Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Community > Team MacNN > Rosetta Downgrading Credits From Optimized Clients

Rosetta Downgrading Credits From Optimized Clients
Thread Tools
OneMacGuy
Senior User
Join Date: Oct 2002
Location: God's Country, The South
Status: Offline
Reply With Quote
Aug 17, 2006, 06:55 PM
 
I just happened to notice that Rosetta has started giving SUBSTANTIALLY less credits that claimed when you are running an optimized client. Thier forums are on fire over this issue and it does really suck. I have just about had it with this crap. Spend all this time and money buying, building and tuning crunchers to get the max performance only to hav ethem change the rules again.

I am seriously thinking of quitting this entirely, think of the money I would save!
Chappaquidick 1, Cheney 0
     
OneMacGuy  (op)
Senior User
Join Date: Oct 2002
Location: God's Country, The South
Status: Offline
Reply With Quote
Aug 17, 2006, 07:14 PM
 
Originally Posted by OneMacGuy
I just happened to notice that Rosetta has started giving SUBSTANTIALLY less credits that claimed when you are running an optimized client. Thier forums are on fire over this issue and it does really suck. I have just about had it with this crap. Spend all this time and money buying, building and tuning crunchers to get the max performance only to hav ethem change the rules again.

I am seriously thinking of quitting this entirely, think of the money I would save!
After looking some more at my new credits granted. I am outta there.

<table border="1" cellpadding="5" width="100%"><tbody><tr><td>32999642</td> <td>28597714</td> <td>16 Aug 2006 6:35:15 UTC</td> <td>17 Aug 2006 5:51:54 UTC</td> <td>Over</td> <td>Success</td> <td>Done</td> <td>27,448.13</td> <td>142.27</td> <td>26.29</td> </tr> <tr><td>32987194</td> <td>28585856</td> <td>16 Aug 2006 4:35:00 UTC</td> <td>17 Aug 2006 5:25:29 UTC</td> <td>Over</td> <td>Success</td> <td>Done</td> <td>28,046.95</td> <td>145.37</td> <td>26.68</td> </tr> <tr><td>32964659</td> <td>28564534</td> <td>16 Aug 2006 0:59:05 UTC</td> <td>17 Aug 2006 5:25:29 UTC</td> <td>Over</td> <td>Success</td> <td>Done</td> <td>26,462.12</td> <td>137.16</td> <td>23.49</td> </tr> <tr><td>32900415</td> <td>28503943</td> <td>15 Aug 2006 13:09:14 UTC</td> <td>16 Aug 2006 16:31:25 UTC</td> <td>Over</td> <td>Success</td> <td>Done</td> <td>23,677.14</td> <td>122.72</td> <td>122.72</td> </tr> <tr><td>32858797</td> <td>28464636</td> <td>15 Aug 2006 6:26:06 UTC</td> <td>16 Aug 2006 8:13:53 UTC</td> <td>Over</td> <td>Success</td> <td>Done</td> <td>24,808.14</td> <td>128.58</td> <td>128.58</td></tr></tbody></table>
Chappaquidick 1, Cheney 0
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 17, 2006, 07:53 PM
 
Originally Posted by OneMacGuy
After looking some more at my new credits granted. I am outta there.
A few bits of info:

1) They are currently showing the results for both methods (old and new). Stat sites are still looking at the old version.

2) The new method is still being tested. People are complaining it is not consistent or fair yet.

3) (And here's the *real* problem) R@H has announced that once they work the bugs out of the new system, it will be applied retoractively back to February. Some people are saying "that's not true! They never said that!" But they sure did over on the RALPH forum. Of course that site is now conveniently down, so no one can verify...and the saff is silent on the issue. We'll see.

Personally, I don't have a problem if they want to change it going forward. It's their choice, and we can freely make the decision to stay or go.

But to apply it retroactively is complete BS. Had I known, I would have been using those hours to cruch something else like SETI with our mac optomized clients.

R@H cannot make up for my/our lost opportunity. It's nothing short of theft and deceit. They lured us there with a sweet scoring system, and openly said time after time that optomized clients were not against the rules. They got the cycles they wanted from us with the promise of fat credits, and are now taking it back after real time invested that we cannot get back.
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 17, 2006, 08:14 PM
 
Stats mean nothing to me.
Pretty funny since I have the most power and make optimized clients.
I am happy to see them make everything fair.
In my mind, if things are based on speed, the slower the machine, the higher the penalty should be because that user is wasting more resources (power, time, etc). This would mean those with the latest and greatest would almost always be best.

Really though, the competition shouldn't have anything to do with speed but with results.
The coolest folded protein, most useful, or whatever should get the bonus points.
That is kinda hard to judge if nothing is ever really achieved.
In 10+ years of DC, I've seen nothing significant actually come of any of it.
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 17, 2006, 08:34 PM
 
Originally Posted by mikkyo
Stats mean nothing to me.
Pretty funny since I have the most power and make optimized clients.
I am happy to see them make everything fair.
In my mind, if things are based on speed, the slower the machine, the higher the penalty should be because that user is wasting more resources (power, time, etc). This would mean those with the latest and greatest would almost always be best.

Really though, the competition shouldn't have anything to do with speed but with results.
The coolest folded protein, most useful, or whatever should get the bonus points.
That is kinda hard to judge if nothing is ever really achieved.
In 10+ years of DC, I've seen nothing significant actually come of any of it.
I agree that the more work completed, the more credits should be given. I would even applaud a changing the credit system accordingly. My problem is with making the changes retroactive.

As for credits, they are very meaningful to me, and a boon to the projects. The competition is why I am crunching with 14 machines instead of just my daily box. Those extra boxes wouldn't exist without it.
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 18, 2006, 12:18 AM
 
Originally Posted by OneMacGuy
After looking some more at my new credits granted. I am outta there.
Don't feel so bad..<table border="1"><tbody><tr><td>Claimed Credit</td><td>Granted Credit</td></tr>
<tr><td>237.58</td><td>38.27</td></tr>
<tr><td> 239.30</td><td>38.23</td></tr>
<tr><td> 255.15</td><td>44.40</td></tr>
<tr><td>238.77</td><td>38.71</td></tr>
<tr><td>249.62</td><td>56.21</td></tr>
<tr><td>232.26</td><td>38.99</td></tr></tbody></table>
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 18, 2006, 01:33 AM
 
Originally Posted by mikkyo
Don't feel so bad..
'fess up...

You have the *other* quad G5 in the top 5*...right?

*Average Credit
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 18, 2006, 02:44 AM
 
Those stats are from an Intel machine.
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 18, 2006, 03:48 AM
 
Originally Posted by mikkyo
Those stats are from an Intel machine.
Sly dog. That wasn't an answer.
     
Knightrider
Dedicated MacNNer
Join Date: Sep 2004
Location: London
Status: Offline
Reply With Quote
Aug 18, 2006, 05:34 AM
 
Originally Posted by mikkyo
Don't feel so bad..<table border="1"><tbody><tr><td>Claimed Credit</td><td>Granted Credit</td></tr>
<tr><td>237.58</td><td>38.27</td></tr>
<tr><td> 239.30</td><td>38.23</td></tr>
<tr><td> 255.15</td><td>44.40</td></tr>
<tr><td>238.77</td><td>38.71</td></tr>
<tr><td>249.62</td><td>56.21</td></tr>
<tr><td>232.26</td><td>38.99</td></tr></tbody></table>
As we all know, these kind of results are inherent to the quorum system, which was designed specifically to filter out any 'adjustments' to the amount of credit given for wu's. Fair enough.

Originally Posted by mikkyo
In my mind, if things are based on speed, the slower the machine, the higher the penalty should be because that user is wasting more resources (power, time, etc). This would mean those with the latest and greatest would almost always be best.
I agree with this. What is unfair is that by the quorum system, faster, gives lower credit.This is called discrimination. It is not a level playing field. In the real world, a level playing field means that every one starts at the same point and first passed the post wins. Not that better players have to have one leg hobbled. Yes, it is all for the sake of the science, but there has to be some kind of quid pro quo here.

What I don't understand is why, when there are perfectly good optimisations that do the work and produce more results, they are not incorporated into the project.That's what 'open source' is all about, isn't it?

K.



.
     
OneMacGuy  (op)
Senior User
Join Date: Oct 2002
Location: God's Country, The South
Status: Offline
Reply With Quote
Aug 19, 2006, 08:02 AM
 
Originally Posted by Knightrider
As we all know, these kind of results are inherent to the quorum system, which was designed specifically to filter out any 'adjustments' to the amount of credit given for wu's. Fair enough.


I agree with this. What is unfair is that by the quorum system, faster, gives lower credit.This is called discrimination. It is not a level playing field. In the real world, a level playing field means that every one starts at the same point and first passed the post wins. Not that better players have to have one leg hobbled. Yes, it is all for the sake of the science, but there has to be some kind of quid pro quo here.

What I don't understand is why, when there are perfectly good optimisations that do the work and produce more results, they are not incorporated into the project.That's what 'open source' is all about, isn't it?

K.

.
But, they gave out the source code to Boinc and said basically that you are free to tweak the boinc client and use it. A few people did with great results.

I have spent a LOT of time and pretty bit of money installing this client on a lot of computers and running them here in my home. Time finding and learning to tune these optimized clients and getting them installed several times to keep them up to date.

And NOW, they say that all of that was wrong? How many more times are they gonna change the rules and throw out a lot of what I have worked and paid for?

BUT, the real kicker in the end was a stament made by Mikkyo (not bad Mikkyo!):

Originally Posted by Mikkyo
In 10+ years of DC, I've seen nothing significant actually come of any of it.
Why bother with all this work and expense if it is accomplishing NOTHING?

I am starting to pull my computers off the project as I get around to them.
Chappaquidick 1, Cheney 0
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 23, 2006, 08:31 PM
 
New credit system went live yesterday. Of course, they told no one until today...and that was only after it was pointed out by users.

Mikkyo went from 40k+/day to about 8.5k/day now. I went from a max of around 20k/day to now about 4k/day.
     
Knightrider
Dedicated MacNNer
Join Date: Sep 2004
Location: London
Status: Offline
Reply With Quote
Aug 23, 2006, 08:51 PM
 
Originally Posted by zombie67
New credit system went live yesterday. Of course, they told no one until today...and that was only after it was pointed out by users.

Mikkyo went from 40k+/day to about 8.5k/day now. I went from a max of around 20k/day to now about 4k/day.
I think we can safely conclude that they have no interest in work rates.

K.
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Aug 23, 2006, 09:49 PM
 
An important thing to keep in mind is that it is not only the optimized Mac clients that are affected, but also the optimized Windows and Linux clients. All users are affected equally for the most part.

And the purpose of open sourcing the code was not to allow enhancements per platform but rather to allow individuals to compile for other alternate platforms. The ability to optimize was just a side effect of that.

An life goes on, as does crunching.
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 23, 2006, 10:31 PM
 
Originally Posted by Shaktai
An important thing to keep in mind is that it is not only the optimized Mac clients that are affected, but also the optimized Windows and Linux clients. All users are affected equally for the most part.
Absolutely. There are two issues I have with it:

1) Again, they told everyone about it after the fact. You cannot afford to piss off people when you are asking them for handouts. Clear communication with plenty of advanced warning for changes is critical in keeping the people happy. Even after the cluster F they cased two weeks ago, they still have not learned the lesson.

2) They said that everyone would get the same credit per job. Work faster get more credits. Fair enough. Turns out that is NOT the case. It uses a running average per structure. So the first guy gets exactly whatever he claims. The second gets the average of the two, the third gets the average of the 3, and so on. So if you use an optomized client, and get lucky enough to return first, you hit the lotto. And the further away you get from being first, the less advantage you get. The unintended consequence is that everyone (who cares) will set their run time to 1 hour. Not sure what the impact of that will be in the long run.
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 24, 2006, 02:31 PM
 
I'm sure number 1 annoyed a ton of people and is not a very nice way to do things.
Number 2 is just plain silly. Just give everyone credit based on WU or use the new system Seti is.

The new silly average system is great for one thing though, pushing Macs back up into the top ten.
I imagine in a few days, once the RAC stabilizes, the power of macs will be visible to all again.
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 26, 2006, 02:20 AM
 
Originally Posted by mikkyo
Don't feel so bad.. (table deleted)
mikkyo, did you pull out, or are the stats showing what you are generating with the new credit system?

Two points:

1) My daily rate went up when I increased my run time from 1 to to 6 hours (from high 5k to low 7k total credits/day). I am trying out 8 hours now.

2) Changing the power setting from auto to max on my quad g5 increased the credits from ~40 to ~70 credits per 6 hr run. Odd because the menumeters were maxed. I am guessing it wasn't really running at max because the process/nice is so low. I just noticed the settings a few days ago.
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 26, 2006, 01:43 PM
 
Im still crunching away.
Using the -return_results_immediately flag will help RAC too.
I run everything at nice -20 when the machines arent in use for other things.
( Last edited by mikkyo; Aug 29, 2006 at 04:34 AM. )
     
zombie67
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Aug 27, 2006, 10:19 AM
 
Originally Posted by zombie67
1) My daily rate went up when I increased my run time from 1 to to 6 hours (from high 5k to low 7k total credits/day). I am trying out 8 hours now.
At 8 hour run time, I'm in the high 6k range over the past day. There appears to be no patern to this at all. I'm going back to 6 for a while.
     
ChillyWilly5280
Junior Member
Join Date: Jun 2006
Status: Offline
Reply With Quote
Aug 28, 2006, 11:12 PM
 
Originally Posted by mikkyo
Im still crunching away.
Using the -return-results-immediately flag will help RAC too.
I run everything at nice -20 when the machines arent in use for other things.
Which file is the -return-results-immediately flag in and what should it read?

Thanks, Chris
     
mikkyo
Senior User
Join Date: Feb 2002
Location: Silly Valley, Ca
Status: Offline
Reply With Quote
Aug 29, 2006, 04:34 AM
 
it is a command line argument
I dont know if it is in the GUI anywhere.
./boinc_5.4.9Superbench -return_results_immediately
     
ChillyWilly5280
Junior Member
Join Date: Jun 2006
Status: Offline
Reply With Quote
Aug 29, 2006, 10:14 AM
 
Originally Posted by mikkyo
it is a command line argument
I dont know if it is in the GUI anywhere.
./boinc_5.4.9Superbench -return_results_immediately
Gotcha, thanks.
Chris
     
   
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 10:31 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,