Pages: [1] 2 3
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 140,887,160
World-rank: 7,394

2007-09-07 13:54:26
last modified: 2007-09-07 13:54:53

Project abstract

Radio Network Design (RND) is a telecommunication problem which consists in covering a geographic area with radio signal using the fewer number of transmitters that cover the maximum area. Therefore, it is an important topic to solve nowadays, for example in the mobile wireless technology domain, mobile telephony,... Bio-inspired algorithms are a good choice to solve the RND problem, because this is an NP-hard optimization problem which fits bio-inspired algorithms very well. In this project it is used the PBIL (Population-Based Incremental Learning) algorithm. This is a modern algorithm, and we hope to get good results with it. PBIL is based on genetic algorithms and competitive learning (typical in neural networks).

Homepage of project
They have no forum, so I asked the admin to post here until they managed to start one Here's my mail correspondence with him from today:
Hello Uwe:

First of all, thanks for your interest in RND@home.

I answer your doubts:

* Is your project indented to be open?

Yes, it is an open project. This is our first experience with BOINC projects and we did not know what steps to follow in order to diffuse the project. For this reason, initially, only 7 crunchers were present.

* Have I invaded a private project?

No. Don't worry. It is not a private project.

* Do you plan to open up the forum?

Yes, we have planned several improvements in our project (including to open up the forum). We need some time (this is our first experience with BOINC).

* Is there a way to look at the results?

This is other planned improvement: the development of screensaver graphics in order to show to each cruncher his/her achievements in the RND@home project. Anyway, I attach you a scientific paper showing some of the most important scientific results of this BOINC project. This paper was published in the Spanish Conference on e-Science Grid Computing 2007, as you can see in http://webrt.ciemat.es:8000/e-science/program.html (look for "Vega&quot.

* Will you put the credits in better relation to other projects?

You have reason. We have multiplied the credits by 100. We did that because our WU are very heavy/hard (from 1 hour to 12 hours, that is, they are difficult to complete succesfully), and this is a way to give our crunchers many credits for few WUs. We have to reward their fidelity in some way. Anyway, all the crunchers in our BOINC project obtain the same benefits.

Best regards, and again, thanks for your interest in RND@home.

Miguel.-

--
------------------------------------------------------------
Dr. Miguel Angel Vega-Rodríguez <mavega@unex.es>
http://arco.unex.es/mavega
Univ. Extremadura
Dept. Tecnología de los Computadores y de las Comunicaciones
(Dept. Technologies of Computers and Communications)
Escuela Politécnica. Campus Universitario, s/n
10071 Cáceres. España (Spain)
------------------------------------------------------------

Saenger escribió:
> Hello David and Migeul Ángel,
>
> Since yesterday your project RND@Home is public knowledge, I joined as well, but after crunching my first WU I became doubts about the openness of the project, whether it was really intended to get new crunchers and how to edal with it further on.
>
> First, my first WU crunched obviously fine during the night, my puter still runs and has succesfully completed and reported it back to your server. Unfortunately I can't see the result, but I can see the granted credit in my account. They are plain absurd, even without the proper knowledge of the real runtimes. It was about 10% for the first hour, it ran 5:46:23 in clock time on my puter from start until upload and generated 12,335.37 credits, that's about 100 times the usual amount.
>
> There were only 7 crunchers with credit as I started, with dubious numbers, so I decided to ask now what it's about:
>
> * Is your project indented to be open?
> * Have I invaded a private project?
> * Do you plan to open up the forum?
> * Is there a way to look at the results?
> * Will you put the credits in better relation to other projects?
>
> I've aborted all but 2 of the 19 WUs you sent me, they would never have made the deadline. I won't DL any more WU until some notice from you, either in the news on the page or per mail.
>
> Thx in advance,
> Uwe (Saenger)
> http://arcoboinc.unex.es/rnd/show_user.php?userid=12
Grüße vom Sänger
PovAddict
BAM!ID: 115
Joined: 2006-05-10
Posts: 1013
Credits: 4,359,181
World-rank: 86,867

2007-09-07 17:26:04

You have reason. We have multiplied the credits by 100. We did that because our WU are very heavy/hard (from 1 hour to 12 hours, that is, they are difficult to complete succesfully), and this is a way to give our crunchers many credits for few WUs. We have to reward their fidelity in some way. Anyway, all the crunchers in our BOINC project obtain the same benefits.

What the... 1 hour to 12 hours is "very heavy/hard" in comparison to WHAT? And anyway, there is no reason to multiply the credits. BOINC has always given more credits to longer workunits. Not crunching until their credit matches other projects; it's just not fair.
Nightbird
BAM!ID: 132
Joined: 2006-05-11
Posts: 85
Credits: 0
World-rank: 0

2007-09-07 18:03:32
last modified: 2007-09-07 18:07:18

We have multiplied the credits by 100. We did that because our WU are very heavy/hard (from 1 hour to 12 hours, that is, they are difficult to complete succesfully), and this is a way to give our crunchers many credits for few WUs. We have to reward their fidelity in some way.

Seems to me that now we know also the credit - inflation thanks to some new projects.
Since they claim that "this is our first experience with BOINC", maybe they can stop to play with credits ?
What's the next step ? they will export stats without meaning ?
zombie67
BAM!ID: 1560
Joined: 2006-06-04
Posts: 606
Credits: 170,281,612,131
World-rank: 47

2007-09-07 18:08:55

Yowza. A P4 with a RAC of over 11,000.
Reno, NV
Team: SETI.USA
Nightbird
BAM!ID: 132
Joined: 2006-05-11
Posts: 85
Credits: 0
World-rank: 0

2007-09-07 18:20:13

See the first
credits : 13,304,441.15
rac : 59,324.65

They don't control their credit system but they are going public !
Guest

2007-09-07 19:06:59

You have reason. We have multiplied the credits by 100. We did that because our WU are very heavy/hard (from 1 hour to 12 hours, that is, they are difficult to complete succesfully), and this is a way to give our crunchers many credits for few WUs. We have to reward their fidelity in some way. Anyway, all the crunchers in our BOINC project obtain the same benefits.

What the... 1 hour to 12 hours is "very heavy/hard" in comparison to WHAT? And anyway, there is no reason to multiply the credits. BOINC has always given more credits to longer workunits. Not crunching until their credit matches other projects; it's just not fair.


I just downloaded about 10 WUs on my single, modest, host, should have been about 4 hours work. But each WU downloaded looks like coming in at 72 hours! No way am I going to complete that workload. Thankfully I didn't take a full days worth ;-)

I may let one WU run for a day or so, but that's all. I run a semi-random mix of projects and don't want to be hijacked by crazy WU runtimes. Maybe that's why the credit is so high - To get some half decent credit from a tiny WU that takes days to run?

Al.


Al.

Nightbird
BAM!ID: 132
Joined: 2006-05-11
Posts: 85
Credits: 0
World-rank: 0

2007-09-07 19:17:35

I just downloaded about 10 WUs on my single, modest, host, should have been about 4 hours work. But each WU downloaded looks like coming in at 72 hours! No way am I going to complete that workload. Thankfully I didn't take a full days worth ;-)

I may let one WU run for a day or so, but that's all. I run a semi-random mix of projects and don't want to be hijacked by crazy WU runtimes. Maybe that's why the credit is so high - To get some half decent credit from a tiny WU that takes days to run?

Al.


Al.


They're not ready to go public, that's all.
PovAddict
BAM!ID: 115
Joined: 2006-05-10
Posts: 1013
Credits: 4,359,181
World-rank: 86,867

2007-09-07 21:50:47

I may let one WU run for a day or so, but that's all. I run a semi-random mix of projects and don't want to be hijacked by crazy WU runtimes. Maybe that's why the credit is so high - To get some half decent credit from a tiny WU that takes days to run?

Credit is based on the actual time the workunit takes, not the estimation done from the project.

@Willy, please don't add the project to BOINCstats until the credits are sane.
DoctorNow
    Donator
Volunteer tester
BAM!ID: 261
Joined: 2006-05-12
Posts: 350
Credits: 2,422,701,798
World-rank: 944

2007-09-07 22:09:52
last modified: 2007-09-07 22:10:11

The credits are really incredible...
I crunched 2 WUs under Linux 64 Bit, that were the results:
ca. 3h 50 min -> 7.103,01 Credits
ca. 4h 03 min -> 7.547,86 Credits
That would make an hour average of about 3.100 credits (6.200 for my AMD X2 )


That's really no comparison to any of the other projects and completely crazy.
BOINC@Heidelberg & BOINC Confederation-member

My BOINCstats/[url=http://signature.statseb.fr/sig-335.png]My badges
Barbud [USA]
 
BAM!ID: 5198
Joined: 2006-09-02
Posts: 83
Credits: 1,445,882
World-rank: 169,714

2007-09-08 19:32:49

Yowza. A P4 with a RAC of over 11,000.


Hey Zombie, lets see what you can get on this. 500k rac?
zombie67
BAM!ID: 1560
Joined: 2006-06-04
Posts: 606
Credits: 170,281,612,131
World-rank: 47

2007-09-08 22:44:35

Yowza. A P4 with a RAC of over 11,000.


Hey Zombie, lets see what you can get on this. 500k rac?


Let's see...

My 8-way mac on SETI, with alexkan's optimized application, generated a RAC to 10k. My P4 (same as the one I mentioned above), is 1/20th as fast (if that). So I should be able to get at least 220k RAC for my mac alone. Add my 2 other 8-way (X5355) windows machines, and I should be well over 600k. Add all of my machines, and I should be 1M+ RAC.

Hey! That's more than NEZ!!!
Reno, NV
Team: SETI.USA
Guest

2007-09-10 13:03:24
last modified: 2007-09-10 13:16:16

I have an Account at the Project but pulled my PC's from it once I found out the Credit Situation, anybody that Processes WU's for the Project is basically Cheating knowing the Credits are completely ludicrous.

Yes it's the Project giving the Credit but everybody knows it's not right so to go and capitalize on the Credits is wrong. One of the French Team Members has put up close to 240,000 Credits from 1 8 Core PC in less than 3 days Time and other people will probably do the same trying to leapfrog ahead of people in the Stats.

All the Stats Sites need to take a stand against this Project and make an Announcement that their Stats will not be put up until the Project rectifies the Credits since they have admitted to increasing them 100 Fold.
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 140,887,160
World-rank: 7,394

2007-09-10 13:56:09

I recieved an email from the admin with thanks about this thread, I thought he would post here as well. I've told him in a second mail, an answer to the one qouted in this thread, that I don't think his decision regarding credits would be recieved well within the community to say the least. I see that this seems to have been quite right.

I hope he will come here with some explanations, and imho a solution would be really simple: divide all credits by 100, everything's fine. It should be easy to implement even retroactive and doesn't disturb the standing in the project intern stats at all.
Grüße vom Sänger
[AF>HFR>RR]Sp0wn
BAM!ID: 15321
Joined: 2006-12-21
Posts: 1
Credits: 123,548,345
World-rank: 8,043

2007-09-10 19:40:19

I have an Account at the Project but pulled my PC's from it once I found out the Credit Situation, anybody that Processes WU's for the Project is basically Cheating knowing the Credits are completely ludicrous.

Yes it's the Project giving the Credit but everybody knows it's not right so to go and capitalize on the Credits is wrong. One of the French Team Members has put up close to 240,000 Credits from 1 8 Core PC in less than 3 days Time and other people will probably do the same trying to leapfrog ahead of people in the Stats.

All the Stats Sites need to take a stand against this Project and make an Announcement that their Stats will not be put up until the Project rectifies the Credits since they have admitted to increasing them 100 Fold.



Hi poorboy , that's why I ve put this project in NMW, because I don't want to be consider as a cheater !
Bruno G. Olsen & ESEA @ greenholt
 
BAM!ID: 2303
Joined: 2006-06-17
Posts: 43
Credits: 39,571,456
World-rank: 17,938

2007-09-10 20:29:04

Hi poorboy , that's why I ve put this project in NMW, because I don't want to be consider as a cheater !


So have I, along with ABC due to their 1k wu's
Nightbird
BAM!ID: 132
Joined: 2006-05-11
Posts: 85
Credits: 0
World-rank: 0

2007-09-11 11:00:36

Hi poorboy , that's why I ve put this project in NMW, because I don't want to be consider as a cheater !


So have I, along with ABC due to their 1k wu's

Would be curious to see this project here.
vorik3
BAM!ID: 962
Joined: 2006-05-29
Posts: 5
Credits: 0
World-rank: 0

2007-09-11 12:50:23
last modified: 2007-09-11 12:50:48


Would be curious to see this project here.


Not possible, they don't export stats. And if they are THAT far off from the rest of the BOINC world, I would consider not adding them at all...
[AF>EDLS>BIOMED] Heyoka
BAM!ID: 1578
Joined: 2006-06-04
Posts: 3
Credits: 8,405,813
World-rank: 55,560

2007-09-11 14:17:24
last modified: 2007-09-11 14:18:51


Would be curious to see this project here.


A good idea would be to put ratio for counting the Boinc combined stats.
This to avoid the project administrator increase the credit for the sake of attract more computing power

ABC, Cosmology Riesel, QMC, SHA-1 Collision Search Graz attract many people, that isn't this projects are more interesting than the other. It's because the project administrator are cheating.

Currently many people choose their project for credit and not for the matter of the project and it's not good.
stwainer
BAM!ID: 210
Joined: 2006-05-12
Posts: 31
Credits: 37,195,767
World-rank: 18,760

2007-09-11 15:19:30

It's because the project administrator are cheating.

Currently many people choose their project for credit and not for the matter of the project and it's not good.


And how are you determining these facts?

I crunch one of the projects mentioned because I think it's worth doing.
ebahapo
 
BAM!ID: 239
Joined: 2006-05-12
Posts: 661
Credits: 33,388,440
World-rank: 20,330

2007-09-11 15:26:15

Oh, you babies! How can there be cheating when there's nothing to be gained?

Yes, RND has a screwed up credit system, but the question is rather whether its goals are worthy of our systems idle cycles. As far as I can tell, they are.

Other projects made mistakes in the beginning too. And although it's not stated anywhere int he project site, I assume that it's still in alpha or beta, so they have their quota of mistakes to fulfill. Of course, I hope that RND will shape up soon.
[AF>EDLS>BIOMED] Heyoka
BAM!ID: 1578
Joined: 2006-06-04
Posts: 3
Credits: 8,405,813
World-rank: 55,560

2007-09-11 15:34:35
last modified: 2007-09-11 15:38:18


And how are you determining these facts?

I crunch one of the projects mentioned because I think it's worth doing.


Yes, you are not to blame. you crunch for this project you find interesting.
It's the project administrator who deal too much credit for attract more people. And maybe 25 % of the people on this project are here strictly for credits

Example with 2 projects pretty much similar : Riesel Sieve and PrimeGrid

Riesel Sieve : 5,879.6 GigaFLOPS
PrimeGrid : 2,316.0 GigaFLOPS

The gap is due to the increase of credit on Riesel Sieve + people attract with easy credit


Guest

2007-09-11 21:56:57

The gap is due to the increase of credit on Riesel Sieve + people attract with easy credit


Some of the gap is due to the fact that PrimeGrid has a Quorum on their WU's and Riesel Sieve doesn't, unless you run the LLR Wu's which few people do.

I know if I have my choice between 2 similar Projects & 1 has a Quorum & 1 doesn't I'm going with the one 1 that doesn't every time.
picantecomputing
BAM!ID: 19319
Joined: 2007-02-14
Posts: 172
Credits: 2,486,623
World-rank: 123,914

2007-09-12 00:47:02
last modified: 2007-09-12 00:47:39

Hey Willy, in light of this thread, is there any way to give each project an average credit per hour crunched rating by taking their total credit and dividing by total time crunched? I know there's already kind of something like that here, but that's way too complicated for my taste. And yes, I know every processor gets a different amount of credit, and I know there's no way to have a consistent benchmark. But I'm talking about taking just the grand totals and calculating a master average for each project. In a way I think it's the most sensible way to compare credit among multiple projects, as it takes all crunchers working on those projects into account, thereby smoothing out the differences between processors. That way it'd be easy to get a big-picture view of which projects are out of line (this one obviously is), and how each ranks. Would also be very informative for most who use this site, I think.

This sort of goes along with this post I made a few weeks ago in Comments & Suggestions, as far as how it might be calculated, if the raw time numbers aren't available in the exported stats.
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 140,887,160
World-rank: 7,394

2007-09-13 15:44:55

I've got mail:
Hello Uwe,

Thanks for your e-mail.

We have already corrected the credit system of RND@home:
1. Since all the credits were multiplied by 100, we have divided the current credits of all our crunchers by 100.
2. We have fixed the problem. Now, the credits are not multiplied by 100, so we give credits in the same way as other BOINC projects.

Please, if it is possible for you, check all this by crunching a WU. Also, please, inform all the people that now our credit system is OK. Thanks in advance.

Another important issue. In the last days we have observed that the WUs of RND@home works very fine under Linux, but they are very slow under Windows (of course, thanks to Windows ), and for this reason, many WUs executed under Windows have problems with the deadlines. Please, if it is possible, inform all the people that in order to crunch our WUs it is better to use a Linux BOINC client (of course, if it is possible).

Best regards, and again, thanks for all your help.

Miguel.-


So it's OK to crunch now
Grüße vom Sänger
zombie67
BAM!ID: 1560
Joined: 2006-06-04
Posts: 606
Credits: 170,281,612,131
World-rank: 47

2007-09-13 16:48:07

Great! Can it please be added to stats and BAM?
Reno, NV
Team: SETI.USA
Pages: [1] 2 3

Index :: New projects :: RND@Home
Reason: