2 F@H clients on 1 machine

This forum is for discussions about the Motherboards.org Folding team. What is folding? Venture on in for a look.

Moderator: The Mod Squad

Postby Tulatin » Tue Sep 23, 2003 5:48 pm

you were just screwing with me weren't you Josh? As to where you left off, are you going to just leave your legacy? A dusty work unit count idly sitting by as Aussie's clims far above and beyond? As the might of your machines is forgotten as even partial users (like me) pass your work unit count... surely this isn't a fitting end.
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby joshvee » Tue Sep 23, 2003 5:57 pm

hehe nicely put....but yup thats what i'm gonna do....
joshvee
Black Belt 4th Degree
Black Belt 4th Degree
 
Posts: 4411
Joined: Sat Jul 12, 2003 11:13 pm
Location: Australia

Postby Tulatin » Tue Sep 23, 2003 5:58 pm

ah, here it is - page 9 -

Thanks. At the rate those two machines will fold , we should be seeing results posted today. I for one will be watching with interest. I remember Tulatin mentioning running an AMD test vs his 2.0 P4 and was wondering what he had in mind as being a fair challenge. Thought I might slap something together out of the scrape parts bin

Fair Challenge? 2600+, but it would be short term... about 8 hours at most, so that's shot.
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby Tulatin » Tue Sep 23, 2003 5:58 pm

You'll still sponsor that badge contest right, and what will it take to lure you back?
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby joshvee » Tue Sep 23, 2003 6:05 pm

Tulatin wrote:You'll still sponsor that badge contest right, and what will it take to lure you back?


heh what contest? smurf-pimp is the only contestant....pretty sad outcome. and he's missing in action.

Naa this folding forum is to small for Aussie and me and seeing he's a mod and all, I guess I'll just bow out and float away with the mist in the night 8) with my crappy Intels in tow. :D
Last edited by joshvee on Thu Sep 25, 2003 4:05 am, edited 2 times in total.
joshvee
Black Belt 4th Degree
Black Belt 4th Degree
 
Posts: 4411
Joined: Sat Jul 12, 2003 11:13 pm
Location: Australia

Postby Tulatin » Tue Sep 23, 2003 6:08 pm

aw come on dude. I can rig up another contestant, but two is sort of a sad sack showing. If we get it in the news, it'll take...
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby evasive » Tue Sep 23, 2003 6:14 pm

If I had serious power here, but my cellies would get blown out of the water either way (1.8 and 2.0)...
We hate rut, but we fear change.
********************************
System error, strike any user to continue...
evasive
Mobo-fu Master
Mobo-fu Master
 
Posts: 37389
Joined: Sun May 06, 2001 12:01 am
Location: Netherlands

Postby Tulatin » Tue Sep 23, 2003 6:15 pm

Time for creative clocking evasive (2.0 and 2.2 respectively)
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby Aussie » Wed Sep 24, 2003 10:45 am

I present the following information for the benefit of fellow forum users and especially those that were following the joshvee - Aussie challenge, you can draw your own conclusions from it.

Firstly I would like to point out that joshvee scored 199 points on the team points page in aproximately 24 hours:-

http://folding.stanford.edu/cgi-bin/teampage?q=33258

joshvee_challenge 199 4



Secondly I would like to point out joshvee's post where he stated his claimed output in 27 hours:-


http://www.motherboards.org/forums/viewtopic.php?t=31568&postdays=0&postorder=asc&start=130&sid=0fef2ffd0606ca25a18154c2ef786943

= 257 all after 27 hours with 2 virtual CPUs registered in the stats.



Thirdly I would like to point you to the site below where Folding@home performance of a P4 3.06, with and without hyperthreading, is compared against an Athlon 2.8. As you will see when running a single client the P4 3.06 is marginally faster than the XP2800+ and when running two clients the P4 gained a 50% performance increase as stated on the conclusion page (3rd page).

http://www.sudhian.com/showdocs.cfm?aid=280&pid=1032



Next I would like to show you the following extracts from the actual FAHlog files of a P4 3.06 HT running two instances of the folding client. These logs were kindly sent to me by a member of another forum.


Arguments: -local -config -advmethods -forceasm

[11:35:18] - Ask before connecting: No
[11:35:18] - Use IE connection settings: Yes
[11:35:18] - User name: XXXXXXXX (Team XXXXX)
[11:35:18] - User ID = XXXXXXXXXXXXXXXX
[11:35:18] - Machine ID: 1



[22:54:46] + Processing work unit
[22:54:46] Core required: FahCore_78.exe
[22:54:46] Core found.
[22:54:46] Working on Unit 04 [September 19 22:54:46]
[22:54:46] + Working ...
[22:54:46]
[22:54:46] *------------------------------*
[22:54:46] Folding@home Gromacs Core
[22:54:46] Version 1.49 (June 23, 2003)
[22:54:46]
[22:54:46] Preparing to commence simulation
[22:54:46] - Assembly optimizations manually forced on.
[22:54:46] - Not checking prior termination.
[22:54:47] - Expanded 361726 -> 1757153 (decompressed 485.7 percent)
[22:54:47] - Starting from initial work packet
[22:54:47]
[22:54:47] Project: 377 (Run 19, Clone 49, Gen 1)
[22:54:47]
[22:54:47] Assembly optimizations on if available.
[22:54:47] Entering M.D.
[22:54:53] Protein: p377_gnra_pf2_22i
[22:54:53]
[22:54:53] Writing local files
[22:55:29] Extra SSE boost OK.
[22:55:30] Writing local files
[22:55:30] Completed 0 out of 500000 steps (0)

[03:20:09] Completed 100000 out of 500000 steps (20)

[07:45:23] Completed 200000 out of 500000 steps (40)

[12:10:01] Completed 300000 out of 500000 steps (60)

[16:34:51] Completed 400000 out of 500000 steps (80)

[21:00:46] Completed 500000 out of 500000 steps (100)
[21:00:50] Writing final coordinates.
[21:00:50] Past main M.D. loop
[21:01:50]
[21:01:50] Finished Work Unit:
[21:01:50] - Reading up to 289968 from "work/wudata_04.arc": Read 289968
[21:01:50] - Reading up to 471072 from "work/wudata_04.xtc": Read 471072
[21:01:50] goefile size: 0
[21:01:50] Leaving Run
[21:01:51] - Writing 770496 bytes of core data to disk...
[21:01:51] ... Done.
[21:01:51] - Shutting down core
[21:01:51]
[21:01:51] Folding@home Core Shutdown: FINISHED_UNIT
[21:01:53] CoreStatus = 64 (100)
[21:01:53] Sending work to server

Start time 22:54, Finish time 21:01 = 22 hours 6 minutes 53.3 points (24.9 min per point)


[23:10:56] + Processing work unit
[23:10:56] Core required: FahCore_78.exe
[23:10:56] Core found.
[23:10:56] Working on Unit 05 [September 20 23:10:56]
[23:10:56] + Working ...
[23:10:56]
[23:10:56] *------------------------------*
[23:10:56] Folding@home Gromacs Core
[23:10:56] Version 1.49 (June 23, 2003)
[23:10:56]
[23:10:56] Preparing to commence simulation
[23:10:56] - Assembly optimizations manually forced on.
[23:10:56] - Not checking prior termination.
[23:10:57] - Expanded 591255 -> 3025757 (decompressed 511.7 percent)
[23:10:57] - Starting from initial work packet
[23:10:57]
[23:10:57] Project: 364 (Run 11, Clone 31, Gen 31)
[23:10:57]
[23:10:57] Assembly optimizations on if available.
[23:10:57] Entering M.D.
[23:11:03] Protein: p364_tip5p_pf1_Mg
[23:11:03]
[23:11:03] Writing local files
[23:11:39] Extra SSE boost OK.
[23:11:40] Writing local files
[23:11:40] Completed 0 out of 100000 steps (0)

[01:50:06] Completed 20000 out of 100000 steps (20)

[04:42:13] Completed 40000 out of 100000 steps (40)

[07:22:16] Completed 60000 out of 100000 steps (60)

[10:01:59] Completed 80000 out of 100000 steps (80)

[12:40:17] Completed 100000 out of 100000 steps (100)
[12:40:21] Writing final coordinates.
[12:40:22] Past main M.D. loop
[12:41:22]
[12:41:22] Finished Work Unit:
[12:41:22] - Reading up to 470280 from "work/wudata_05.arc": Read 470280
[12:41:22] - Reading up to 204896 from "work/wudata_05.xtc": Read 204896
[12:41:22] goefile size: 0
[12:41:22] Leaving Run
[12:41:27] - Writing 678616 bytes of core data to disk...
[12:41:27] ... Done.
[12:41:27] - Shutting down core
[12:41:27]
[12:41:27] Folding@home Core Shutdown: FINISHED_UNIT
[12:41:30] CoreStatus = 64 (100)
[12:41:30] Sending work to server

Start time 23:10, Finish time 12:41 = 13 hours 31 minutes 29.4 points (27.6 min per point)


Arguments: -local -forceasm -advmethods -verbosity 9

[11:32:08] - Ask before connecting: No
[11:32:08] - Use IE connection settings: Yes
[11:32:08] - User name: XXXXXXXX (Team XXXXX)
[11:32:08] - User ID = XXXXXXXXXXXXXXXX
[11:32:08] - Machine ID: 2


[22:53:55] + Processing work unit
[22:53:55] Core required: FahCore_78.exe
[22:53:55] Core found.
[22:53:55] Working on Unit 08 [September 19 22:53:55]
[22:53:55] + Working ...
[22:53:56]
[22:53:56] *------------------------------*
[22:53:56] Folding@home Gromacs Core
[22:53:56] Version 1.49 (June 23, 2003)
[22:53:56]
[22:53:56] Preparing to commence simulation
[22:53:56] - Assembly optimizations manually forced on.
[22:53:56] - Not checking prior termination.
[22:53:56] - Expanded 304957 -> 1767785 (decompressed 579.6 percent)
[22:53:56] - Starting from initial work packet
[22:53:56]
[22:53:56] Project: 344 (Run 33, Clone 84, Gen 0)
[22:53:56]
[22:53:56] Assembly optimizations on if available.
[22:53:56] Entering M.D.
[22:54:02] Protein: p344_gnra_unf_1
[22:54:02]
[22:54:02] Writing local files
[22:54:38] Extra SSE boost OK.
[22:54:38] Writing local files
[22:54:38] Completed 0 out of 500000 steps (0)

[03:19:13] Completed 100000 out of 500000 steps (20)

[07:44:33] Completed 200000 out of 500000 steps (40)

[12:09:21] Completed 300000 out of 500000 steps (60)

[16:34:21] Completed 400000 out of 500000 steps (80)

[21:00:29] Completed 500000 out of 500000 steps (100)
[21:00:33] Writing final coordinates.
[21:00:34] Past main M.D. loop
[21:01:34]
[21:01:34] Finished Work Unit:
[21:01:34] - Reading up to 291960 from "work/wudata_08.arc": Read 291960
[21:01:34] - Reading up to 475496 from "work/wudata_08.xtc": Read 475496
[21:01:34] goefile size: 0
[21:01:34] Leaving Run
[21:01:34] - Writing 776912 bytes of core data to disk...
[21:01:34] ... Done.
[21:01:35] - Shutting down core
[21:01:35]
[21:01:35] Folding@home Core Shutdown: FINISHED_UNIT
[21:01:37] CoreStatus = 64 (100)
[21:01:37] Sending work to server

Start time 22:53, Finish time 21.01 = 22 hours 8 minutes 53.3 points (24.9 mins per point)



[23:10:59] + Processing work unit
[23:10:59] Core required: FahCore_78.exe
[23:10:59] Core found.
[23:10:59] Working on Unit 09 [September 20 23:10:59]
[23:10:59] + Working ...
[23:11:00]
[23:11:00] *------------------------------*
[23:11:00] Folding@home Gromacs Core
[23:11:00] Version 1.49 (June 23, 2003)
[23:11:00]
[23:11:00] Preparing to commence simulation
[23:11:00] - Assembly optimizations manually forced on.
[23:11:00] - Not checking prior termination.
[23:11:00] - Expanded 589496 -> 3025017 (decompressed 513.1 percent)
[23:11:00] - Starting from initial work packet
[23:11:00]
[23:11:00] Project: 364 (Run 13, Clone 31, Gen 31)
[23:11:00]
[23:11:00] Assembly optimizations on if available.
[23:11:00] Entering M.D.
[23:11:07] Protein: p364_tip5p_pf1_Mg
[23:11:07]
[23:11:07] Writing local files
[23:11:43] Extra SSE boost OK.
[23:11:43] Writing local files
[23:11:43] Completed 0 out of 100000 steps (0)

[01:49:38] Completed 20000 out of 100000 steps (20)

[04:41:04] Completed 40000 out of 100000 steps (40)

[07:20:21] Completed 60000 out of 100000 steps (60)

[09:59:23] Completed 80000 out of 100000 steps (80)

[12:38:26] Completed 100000 out of 100000 steps (100)
[12:38:30] Writing final coordinates.
[12:38:30] Past main M.D. loop
[12:39:30]
[12:39:30] Finished Work Unit:
[12:39:30] - Reading up to 470160 from "work/wudata_09.arc": Read 470160
[12:39:30] - Reading up to 203868 from "work/wudata_09.xtc": Read 203868
[12:39:30] goefile size: 0
[12:39:30] Leaving Run
[12:39:31] - Writing 677468 bytes of core data to disk...
[12:39:32] ... Done.
[12:39:32] - Shutting down core
[12:39:32]
[12:39:32] Folding@home Core Shutdown: FINISHED_UNIT
[12:39:42] CoreStatus = 64 (100)
[12:39:42] Sending work to server

Start time 23:10, Finish time 12:39 = 13 hours 29 minutes 29.4 points (27.5 mins per point)


The bolded text was added by myself.


As you can see each CPU generates a point in 25 to 27 minutes approximately, at a point every 25 minutes per CPU that totals to 115 points per day. Allowing for variances in WU's and the possibility of tweaks improving on this let's for example say that each CPU can generate a point every 20 minutes, that totals to 144 points per day so let's round that up to 150 points per day to allow a margin for error.

This would seem to indicate that the best we could expect from a P4 3.06 with hyperthreading enabled and running two instances of the client would be 150 points per day or 6.25 points per hour or 169 points in 27 hours.

I would imagine that a sizeable overclock of say around 30% to 50% up to say 4 to 4.6GHz would improve those figures by roughly the same 30% to 50% so that would give us around 195 to 225 points per day or 220 to 254 points in 27 hours.


I will not comment on this matter any further, the challenge is a dud, all communication with joshvee will now cease.

Draw your own conclusions.



Aussie
Our lives begin to end the day we become silent about things that matter - Martin Luther King
Aussie
Mobo-fu Master
Mobo-fu Master
 
Posts: 10244
Joined: Mon Nov 22, 1999 1:01 am
Location: Mackay, Queensland, Australia

Postby joshvee » Wed Sep 24, 2003 1:18 pm

Well you’ve certainly drawn yours haven't u Aussie.
hehe I told u aussie would write an essay :D ahhh so predictable....says more about Aussie than it does me...haha.
There are a 100 and 1 things that he fails to see in his blind effort to discredit me. All those tests were based on 2 WUs to compare 2 machines....and as we all know different WUs perform very differently even If u take the high point WUs one 50+ pointer can take 11mins per frame and another take 4+ mins per frame depending on the molecule. those tests he linked to were not to show what to expect as far as time goes for all WUs but simply to show the difference between 2 machines doing the SAME WU.
And by the way OCing is not the only way to speed up ur WUs. Like I said I was running 4 WUs for that one day and it did only show up as 2 CPUs from the 1 machine. (if u all want I'll show u how to do it.) I never denied I was doing that for the first day. There was a good advantage to running 4 clients and that is mostly why I did 199 points in 24 hours or so. (there were other tweeks involved)

Personally I don't see any thing wrong with running multiple clients in a challenge (or any other time for that matter) if your computer can do it because we are testing the computers ability to fold as much as it can in a given time. if the computer is able to do it then by all means it should be done. Because we are testing the comps ability to fold at its best.....to do otherwise and not take advantage of the comps full potential is not testing the comp to its fullest. :D

Basically Aussie seems to be trying to point out that I was OCing to 4+gig which I was not.....Like I said there are other ways to improve ur folding on a hyper comp and running 4 clients is only 1 of them.

Anyway I think Aussie is being a very bad sport about all this and is trying to dig up dirt where there is none....that’s pretty low
Last edited by joshvee on Wed Sep 24, 2003 4:23 pm, edited 1 time in total.
joshvee
Black Belt 4th Degree
Black Belt 4th Degree
 
Posts: 4411
Joined: Sat Jul 12, 2003 11:13 pm
Location: Australia

PreviousNext

Return to Motherboards.org Folding Team

Who is online

Users browsing this forum: No registered users and 1 guest

cron