Maxed (Dec 16 2010)


log in

Advanced search

Message boards : Technical News : Maxed (Dec 16 2010)

Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next
Author Message
Profile Geek@PlayProject donor
Volunteer tester
Avatar
Send message
Joined: 31 Jul 01
Posts: 2467
Credit: 86,143,646
RAC: 1,043
United States
Message 1057912 - Posted: 19 Dec 2010, 20:23:56 UTC
Last modified: 19 Dec 2010, 20:27:49 UTC

I don't understand all this bickering about who should and should not get work and suggesting new plans to implement a new "fairness doctorine".

The only completely fair distribution method is the current one. The client makes a request and then either it is successful for not. If not try again. Any other plan is actually unfair to someone.

If you feel that you are being restricted in some way in getting through to Berkeley servers I would suggest you try a different ISP.
____________
Boinc....Boinc....Boinc....Boinc....

Profile Westsail and *Pyxey*
Volunteer tester
Avatar
Send message
Joined: 26 Jul 99
Posts: 338
Credit: 20,538,216
RAC: 0
United States
Message 1057917 - Posted: 19 Dec 2010, 20:31:17 UTC - in response to Message 1057912.

I don't understand all this bickering about who should and should not get work and suggesting new plans to implement a new "fairness doctorine".

The only completely fair distribution method is the current one. The client makes a request and then either it is successful for not. If not try again. Any other plan is actually unfair to someone.

If you feel that you are being restricted in some way in getting through to Berkeley servers I would suggest you try a different ISP.

+100000
____________
"The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, 'hmm... that's funny...'" -- Isaac Asimov

tbretProject donor
Volunteer tester
Avatar
Send message
Joined: 28 May 99
Posts: 2860
Credit: 215,464,426
RAC: 185,701
United States
Message 1057923 - Posted: 19 Dec 2010, 20:41:17 UTC - in response to Message 1057007.

This isn't the politics board, but I didn't want to let either idea pass without adding a "correction," make of it what you will.

I believe the first quote from Marx was intended as sarcasm.

Not even Marx used that as a slogan.

The "From each..." line is supposed to be the end result of establishing a Socialist utopia where Socialism has created an over-abundence of everything for everyone; not a motto or method for reaching utopia. That's just simply a statement of fact and not an opinion about that fact, one way or the other.

Not even Marx (who was both nuts and a hypocrite - look it up) was so deluded as to think that people would work slavishly to live like their neighbor who doesn't work at all but gets a lot by needing a lot. That's a misuse of that quote and it's done all the time.

Frankly, I can see where work unit distribution would be best for the project if those who produced more, got more. I *think* that was the point of the sarcasm; it makes little sense to feed everyone equally when all are not equally committed.

I don't know how anyone would enforce such a thing, and if it were done I might well find myself with less work (I don't have any big number-crunchers). I can also see the frustration of those who have spent thousands of dollars and hundreds or thousands of hours optimizing their systems, being capable of thru-putting massive amounts of work units then find themselves unable to get the work.

That's a lot more work and commitment than just downloading a screensaver. Shouldn't they get first dibs on the work?

I just wanted to make clear that my opinion isn't self-serving; if there were only enough work to keep the big guns in work units I might find myself without work units. Still, to have all that optimized computing sitting around idle so I can let my P4 1.8MHz plod along for days doing the work someone else wanted and could have completed in a few minutes makes very little sense.

Bret
____________

tbretProject donor
Volunteer tester
Avatar
Send message
Joined: 28 May 99
Posts: 2860
Credit: 215,464,426
RAC: 185,701
United States
Message 1057926 - Posted: 19 Dec 2010, 20:52:37 UTC - in response to Message 1057923.

... and quickly adding so I don't get ambushed:

I don't CARE one way or the other.

Me? I'm STILL content.

Bret
____________

Profile Robert Gammon
Send message
Joined: 29 Aug 01
Posts: 19
Credit: 847,876
RAC: 4,150
United States
Message 1057959 - Posted: 19 Dec 2010, 23:13:39 UTC - in response to Message 1057912.

In large cities, choice of ISP is difficult as there are so many offerings

Not all of us live in large cities. so we may be stuck with either Dialup, Satellite or Packet radio (if offered)

Slightly closer to a town and we may add Cable to the mix, and still closer, DSL or Uverse if offered

The plethora appears to occur when local populations exceed roughly 150,000

The wealth of America is unevenly spread
____________

Aurora Borealis
Volunteer tester
Avatar
Send message
Joined: 14 Jan 01
Posts: 2981
Credit: 5,089,440
RAC: 2,014
Canada
Message 1057960 - Posted: 19 Dec 2010, 23:41:48 UTC

The top couple thousand may produce results faster, but they aren't the backbone of the project. When the project goes for a grant they don't say we have 2000 volunteers that do this much work. The project selling point is that it is supported by 150,000 volunteers. If you don't feed the masses of users then that support will dwindle away even more than it has in the past. The project will be seen as elitist and grants will be even harder to come by. Can those 2000 top crunchers come up with the half million dollars it takes to keep this project running... I don't think so.

As for the ridiculous comment about changing your ISP. This isn't an American project. It's global. Most of the world doesn't have the Internet choices available in North American cities.
____________

Boinc V7.0.27
Win7 i5 3.33G 4GB, GTX470

Profile Geek@PlayProject donor
Volunteer tester
Avatar
Send message
Joined: 31 Jul 01
Posts: 2467
Credit: 86,143,646
RAC: 1,043
United States
Message 1057967 - Posted: 20 Dec 2010, 0:16:46 UTC

I live within 20 miles of Houston. I have only one ISP available to me and the fastest DSL available in my area is 3GB. It never even aproaches that kind of speed, usually half that.

When the cricket graph is maxed out I cannot get through to Berkeley. NOTHING. You may consider my comment ridiculous but if I could change ISP I would. Instead I have to live with it or move my family. And of course I know this is a global project, America has it's own (major) problems with bandwidth in most all rural areas.


____________
Boinc....Boinc....Boinc....Boinc....

Cheopis
Send message
Joined: 17 Sep 00
Posts: 139
Credit: 11,311,233
RAC: 8,073
United States
Message 1058056 - Posted: 20 Dec 2010, 12:09:54 UTC

The only possible "fair" method I can see for distributing work units would be to set teired levels of uploads. This distribution method would be triggered by available work units dropping to zero on the servers.

For example:

1) For whatever reason, the servers have not split data recently, and the available work units dropped to zero.
2) When work units become available again, the work units going to each client would be restricted to one day's worth of work.
3) When work units available to distribute stays greater than zero for 24 hours, the restriction increments by a day, allowing each client to get X+1 days work.

Your RAC would only be important inasmuch as it means you would get more work units assigned because you do more in 24 hours. A high RAC user getting 10 days worth of work slows the whole project down because 50 or 100 low end users are getting no work units at all. Idle machines do no good for SETI.

Eventually, provided that the splitters can catch up and keep work units flowing, in a couple weeks, everyone should have 10 days of work units if that's what they want.

jravin
Send message
Joined: 25 Mar 02
Posts: 941
Credit: 102,471,663
RAC: 88,276
United States
Message 1058059 - Posted: 20 Dec 2010, 12:57:20 UTC
Last modified: 20 Dec 2010, 12:59:02 UTC

Actually, I thought the recent server-based distribution of work (before the Great Outage of 2010) worked fine. At first, set limits of n WU/CPU and m WU/GPU, regardless of machine capability.

This allowed smaller users to get their share, along with some work for heavy crunchers. Then the Team would increase the allowances as b/w permitted, and eventually waive them entirely, so big crunchers could try to fill their caches.

That policy seemed to work reasonably fairly in terms of access to WUs.
____________

Profile Robert Gammon
Send message
Joined: 29 Aug 01
Posts: 19
Credit: 847,876
RAC: 4,150
United States
Message 1058064 - Posted: 20 Dec 2010, 13:21:53 UTC

I like the ideas recently espoused concerning fairness. The question is how to implement. When work unit production/availability is low, limit each responder to 1 WU (regular or Astro). When work unit has been returned to normal, but only for a short while, limit each responder to 1 days worth (regular or AP) As workflow returns to normal, and the servers are normal, increase the number of WU's allowed by one day's worth for each day of uptime on the servers.

And yes absolutely, the reputation of the project is that it allows people from all over the world to participate so long as they have a computer that meets the minimal requirements and an internet connection. 150,000 users who turn in 1 WU per day is a POWERFUL amount of work. 10 users who turn in 1,000 WUs a day are piddling by comparison
____________

Profile John Neale
Volunteer tester
Avatar
Send message
Joined: 16 Mar 00
Posts: 514
Credit: 3,178,729
RAC: 2,053
South Africa
Message 1058094 - Posted: 20 Dec 2010, 14:54:32 UTC - in response to Message 1058064.

150,000 users who turn in 1 WU per day is a POWERFUL amount of work. 10 users who turn in 1,000 WUs a day are piddling by comparison


If that were so, it would be true. But it's not so.

Taking the Scarecrow's data of 19 December, and basing my calculations on active users (those with an RAC > 1):

At the bottom end, there are 102,748 users with an RAC between 1 and 99. They represent 68.4 % of the active users, and contribute 5.8 % towards the total daily RAC.

At the top end, there are 82 users with an RAC > 30,000. They represent 0.0546 % of the active users, and contribute 9.7 % towards the total daily RAC.

Between them, there are 47,424 users with an RAC between 100 and 29,999. They represent 31.6 % of the active users, and contribute 84.5 % towards the total daily RAC.

I conclude that:

Over one hundred thousand little guys contribute ... little.

Less than 100 big guys contribute ... almost twice as much as 100,000 little guys.

Most of the work is done by around 50,000 hobbyists.
____________

Profile James Sotherden
Avatar
Send message
Joined: 16 May 99
Posts: 8918
Credit: 35,923,492
RAC: 43,370
United States
Message 1058107 - Posted: 20 Dec 2010, 16:07:34 UTC

All this rationing talk and fairness for work units, or my computer is old and yours is new is totaly rubbish. For crying out loud, when your computer asks for stinking work you either get it or not. Has nothing to do with what the hell your RAC is or what kind of computer your running.

Before the servers started crapping out and the 3day outage started I ran a .75 day cache and never ran out of work. I have a 3 day cache now and quess what? Out of the 3 machines I run seti on, the i7 usually ran out first. the old p4 second and then the mac. Do I care no,I do other projects when it happens.

Matt has stated that 3 day outages will be back for who knows how long. So if you dont have a 3 day cache at least, Dont be crying you cant get work for your one work unit a day machine.

And that is MHO.


____________

Old James

Profile Ageless
Avatar
Send message
Joined: 9 Jun 99
Posts: 12326
Credit: 2,631,080
RAC: 1,094
Netherlands
Message 1058128 - Posted: 20 Dec 2010, 17:02:44 UTC - in response to Message 1058107.

So if you dont have a 3 day cache at least, Dont be crying you cant get work for your one work unit a day machine.

Getting it actually in is another thing. Downloading the last 8 with blistering speeds of 0.46KB and 1.12KB/sec. Not going to work. Been doing that all day long.

Oh and before anyone blames my ISP, I have 50Mbit down/5Mbit up. Not the problem on my side, mate. :P
____________
Jord

Fighting for the correct use of the apostrophe, together with Weird Al Yankovic

Profile Chris SProject donor
Volunteer tester
Avatar
Send message
Joined: 19 Nov 00
Posts: 32089
Credit: 13,772,324
RAC: 25,508
United Kingdom
Message 1058132 - Posted: 20 Dec 2010, 17:17:58 UTC

Interesting figures! This would seem to more or less bear out the old 80/20 rule, e.g. 80% of the wealth of a country is owned by 20% of the population. Here 84.5% of the work is done by 31.6% of crunchers. Nice to see Seti maintaining the Status Quo!

Best Wishes,

From a Hobbyist :-)

Between them, there are 47,424 users with an RAC between 100 and 29,999. They represent 31.6 % of the active users, and contribute 84.5 % towards the total daily RAC.

I conclude that:

Over one hundred thousand little guys contribute ... little.

Less than 100 big guys contribute ... almost twice as much as 100,000 little guys.

Most of the work is done by around 50,000 hobbyists.

____________
Damsel Rescuer, Uli Devotee, Julie Supporter, ES99 Admirer,
Raccoon Friend, Anniet fan, didn't take pot advice!


bill
Send message
Joined: 16 Jun 99
Posts: 861
Credit: 23,975,287
RAC: 13,839
United States
Message 1058136 - Posted: 20 Dec 2010, 17:29:41 UTC - in response to Message 1058128.

So if you dont have a 3 day cache at least, Dont be crying you cant get work for your one work unit a day machine.

Getting it actually in is another thing. Downloading the last 8 with blistering speeds of 0.46KB and 1.12KB/sec. Not going to work. Been doing that all day long.

Oh and before anyone blames my ISP, I have 50Mbit down/5Mbit up. Not the problem on my side, mate. :P


Fired up, Tornado, which had been turned off the last week, and after installing
the new improved apps (Thank you JG) asked for work. It took 2 minutes and 35 secands to download 21 work units.

Make of that what you will.

BTW, that's a four day cache on Tornado.

Profile Cliff HardingProject donor
Volunteer tester
Avatar
Send message
Joined: 18 Aug 99
Posts: 1005
Credit: 52,991,835
RAC: 42,757
United States
Message 1058186 - Posted: 20 Dec 2010, 21:04:06 UTC

Maybe I have missed the point in my old age, but I hear & read about all of this bickering concerning the fairness of obtaining WUs and whether or not someone has enough cache during the 3-day downtime or for any time period for that matter. What I don't hear or see is anyone suggesting that people increase the size of their cache by increasing the number of days of processing to hold on their machine. One of the things I did many moons ago when I was a SINGLE LITTLE machine, can't remember what it was but it was prior to a Pent II, dial-up, and took several days to do ONE wu, compared to what is out there today, was to increase the 'Maintain enough work for an additional x days' option, which is located in the NETWORK area under 'When and how BOINC uses your computer' in an individual's account page. As far as I know the upper limit is still 10 days.

If one were to increase the number of days under this option, they should have enough to hold them over. I realize that it might take a bit longer, for those of us that do not have the optimal capabilities to d/l the data to your machine(s), but with patience being the operative word, and your machine is connected; you will get the addtional WUs.

Even before I got DSL I never was at a point where I didn't have at least one WU, except when there was an extended outage, and back then extended outages went on for months, not like the small one we recently had.


BTW... HOW BOUT THOSE BIRDS.... GO EAGLES!!!!

Profile platium
Avatar
Send message
Joined: 5 Jul 10
Posts: 212
Credit: 262,426
RAC: 0
United Kingdom
Message 1058201 - Posted: 20 Dec 2010, 22:00:01 UTC

you are all lucky as far as internet speed goe's here in the uk we have the British telecom it siad i get upto 10 mb but it best is 1.8 m.b and its drops down to 500 bytes sometimes

WinterKnight
Volunteer tester
Send message
Joined: 18 May 99
Posts: 8685
Credit: 24,996,467
RAC: 29,346
United Kingdom
Message 1058318 - Posted: 21 Dec 2010, 4:11:29 UTC - in response to Message 1058201.

you are all lucky as far as internet speed goe's here in the uk we have the British telecom it siad i get upto 10 mb but it best is 1.8 m.b and its drops down to 500 bytes sometimes

If you can, and I know their cable doesn't reach all parts of the UK, go to Virgin. I'm on 20Mb and that's what I get, testing to UK servers. Last time I tested to SF got >18Mb.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next

Message boards : Technical News : Maxed (Dec 16 2010)

Copyright © 2014 University of California