How to select projects - more points/day?

A forum just for SPCR's folding team... by request.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

How to select projects - more points/day?

Post by dukla2000 » Fri Jul 11, 2003 3:38 am

It occurs to me it should be possible to select the projects you are assigned (beyond -advmethods to get GROMACS).

For example, on my Athlon XP systems the current project 910 gives around twice the daily point count as project 340 stuff. And at this Stanford pagethere is a list of which servers assign which projects. So if I could persuade my boxen to go to 171.64.122.144 for work I have a reasonable chance of doubling my daily output.

I read in a Stanford folding thread (or maybe misinterpreted) that some guy was selecting his servers - is it perhaps as simple as putting a line in the hosts file? Except it also seems to me the core.exe has a hardcoded list that it polls - maybe we need to do a hex patch?
[size=75]2009/Oct: Jetway JNC81-LF * 4850e naked under fanless Xigmatek Apache * Antec mini Skeleton w/Nexus 120mm PWM fan * Delta 90W brick w/Skeleton DC-DC board * WD2500BEVT 250Gb blue[/size]

WarpedPlatter
Friend of SPCR
Posts: 37
Joined: Tue Jun 10, 2003 12:40 pm
Location: Northern Virginia, USA

Post by WarpedPlatter » Fri Jul 11, 2003 6:59 am

If the client can't contact one IP address, will it try another after a certain amount of time? If so, all that would be needed is to block access to the IP addresses of the servers with less desirable projects.

I'll try to fire up a packet sniffer when future WUs are transmitted and received to see what I can figure out.

randalee
Posts: 34
Joined: Fri Jul 11, 2003 9:57 am
Location: Salt Lake City, UT
Contact:

Post by randalee » Fri Jul 11, 2003 10:01 am

Easiest way to block access to a site is add an entry in the HOSTS file on your PC pointing to 127.0.0.1 (this is the address reserved for localhost).

So if I wanted to block yahoo.com I'd put an entry in HOSTS:

127.0.0.1 yahoo.com

However, I don't know if the Folding client is querying by a name or directly by IP address. If they're going direct for a specific IP address, then my HOSTS method wouldn't work.

Randy Clements
Salt Lake City, UT

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Fri Jul 11, 2003 10:46 am

WarpedPlatter wrote:If the client can't contact one IP address, will it try another after a certain amount of time?
Yup - at the end is an extract from a log where 1 of my boxen spent 2 hours trying to get more work, most of the time trying to access 171.64.122.110 until it for some reason tried 171.64.122.119.

The problem with this 'unnatural selection' is if the server of 'generous' units is having problems, we could sit without work for a while. But I am still curious to try.

[19:01:01] + Attempting to get work packet
[19:01:01] - Connecting to assignment server
[19:01:02] - Successful: assigned to (171.64.122.110).
[19:01:02] + News From Folding@Home: Welcome to Folding@Home
[19:01:02] Loaded queue successfully.
[19:01:03] + Could not connect to Work Server
[19:01:03] - Error: Getwork #9 failed, and no other work to do. Waiting before retry
[19:22:29] + Attempting to get work packet
[19:22:29] - Connecting to assignment server
[19:22:30] - Successful: assigned to (171.64.122.119).
[19:22:30] + News From Folding@Home: Welcome to Folding@Home
[19:22:30] Loaded queue successfully.
[19:22:31] - Deadline time not received.
[19:22:35] + Closed connections
[19:22:35]
[19:22:35] + Processing work unit
[size=75]2009/Oct: Jetway JNC81-LF * 4850e naked under fanless Xigmatek Apache * Antec mini Skeleton w/Nexus 120mm PWM fan * Delta 90W brick w/Skeleton DC-DC board * WD2500BEVT 250Gb blue[/size]

Wrah
Patron of SPCR
Posts: 316
Joined: Thu Apr 10, 2003 1:56 am

Post by Wrah » Mon Jul 14, 2003 2:18 am

I think we have to continue this thread on our secret forum..

http://kwsnforum.com/viewtopic.php?t=2030
[size=84]XP 2800@2270mhz+SLK800A+Nexus fan, Aopen 8RDA+@414mhz, Samsung 80gb on foam, XFX6800GT with a Nexus 120mm fan taped onto it, Nexus 4090 PSU, Chieftec Dragon+Nexus fan@1200rpm[/size]

Fart in your gen direxion
Posts: 59
Joined: Sun Jun 01, 2003 2:35 pm
Location: Regrettably for you, I'm Upwind in Upstate N.Y. U.S.A.
Contact:

Post by Fart in your gen direxion » Mon Jul 14, 2003 7:46 am

I thought it was all for science ? Guess not, huh ? :roll: Bad dog, no humping allowed for you. :P

Zhentar
Patron of SPCR
Posts: 918
Joined: Thu Mar 27, 2003 4:00 pm
Location: Madison, WI, USA

Post by Zhentar » Mon Jul 14, 2003 10:34 am

science? SCIENCE? what kind of newbie are you! This isn't about science, its about getting more points! that "science" crap is just for justifying buying more parts.

Rusty075
SPCR Reviewer
Posts: 4000
Joined: Sun Aug 11, 2002 3:26 pm
Location: Phoenix, AZ
Contact:

Post by Rusty075 » Mon Jul 14, 2003 3:56 pm

Bad idea guys, for several reasons:

1. The work units aren't server dependant. Just because you got a 910 from 1 server, doesn't mean it won'r give you a Tinker next time.

2. The units are given out based on Stanfords need. If they don't need 910's, you won't get them, from any server.

3. The whole reason Stanford has multiple servers is to lessen the strain on any individual one, and to increase reliability. Think what would happen if 75% of people kept hammering one server.

4. This will probably result in lower folding scores for you, not higher. We all know how often the servers go down, you machine may spend 12 hours hammering a down server, when it could be folding.

5. And lastly, trying to rig to system to your benefit, rather the Stanfords, flies in the face of the whole point of doing this. It is a charity, after all.

If I told you how to hack the core to have it send back units 10 times as fast, even though the units would be useless to Stanford, would you do it?
[size=75][b]Senior Contributing Writer, SPCR[/b][/size]

Zhentar
Patron of SPCR
Posts: 918
Joined: Thu Mar 27, 2003 4:00 pm
Location: Madison, WI, USA

Post by Zhentar » Mon Jul 14, 2003 4:58 pm

well, I wouldn't say taking the higher point WUs and leaving lower point ones to "anonymous" users would undermine the point, but I will agree that the other 4 are very true.

And the last bit would be very wrong, undermining an effort to get credit indication that you are highly benefitting it....

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Tue Jul 15, 2003 3:49 am

Rusty075 wrote:5. And lastly, trying to rig to system to your benefit, rather the Stanfords, flies in the face of the whole point of doing this. It is a charity, after all.

If I told you how to hack the core to have it send back units 10 times as fast, even though the units would be useless to Stanford, would you do it?
There is no intent implied to return results that are incomplete/inaccurate whatever. Merely (as per Zhentar) to do work Stanford want, properly, but that improves scoring.
Rusty075 wrote:1. The work units aren't server dependant. Just because you got a 910 from 1 server, doesn't mean it won'r give you a Tinker next time.
Disagree - my understanding is that projects are served by 1 and only 1 server. However if you do stick to 1 server, you still are at the control of Stanford which WUs that server is issuing at the instant you request.

But your general tone/thoughts are taken to heart - in particular the risk of point 4 means that losing folding time because of some weird Holy Grail would indeed be a very poor result.

ps - anything I am interested in will NOT breach the Stanford license.
[size=75]2009/Oct: Jetway JNC81-LF * 4850e naked under fanless Xigmatek Apache * Antec mini Skeleton w/Nexus 120mm PWM fan * Delta 90W brick w/Skeleton DC-DC board * WD2500BEVT 250Gb blue[/size]

tragus
*Lifetime Patron*
Posts: 356
Joined: Fri Apr 18, 2003 11:19 am
Location: Baltimore, MD

Post by tragus » Tue Jul 15, 2003 6:02 am

dukla2000 wrote:n particular the risk of point 4 means that losing folding time because of some weird Holy Grail would indeed be a very poor result. [[emphasis added]]


Methinks that the Knights Who Say Ni might have observations on this point.
This space for rent. Will oppress peasants for food.

wussboy
Posts: 635
Joined: Wed Oct 23, 2002 12:34 pm
Location: Southampton, UK

Post by wussboy » Tue Jul 15, 2003 8:53 pm

And me also thinks said observations would generally refer to farting.
Everything is exactly what it seems.

riffst3r
Posts: 221
Joined: Tue Apr 08, 2003 8:43 am
Contact:

Post by riffst3r » Tue Jul 15, 2003 11:19 pm

me protests, why not burping?
[url=http://forums.silentpcreview.com/viewtopic.php?p=17121#17121][img]http://groups.msn.com/_Secure/0PADVAtURxRVSQVlAFs285ZenYORuaaSACZNaNxzPz*JXEWqAkkQLwQ1AaRTlwU0VP!abPbjXK1f0jKipyX4LYXM0Wxar!Hyz/sib.jpg?dc=4675427731303618688[/img][/url]

http://www.riffst3r.com

Fart in your gen direxion
Posts: 59
Joined: Sun Jun 01, 2003 2:35 pm
Location: Regrettably for you, I'm Upwind in Upstate N.Y. U.S.A.
Contact:

Post by Fart in your gen direxion » Wed Jul 16, 2003 8:17 pm

tragus wrote:
dukla2000 wrote:n particular the risk of point 4 means that losing folding time because of some weird Holy Grail would indeed be a very poor result. [[emphasis added]]


Methinks that the Knights Who Say Ni might have observations on this point.


We seek the Holy Grail, if you have it you'd better hand it over or we'll dispatch the horde on you :x . And stop trying to "work" the system you dawgs, chew on whatever bones Stanford gives you :P .

wussboy
Posts: 635
Joined: Wed Oct 23, 2002 12:34 pm
Location: Southampton, UK

Post by wussboy » Wed Jul 16, 2003 9:28 pm

Whoa whoa whoa. Who said anything about "work"?
Everything is exactly what it seems.

Post Reply