Welcome to MilkyWay@home

And if we DONT run a "FASTER" App.?


Advanced search

Message boards : Number crunching : And if we DONT run a "FASTER" App.?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · Next

AuthorMessage
ProfileKevint
Avatar

Send message
Joined: 22 Nov 07
Posts: 285
Credit: 1,076,786,368
RAC: 0
1 billion credit badge10 year member badge
Message 5904 - Posted: 4 Nov 2008, 5:25:18 UTC - in response to Message 5903.  



So I guess I am to assume that the limit is 108 credit per hour... Is this per core or per computer ????
ID: 5904 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Benjamin Rietveld

Send message
Joined: 26 Sep 08
Posts: 11
Credit: 17,597
RAC: 0
10 thousand credit badge10 year member badge
Message 5905 - Posted: 4 Nov 2008, 6:21:39 UTC

it is per core

ps. the ignorance in this thread is really surprising
ID: 5905 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 519
Credit: 284,026,803
RAC: 5,988
200 million credit badge10 year member badgeextraordinary contributions badge
Message 5906 - Posted: 4 Nov 2008, 6:57:42 UTC - in response to Message 5885.  

Ah -- that's what I meant -- so you get the completed work for your use and we don't get 'credit' for it. OK -- so then the question is, what is the hourly credit limit? I don't think I bounce up against it even on my quad cores as I run other projects as well. Based on the numbers though, it may change the resource shares I have running.


The work doesnt get zeroed out, you just don't get credit for it if you're over the limit. Also, regarding the Mac version, we're raising the credits back up because we think it's unfair to those that don't use the optimized app and either lower the per hour limit or keeping it the same (that's travis's deal).


ID: 5906 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cherry

Send message
Joined: 31 Oct 08
Posts: 1
Credit: 989,726
RAC: 0
500 thousand credit badge10 year member badge
Message 5907 - Posted: 4 Nov 2008, 7:25:37 UTC
Last modified: 4 Nov 2008, 7:26:37 UTC

The limit is 0.03 cr/sec. So just look at how long a WU needs on your system, multiply it bei .03 and you know how much credit you'll get for that WU. So no "zero credit scenario", you just get lower credits per WU, the faster you crunch.
So just reactivate all your stone-age computers, even pentium 200 should be fast enough to reach the .03 cr/sec.

Or pressure Travis so he finally does his work and invest the few days that are needed (as shown by the milksop-guys) to bring a decent app to _all_ milkyway crunchers. Then there wouldn't be a need for the optimised app, and they could give fair credit based on their own, fast stock application.

As long as Travis doesn't do his homework, i don't see a solution for milkyway, that'll make everyone happy.
ID: 5907 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
6dj72cn8

Send message
Joined: 26 Dec 07
Posts: 41
Credit: 722,567
RAC: 1,497
500 thousand credit badge10 year member badge
Message 5908 - Posted: 4 Nov 2008, 8:07:27 UTC - in response to Message 5885.  

Also, regarding the Mac version, we're raising the credits back up because we think it's unfair to those that don't use the optimized app and either lower the per hour limit or keeping it the same (that's travis's deal).


Without buying in to the rights and wrongs of the current situation, I merely want to say thanks to Dave and Travis for putting the credit back up so that those of us without access to the optimised app still have the option of continuing on as before. Even if it does take us hours to do what others can do in minutes. That said, I do hope a new stock app, equally speedy for all platforms, is released soon.

ID: 5908 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
50 million credit badge10 year member badge
Message 5909 - Posted: 4 Nov 2008, 8:57:12 UTC - in response to Message 5908.  

That said, I do hope a new stock app, equally speedy for all platforms, is released soon.


I believe that is planned to be released this week. So, that new faster MW stock client will be the norm and all credit argument should stop.

This new stock client should replace the current slow stock client. As I understand it, the new stock client has been changes slightly, for the purposes of the science, but it uses the same/similar fast code to Milksop's.

As this is becomes the stock client, the cross project credit will be based on it and people have the choice to move to the new stock client (or not). The output of results will then be on their choice of client, and should put a stop to all this whinging about the credit given.
ID: 5909 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileArion
Avatar

Send message
Joined: 10 Aug 08
Posts: 218
Credit: 41,846,854
RAC: 0
30 million credit badge10 year member badge
Message 5912 - Posted: 4 Nov 2008, 11:06:26 UTC - in response to Message 5909.  
Last modified: 4 Nov 2008, 11:16:36 UTC

As this is becomes the stock client, the cross project credit will be based on it and people have the choice to move to the new stock client (or not). The output of results will then be on their choice of client, and should put a stop to all this whinging about the credit given.


Well after spending an hour playing with the calculator thinking I was going to accomplish anything worth discussing in the way of credits I've given up and decided to just go with the flow.

About the best I can make out, I work with one project and it takes about 8 hours to finish a work unit that I get 222 credits for, another I get about a little over 370 credits for 24 hours. Both projects I've been with for a number of years and I'm happy with the credit I'm getting for the work I'm doing. To be perfectly honest I was happy with the credit I was getting here with the unoptimized client. Since the client was so inefficient and has now been optimized either as an outside party's application or the official one from MW, I don't really care what the credit per WU is so long as it is something above 0.

The way I look at it if I get 108 credits an hour limit no matter how many WU's I've done I'm still about 3 times or more than what I get with any other project based on credits per hour. Of course this is all based on the fact that the new optimized client or something close to it becomes the standard application since it will be about the equivilant of other projects optimizations of their clients over time.

For those who won't be able to use the new standard application I don't know what to do to make it fair. I would like to think that the project would try to make it a priority to have other optimized clients for the most common OS's/processors as soon as possible. It's obvious that there is a lot of talented volunteers who would be willing to work with the project in this regard.

Just my thoughts on all this.

[Edit] As a side note I think any work done on optimized clients should be done in a seperate manner and have whatever credits are gained during the process kept out of the mainstream project so as not to repeat what has gone on here during the process of this latest optimization.
[/Edit]
ID: 5912 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profileveebee

Send message
Joined: 15 Oct 07
Posts: 14
Credit: 150,634,885
RAC: 0
100 million credit badge10 year member badge
Message 5913 - Posted: 4 Nov 2008, 11:46:04 UTC

so, the 108 credit per hour "limit".. does it apply to each core, or each "whole" computer, or each "user" as a whole ?

just want to know so I can set the right amount of wu's without crunching for nothing....
ID: 5913 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
50 million credit badge10 year member badge
Message 5914 - Posted: 4 Nov 2008, 12:08:02 UTC
Last modified: 4 Nov 2008, 12:09:27 UTC

I believe your comment is correct veebee.

So, for a quad - the 4 cores would output 432 credits per hour (108 x 4). Assuming this is consistent, then that rigs RAC would be 432 x 24 = 10,368.

Not a bad output, me thinks?

I find my fastest Quad (a Penryn), might be most effective on a 50:50 share with Einstein. Or perhaps just crunching Einstein alone (one of my other projects), and leaving the rest of my farm (older PCs) to MW alone.
ID: 5914 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileArion
Avatar

Send message
Joined: 10 Aug 08
Posts: 218
Credit: 41,846,854
RAC: 0
30 million credit badge10 year member badge
Message 5915 - Posted: 4 Nov 2008, 12:17:52 UTC - in response to Message 5913.  
Last modified: 4 Nov 2008, 12:19:12 UTC

so, the 108 credit per hour "limit".. does it apply to each core, or each "whole" computer, or each "user" as a whole ?

just want to know so I can set the right amount of wu's without crunching for nothing....



From what I've been able to gather from my 3 systems - 2 dual cores and a so as molasses single core the credit limit applies to each core. At least it does on mine. I share all my systems with another project and my fastest system here does about 12 units per hour and credit awarded is aprox. 108 per hour. The other cores are running different projects so I managed to get a compariosn that way. For what it is worth my wife's single core swaps out with another project and while her grantd credit is higher it still breaks out to .03 credits per CPU second = 108 credits per hour.Personally I woudl rather see a WU taking 1 hour and getting from 50 to 100 credits than trying to allocate so much per second. But that's just my preference based on other projects.

Either way I have had a chance to compare it against my "Dear to Heart Projects: and I'm satisfied with some of the proposals and don't have any further reason to bitch and complain.They want to cut it to 54 credits per hour I'll still be making headway, albeit at a much slower pace.
ID: 5915 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Brian Silvers

Send message
Joined: 21 Aug 08
Posts: 625
Credit: 558,425
RAC: 0
500 thousand credit badge10 year member badge
Message 5916 - Posted: 4 Nov 2008, 12:21:15 UTC

Dave has already said that what I said was correct...

The limit is 108 per hour per core. All that happens is each task completed is adjusted to that ratio. It is all right there in your own stats plain as day. You will not see any zero credit unless you aborted the task, the task was declared redundant, or you had an error. If you complete the task and it validates, the limit is as follows:

granted credit * 3600 / runtime ~= 108

For example, one of my recent tasks

CPU time (sec) claimed credit granted credit
317.80 1.35 9.53

Per the formula above:

9.53 * 3600 / 317.80 = 107.9546 ~= 108

If you have multiple cores, you get that same allotment per core. It's really very simple. You all are overcomplicating things.

As to whether or not the limit is "fair" for fast machines, that's debatable. I'm not getting into that, I'm only stating that the limit is very clearly defined and observable in your own stats...

ID: 5916 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Steven Pletsch

Send message
Joined: 18 Oct 08
Posts: 7
Credit: 676,341
RAC: 0
500 thousand credit badge10 year member badge
Message 5917 - Posted: 4 Nov 2008, 13:22:16 UTC
Last modified: 4 Nov 2008, 13:27:06 UTC

I just wanted to drop in with my 2 cents on this issue

About the optimized application

First off, you can't fault anyone for choosing to run an optimized application, whether they are doing it for the credits, or to produce more work. I think Milksop has done a great service to this project by releasing the application, if it were not for his(or her, i don't know) generosity, this application would not be available in the first place. So many thanks Milksop for your the effort you put into advancing the science of this project. I for one, and I know many other truly appreciate your hard work!

Second, you cannot fault the developers for an inefficient application being run before. Applications such as this one are difficult to write, and for someone that does not fully know all the tricks and the ideal coding methods it can be a daunting task. Anyone who complains about it should sit down, open their favorite text editor and write a better one, or keep their mouth shut. Again many thanks to Milksop for taking the time to write a better one, very few would do the same, whether they knew how or not.

About the credits

All this squabbling about doing the bare minimum to get the most credits is pretty disturbing. You should all be happy to get what you can, and leave it at that. If you are running a stock application, than I do not think that the credits should change, but that's not my decision to make. The optimized application is great, and is significantly increasing the amount of work being done on this project.

Since the application is public, that makes the choice for the project administrators a bit more difficult, but ultimately it is the science which benefits from this. If the administrators choose to implement a per core, per hour limit, it is their project, and their choice. Those who hit that limit, should be grateful that they have the opportunity to contribute so much.

We all joined because of the science in the first place right? I know a lot of people get caught up in their rank, position, credits, etc. But seriously, think back and ask yourself, why are you contributing in the first place? If there were no stats, and no credits, most of us would probably still be here. I like to see credits, I think it's great to see how you compare with the rest of the world. But I don't take it too seriously, I've run projects that grant 50 credits a day, and some that grant 5000 credits a day, on the same machine, same resource share, same everything. Heck I've even run a couple that grant nothing. While cross project comparison is fun, it is not, and has never been fair. I guess the point I'm trying to get to is: "Be grateful for what you get, and don't sweat the details, or people may think your are trying to compensate for your shortcomings elsewhere."

DA will never come out and complain about the project granting 3 credits per hour (Superlink used to), or 1 credit per WU (Current scheme for Ramsey), or 0 credit at all (several alpha and beta projects). Choose whatever projects you want, but don't take things so seriously. I don't know of anyone that can honestly say "I joined Boinc because I wanted to beat everyone else at how many points I can collect."

I'm personally only running this project for a few days, tonight, or tomorrow morning I will detach, and probably not return. It's nothing to do with the credits, I just like to contribute a specific amount of work to different types of projects and then leave to move on to the next one. Whether there is an optimized application or not, I would have done the same, the only difference is that it probably would have taken longer. Either way, whether I did it last month, or this month, the amount of work done for that number of credits would have been exactly the same, perhaps even less with the unoptimized app since I'm hitting the wall on the machines I have attached right now. I don't care about the limit, and will continue to let them crunch until the goal is hit, or the server blows up from getting slammed by 3000 hosts per minute.

Sorry for the rant, but I've seen this discussion now on a dozen different projects, and it's always the same. Everyone is quick to say "I don't care about credits" and follow it up with "but they are getting too much" or "I'm getting screwed by ..."

I'm waiting for the project, that just pays Claimed Credit^5 and refuses to cave in to pressure to change it. It's not far off, and will show just how pointless this whole argument is.

P.S. Perhaps the admins would consider just granting claimed credit until the whole mess is sorted, then I, and everyone else running the optimized app, will get the 1 or so credits per WU we are claiming, and those with the non-optimized app will get whatever their computers think they deserve.
ID: 5917 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profilem4rtyn
Avatar

Send message
Joined: 16 Jan 08
Posts: 18
Credit: 4,111,257
RAC: 0
3 million credit badge10 year member badge
Message 5966 - Posted: 5 Nov 2008, 20:28:34 UTC - in response to Message 5917.  
Last modified: 5 Nov 2008, 20:30:20 UTC

We all joined because of the science in the first place right?


Very wrong!

I don't know of anyone that can honestly say "I joined Boinc because I wanted to beat everyone else at how many points I can collect."


Then maybe your not really looking, I know many.

I can understand and respect the reasons the "science first" types, but I just can't understand why so many of them refuse to give the same respect and understanding to the competative crunchers, after all the work we do is just as helpfull to the project.
m4rtyn
******************************* *******************************

ID: 5966 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileMac-Nic
Avatar

Send message
Joined: 26 Mar 08
Posts: 15
Credit: 2,045,502
RAC: 0
2 million credit badge10 year member badge
Message 5967 - Posted: 5 Nov 2008, 20:56:39 UTC

If one realy only need Credits and don't care about sience I advice to go FreeHAL (or is it FreeFall)

CPU Time 161 sec claimed credits 0.90, granted 100 credits with stock app.

====================================================================================================
330904 95255 5 Nov 2008 10:47:07 UTC 5 Nov 2008 10:48:37 UTC Over Success Done 90.00 0.5045 100.5045

Task ID 368653
Name ConvertDatabases_610215958045_1
Workunit 179153
Created 5 Nov 2008 10:40:47 UTC
Sent 5 Nov 2008 10:45:56 UTC
Received 5 Nov 2008 10:48:37 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 1358
Report deadline 17 Nov 2008 0:32:35 UTC
CPU time 161
stderr out <core_client_version>5.10.45</core_client_version>
<![CDATA[
<stderr_txt>
wrapper: starting
wrapper: running .\runner.exe ()
"my" variable $TEMP_FILE masks earlier declaration in same scope at AI/SemanticNetwork.pm line 722.
"my" variable $sth masks earlier declaration in same scope at AI/SemanticNetwork.pm line 760.
"my" variable $sth masks earlier declaration in same scope at AI/SemanticNetwork.pm line 830.
"my" variable $TEMP_FILE masks earlier declaration in same scope at AI/SemanticNetwork.pm line 851.
"my" variable $sth masks earlier declaration in same scope at AI/SemanticNetwork.pm line 854.

app exit status: 0x1d

</stderr_txt>
]]>

Validate state Valid
Claimed credit 0.902552854818939
Granted credit 100.902552854819
application version 5.24
======================================================================================================

Happy credit harvest ;)
ID: 5967 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Emanuel

Send message
Joined: 18 Nov 07
Posts: 280
Credit: 2,442,757
RAC: 0
2 million credit badge10 year member badge
Message 5976 - Posted: 6 Nov 2008, 0:41:51 UTC - in response to Message 5966.  
Last modified: 6 Nov 2008, 0:46:54 UTC

I can understand and respect the reasons the "science first" types, but I just can't understand why so many of them refuse to give the same respect and understanding to the competative crunchers, after all the work we do is just as helpfull to the project.


The reason is simple: the people who contribute for the science are idealists (in the true sense of the word, not the self-centered one), and idealists get frustrated when other people don't see things the same way. I should know, I am one.

(they also write long posts that have other idealists smiling and nodding their heads in agreement while everyone else either TLDRs or thinks it's unjustifiable nonsense)
ID: 5976 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Annika Kremer

Send message
Joined: 30 Oct 08
Posts: 32
Credit: 60,528
RAC: 0
10 thousand credit badge10 year member badge
Message 5977 - Posted: 6 Nov 2008, 1:01:35 UTC

Sorry, but I haven't come across the term "to TLDR" yet. What does it mean?
ID: 5977 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileLabbie
Avatar

Send message
Joined: 29 Aug 07
Posts: 327
Credit: 116,463,193
RAC: 0
100 million credit badge10 year member badge
Message 5978 - Posted: 6 Nov 2008, 2:16:38 UTC - in response to Message 5977.  

Sorry, but I haven't come across the term "to TLDR" yet. What does it mean?


"Too Long, Didn't Read"

Wikipedia


Calm Chaos Forum...Join Calm Chaos Now
ID: 5978 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Annika Kremer

Send message
Joined: 30 Oct 08
Posts: 32
Credit: 60,528
RAC: 0
10 thousand credit badge10 year member badge
Message 5979 - Posted: 6 Nov 2008, 2:28:12 UTC

Ah, okay. Thanks a lot for the information.
Emanuel: I know too well what you mean. Now try being an idealist among a bunch of mostly childish and completely un-idealistic CS students and you know how I spend my time ;-)
ID: 5979 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Steven Pletsch

Send message
Joined: 18 Oct 08
Posts: 7
Credit: 676,341
RAC: 0
500 thousand credit badge10 year member badge
Message 5983 - Posted: 6 Nov 2008, 14:31:57 UTC - in response to Message 5966.  

We all joined because of the science in the first place right?


Very wrong!

I don't know of anyone that can honestly say "I joined Boinc because I wanted to beat everyone else at how many points I can collect."


Then maybe your not really looking, I know many.

I can understand and respect the reasons the "science first" types, but I just can't understand why so many of them refuse to give the same respect and understanding to the competative crunchers, after all the work we do is just as helpfull to the project.


Fair enough, if this is the motivation, than the system of credits has some value to attract those who wish to compete for credits alone. Though it should be pointed out that the system has never been fair, and probably never will be.

Your effort is appreciated just as much as others, if not more so because of the volume you contribute. Thank you for pointing this out, and I will keep it in mind for future posts. I honestly never considered this being a persons sole motivation, since to me anyways, the credits have never represented a tangible value.

If this is your motivation, I would also suggest investing in some CUDA compatible graphics cards, as some cards can produce easily 10,000 - 20,000 credits/day. They are significantly cheaper than the computers required to obtain that output, and will probably grow to be more widely supported by projects in the coming year.
ID: 5983 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 519
Credit: 284,026,803
RAC: 5,988
200 million credit badge10 year member badgeextraordinary contributions badge
Message 5991 - Posted: 6 Nov 2008, 20:01:26 UTC - in response to Message 5907.  

I suspect Travis has a fair amount of work on his plate -- we might think he exists only to serve us, but I think otherwise.

That being said, what really needs to be done in a new batch of work units is they need to be MUCH longer. Current work units process in the 5 minute area -- so 20 work units per CPU cache is WAY TOO short.


The limit is 0.03 cr/sec. So just look at how long a WU needs on your system, multiply it bei .03 and you know how much credit you'll get for that WU. So no "zero credit scenario", you just get lower credits per WU, the faster you crunch.
So just reactivate all your stone-age computers, even pentium 200 should be fast enough to reach the .03 cr/sec.

Or pressure Travis so he finally does his work and invest the few days that are needed (as shown by the milksop-guys) to bring a decent app to _all_ milkyway crunchers. Then there wouldn't be a need for the optimised app, and they could give fair credit based on their own, fast stock application.

As long as Travis doesn't do his homework, i don't see a solution for milkyway, that'll make everyone happy.


ID: 5991 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · Next

Message boards : Number crunching : And if we DONT run a "FASTER" App.?

©2020 Astroinformatics Group