the law. I altered the benchmark to claim what that computer would have claimed had it been running Windoze. You have the right to call equal pay for equal work cheating if you want to but I just can't imagine why anybody would want to say something that silly in public.

.

So, Doug, you stick with your claim that claiming equal pay for equal work is cheating. I define cheating as claiming more than the amount one earns.

I wish you good luck on your quest for silliness though I don't think you need the luck, you seem to have all the requisite skills and tenacity required to accomplish your noble goal.
.

So, Doug, you stick with your claim that claiming equal pay for equal work is cheating. I define cheating as claiming more than the amount one earns.

I wish you good luck on your quest for silliness though I don't think you need the luck, you seem to have all the requisite skills and tenacity required to accomplish your noble goal.
.

lol. For them there must be other things to do before they go into the credits for Linux specifically. moreover, they surely promised to work on the point in future. What you're doing isn't admirable.

I don't crunch to earn your admiration. I don't owe D@H anything, they are in MY debt not the other way around. Their failure to issue credits fairly is not admirable. For me there are other things to do before I support projects that refuse to tackle issues that have existed for as long as the Linux credits problem has existed. A promise to work on it in the future does not tackle the issue. Other projects have made real efforts and have provided real solutions, I will support them instead.

Doug, Some men have the courage to admit when they've made a mistake and go on with life. You don't seem to have the cahones for that so don't tell me about pompacity and arrogance. Your claim that claiming what I earned is cheating is patently silly and your flying into a rage over it is further proof of your sillyness. Do I care that I don't have a lot of silly friends? Not at all because I have sensible courageous friends.

Doug, Some men have the courage to admit when they've made a mistake and go on with life. You don't seem to have the cahones for that so don't tell me about pompacity and arrogance. Your claim that claiming what I earned is cheating is patently silly and your flying into a rage over it is further proof of your sillyness. Do I care that I don't have a lot of silly friends? Not at all because I have sensible courageous friends.

Yes, you're all right, but you're too impatient.

Not all devs are like superman (lol).

This project can be advanced with other patient linux users, without you, so please participate in other project with precious resource of your linux machine for the humanity:)

But please keep in mind before you go that if linux machines continue crunching as they have been the credit system issue can be set to higher priority since what they should solve other issues before they work on the problem can be completed. I'm not sure, though.

You cannot see past your EGO, to admit, Linux users do not share your beliefs. Do any Linux users beleive that manipulating your Bencmarks is "wrong"?

If you cannot see past your own arrogance to accept that we all have our own opinions,

wrong in your eyes maybe, but, where are other Linux users that agree with you.??

No- more waisting time with "your" arrogance.If you cannot see that, really feel sorry for you. Lonley Boy raging on and on about how Manpulation is correct.

I don't know if manipulating a file this way is cheating. I don't have time for "philosopy of distributed computing" right now, but it's sure that all these questions come from a real problem. It's necessary to speak about it as much as possible, so that the admins can be informed about this issue in the credit system :)

I don't know if manipulating a file this way is cheating. I don't have time for "philosopy of distributed computing" right now, but it's sure that all these questions come from a real problem. It's necessary to speak about it as much as possible, so that the admins can be informed about this issue in the credit system :)

And Dagorith,
One person "may" agree with you in philosophy, yet, where are all the other Linux users, in this Docking experiment.? one individual. and hehe, an Email, donnot
show your approach to a problem will get you very far in life, or with Boinc. Of course you did not manipulate me, you are just fooling yourself.

And Dagorith,
One person "may" agree with you in philosophy, yet, where are all the other Linux users, in this Docking experiment.? one individual. and hehe, an Email, donnot
show your approach to a problem will get you very far in life, or with Boinc. Of course you did not manipulate me, you are just fooling yourself.

Right, Doug, where are all the other Linux crunchers who you seem to think disapprove of claiming what one earns? Foam at the mouth as much as you want but you won't find many people who think claiming what I earned is cheating. In fact not even the project admins here seem to think it's cheating.
.

Dagorath,

Stop stupid cracking and flaming.
No obscene language, and keep forum clean.
____________

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

Dear Doug and Dagorath,

Having a discussion about the boinc credit system is fine with me, but the fact that this discussion is going way out of line regarding all the obscene language and childish behavior that has been thrown around is not. This is a biomedical project's forum, not a hate/flame/obscene language forum, so please keep it down for the other people's sake.

To answer the question: yes, the boinc credit system is not fair to all platforms. yes, the boinc developers are working on finding a solution for this, and yes, we are working to find a better solution for D@H itself, but this will cost some time and some patience. If that is too much to ask for, then by all means please stop alpha-testing D@H since it will only bring frustration to you and to us; we have frustration enough with our app already :-)

For now, I've deleted the posts with the obscene language.

Thanks for your understanding. Let me go back to my credit system research now...

Andre

PS Suguru, thanks for focusing my attention on this!!
____________
D@H the greatest project in the world... a while from now!

Please make sure that the boinc developers hear about this as well from as many people as possible. It would be best to fix the problem at the root and not have multiple different project solutions in place that will only confuse people (and more work for the project devs that could be spent on the app).

Andre

I don't know if manipulating a file this way is cheating. I don't have time for "philosopy of distributed computing" right now, but it's sure that all these questions come from a real problem. It's necessary to speak about it as much as possible, so that the admins can be informed about this issue in the credit system :)

@ Dagorath a.k.a 'Worrall_Is_An_Idiot'
Stop your silly game.
Now you're insulting an user.
Doug Worrall_Is_Not_An_Idiot.

What do you want ? a ban ? you're on the way...

We gave him an hour to change it back to his name, before we'll do that ourselves and disable his user account. We hate to do this, because I love democracy and freedom of speech, but some people just go to far. We'd like to do science and not play psychologist :-)

AK and the D@H team

@ Dagorath a.k.a 'Worrall_Is_An_Idiot'
Stop your silly game.
Now you're insulting an user.
Doug Worrall_Is_Not_An_Idiot.

What do you want ? a ban ? you're on the way...

To answer the question: yes, the boinc credit system is not fair to all platforms. yes, the boinc developers are working on finding a solution for this, and yes, we are working to find a better solution for D@H itself, but this will cost some time and some patience. If that is too much to ask for, then by all means please stop alpha-testing D@H since it will only bring frustration to you and to us; we have frustration enough with our app already :-)

1) Linux earns fewer points per workunit then the other platforms.
...
In order to address issue #1 we have sent our findings to Dr David Anderson at BOINC. He said that they will be compiling the Linux agent in the 5.8 release with optimizations that are comparable to the optimizations used on the other platforms. This should allow the credit claimed by Linux to match the other platforms. This new version should be available with 6 weeks or so.

@ Dagorath a.k.a 'Worrall_Is_An_Idiot'
Stop your silly game.
Now you're insulting an user.
Doug Worrall_Is_Not_An_Idiot.

What do you want ? a ban ? you're on the way...

As someone who has but one part time computer on windoze (the one my 7yo sister uses) and 2 on linux, with a 3rd on the way to also use linux i too get upset about how low i get credit compared to a windoze box. But i have accepted this as being part of choosing to crunch and hope 5.8 is better.
I have personally found those projects that flops count are best for equal credit for equal data processed. Actually those projects who do flops count make up a lot more of my credit than those that dont even though they often get much less CPU time (around 25%).
If anything flops counting or fixed credit by workunit will bring in crunchers who use linux real fast but you will also find many linux users seem to crunch just about nething thats new so there always around.
Ive rambled enough now :)

Fixed or flops counted credits are good and i personally wont use any method of boosting my credits other than a bigger better computer :)
____________

Fixed or flops counted credits are good and i personally wont use any method of boosting my credits other than a bigger better computer :)

Nah i wont be using windoze anymore if i can help it :) Ill leave that for little kids and computer illiterates.

Please this was not a flame. Dont hurt me
____________

Ok i lied and bit the bullet. Loaded Boinc 5.5.0. If it claims more than the standard windoze client it goes out the window but if it gets me to just under ill keep it
____________

Using 5.5.0 client issue will be, I think, widely discussed in future, and I don't regard it good idea to use the client. Since this project is under alpha-testing, I've dared not to deal with the issue:( But perhaps the necessity of discussion around the fairness is coming up...whether willingly or not.

Ok i lied and bit the bullet. Loaded Boinc 5.5.0. If it claims more than the standard windoze client it goes out the window but if it gets me to just under ill keep it

Using 5.5.0 client issue will be, I think, widely discussed in future, and I don't regard it good idea to use the client.

I agree with you.

Using 5.5.0 client issue will be, I think, widely discussed in future, and I don't regard it good idea to use the client.

I don't regard it as a good idea on Windows (and that's putting it politely). I speak only for myself, but given the known issue with BOINC on Linux, using BOINC 5.5.0 on Linux until the official BOINC Client is fixed (or credit does not rely on the BOINC Client's benchmarks) is fair enough.

I think the alpha stage is a good time to look at these issues and I'm happy to see Andre is taking it seriously. But first priority has to be the apps - get them right so the science is right. After that, do what's necessary to keep the volunteer force interested in helping out (and appropriate levels of credit is a big part of it for many of us).

Even i felt it way overclaimed. to close on windows i needed to claim around 80 credits per WU. I claimed around 115 so im taking it off. its just too much of a claim.
____________

The use of 5.5.0 client is of no relevance when used on projects that don't use Boinc benchmarks (this is for Windows).
The use of the version can also cause problems on projects that do use the boinc benchmarks as I found out when I added an older Intel P4. It has been running Rosetta and Seti for ages and gets what the project gives it no matter what it claims. I have not bothered to update it as it always works.
When I added it to Docking I shot myself in the foot as the system saw the benchmark (which is supposed to be an indication of processing power)and said "this is a fast computer so I will download a lot of work to keep it going". It downloaded about 19 WU's at 46 minutes each.
When processing it takes 8 hours 50 minutes not 46 minutes, so I calculated that I was then unable to process all those WU's on a single core cpu and have had to abort 10 of the WU's.
I will be updating this computer with the new standard 5.4.11? when I get a chance and have stopped any more work going to this machine until it can complete all it now has.
____________

Even i felt it way overclaimed. to close on windows i needed to claim around 80 credits per WU. I claimed around 115 so im taking it off. its just too much of a claim.

Even i felt it way overclaimed. to close on windows i needed to claim around 80 credits per WU. I claimed around 115 so im taking it off. its just too much of a claim.

Hmm, you're right - I didn't know it overclaimed that much on Linux too. Of course you still only got 16 credits for 7.5 hours work because of the quorum.
I take back my comments about this client being OK for Linux.

Are these proposed by Boinc, and granted credit, on par with most Linux Boxes?

Are these proposed by Boinc, and granted credit, on par with most Linux Boxes?

I have no idea. I'm just another cruncher (I don't work for BOINC or this project) but with the standard BOINC 5.4.11 on my windows boxes I am getting between 70 and 80 credits per work unit (and as you can see from my credits, I have not yet done many WU).

The closest match I have for your PC is a 2.4GHz Pentium 4, which has only done one work unit to date. It took much longer than your Celeron and claimed 4.5 times as much credit. It's hard to draw any conclusions from that.

My P4 claimed 7.57 credits per hour. Your Celeron claimed about 3.81 credits per hour (based on the average of the last 3 WU). They're not identical machines so I'd expect there be a difference, though I would not think the P4 is twice as powerful.

IMHO you'd have to run the same WU on one machine, once under Linux, once under Windows, to get comparative data (if it could be done)

Are these proposed by Boinc, and granted credit, on par with most Linux Boxes?, or, way below?

Are these proposed by Boinc, and granted credit, on par with most Linux Boxes?, or, way below?

Hi Doug,

I'm still fighting with my Celeron 2.4 GHz Linux box, but I'll put some links to my other 2 boxes below. You can click on the results link in the summary and see their credit history. All are running the latest 5.4.x stock client downloaded from the Boinc website.

Linux FC3 - Socket A Sempron 2500+

Windows XP - Celeron 2.3 GHz

Hope this helps,

-- David