Whoa! Big Point WUs

gustav · Jan 11, 2005

I new there were big packet work units, and I have it set to get those if available, but I've gotten them before and they used up a bunch of memory, that was about it. I just turned in a WU thats been going at it on my machine in sig for about a week. 2-3 hours per frame, depending on what else I'm doing. Well I guess its worth the wait, 600 points in one WU is niiiiiiiiiiiiiice! :clap:

Guess the days of doing 50-100 WUs per month mainly on this machine (have another folding machine, but only 1.6ghz P4) are gone. I remember those days *sniff*

nikhsub1 · Jan 11, 2005

/me wonders where gustav has been? Old news my friend

Arkaine23 · Jan 11, 2005

2 instances will get you even more PPW if you're not using two instances already.... The 600 pointers seem to use between 60-105MB ram each for the folding core.

gustav · Jan 11, 2005

Well I don't stay updated much...I just let the consoles do their thing and check the forum once in a while.

Way ahead of ya Arkaine, been running 2 instances (on this box) almost as long as I've been folding.

Sentential · Jan 11, 2005

Yea I had a couple which were huge...sadly those days are gone. Im stick on Tinkers right now :-/

gustav · Jan 11, 2005

Sentential said:
Yea I had a couple which were huge...sadly those days are gone. Im stick on Tinkers right now

Yeah I used to set flags and what not, used to run EMIII but then it wouldn't work with the new console so I dumped it. Got tired of screwing with it, so I just let them do their thing, do any WU I get, don't care much...just keep on folding is all.

Sentential · Jan 11, 2005

gustav said:
Yeah I used to set flags and what not, used to run EMIII but then it wouldn't work with the new console so I dumped it. Got tired of screwing with it, so I just let them do their thing, do any WU I get, don't care much...just keep on folding is all.

Yea same, plus I have several boxes folding so its a non issue

Leviathan41 · Jan 11, 2005

My P4 has been folding dual 600 pointers for the past few WUs. It will turn in one and grab another, w00t.

stan03 · Jan 11, 2005

yea ive gotten only 600pt wus for a while too

Arkaine23 · Jan 11, 2005

-advmethods + configuring client.cfg to accept BP will get a steady diet of them with the 5.02 and newer clients

They're pretty much worth 2x the PPW of what their benchmarked point value should be and they're gromacs so non-SSE cpu's need not bother.

Gasoline · Jan 11, 2005

I just turned on bigpackets on 3 of my dual Xeon 3.2Ghz temporary borgs and the one single 3.2Ghz Xeon box and it took a few days (from Friday until Monday) to see an impact, but today a bunch of big point WUs finally got turned this afternoon, giving me my alltime highest daily production spike ever since I began folding.

I just added on 6 more dual 3.2GHz Xeon servers tonight, each running 4 instances of the FAH console 5.02, since I can only fold on them for maybe a week until I lose them, and since bigpackets take so long to finish, I just left those 24 new instances running non-bigpackets in hopes of squeezing some decent points out of them before the machines go on to their permanent homes.

gustav · Jan 11, 2005

I've got bigpackets=yes in my cfg, so I assume its on. I don't use flags but this is the first time I've had a 600 point WU. Like I said I've had WUs that took up tons of ram and took forever, but thats was when I first put on big packets when I got the new console. If I remember right they used 150 megs for each core and took several days to complete.

thats a lot of power gasoline, should add a nice chunk of pointage.

Arkaine23 · Jan 11, 2005

Without -advmethods, you'll only get them once in awhile if your CPU is sufficiently fast at the benchmark and turning in work units. The performance quotient of the particular client has something to do with whether or not you get them and how often, if you are not using -advmethods but have BP enabled in the client.cfg.

Sure would help the team production if you used -advmethods....

Leviathan41 · Jan 11, 2005

Man, gasoline has already put up a cool 5600 points today with one more update to go! Nice work :thup:

gustav · Jan 11, 2005

Arkaine23 said:
Sure would help the team production if you used -advmethods....

Done.

Arkaine23 · Jan 11, 2005

Never has a single word so perfectly captured my feelings about something as this:

woot!

KrisMCool · Jan 11, 2005

gustav said:
I just turned in a WU thats been going at it on my machine in sig for about a week. 2-3 hours per frame, depending on what else I'm doing.

This isn't meant to bust your chops, but rather to point out a possible problem with your setup. Your sig rig says P4E @ 3.75. I'm running 2 P4Cs @ ~ 3.75, 2 instances each, running 600 pt BP WUs since they were available. My frame times are ~ 28:30 with one instance and ~ 46:00 with two. If its REALLY taking 2 - 3 hours per frame on said rig, assembly optimizations (SSE) may have been disabled in your configuration. Maybe from a previous crash or nonpolite program exit. Just to be sure, you may want to add the -forceasm flag to see if your frame times dramatically improve. Just a thought.

BTW, I've found that P4Es fold single WUs a fair amount slower than a P4C at same clocks, but fold 2 WUs just a bit faster. So, as I'm sure you already know, 2 instances is definitely the way to go on any P4, but especially P4Es.

Gasoline · Jan 11, 2005

KrisMCool said:
2 instances is definitely the way to go on any P4, but especially P4Es.

Even on the lowly, non-hyperthreading original classic 400fsb Northwoods?

Arkaine23 · Jan 11, 2005

No, without HT there's not much advantage to it, other than one instance is going while the other is uploading or downloading a WU.

gustav · Jan 11, 2005

well here is a clip from the last one I did..dunno if it will help. They dont always get 100% of the cpu, I run bit torrent which sucks up anywhere from 1-10% cpu usage as well as all my other programs, aim, firefox, ventrilo, antivirus, firewall, microsoft anti-spyware, all seeing eye, key counter, aida32 basicly all programs I run normally, as well as two medal of honor allied assault game servers which take little to nothing at idle but use more as people play. But for the most part they get most of the cpu time. Maybe you had a different 600 point wu that went quicker, I've seen several on their site. But if there is a problem you can detect in the log, point it out. I know when I go to get a new WU I get FILE_IO_ERROR a lot. Some people say my overclock is unstable but it primes for several days without error, memtest runs without errors as well.

Code:

--- Opening Log file [January 9 13:17:30] 


# Windows Console Edition #####################################################
###############################################################################

                       Folding@Home Client Version 5.02

                          [url]http://folding.stanford.edu[/url]

###############################################################################
###############################################################################

Launch directory: C:\Program Files\Folding\FAH 2
Executable: C:\Program Files\Folding\FAH 2\FAH502-Console-2.exe


[13:17:30] - Ask before connecting: No
[13:17:30] - Use IE connection settings: Yes
[13:17:30] - User name: OCgustav (Team 32)
[13:17:30] - User ID: 4CC31C3315CD05CE
[13:17:30] - Machine ID: 2
[13:17:30] 
[13:17:30] Loaded queue successfully.
[13:17:30] + Benchmarking ...
[13:17:33] 
[13:17:33] + Processing work unit
[13:17:33] Core required: FahCore_78.exe
[13:17:33] Core found.
[13:17:33] Working on Unit 08 [January 9 13:17:33]
[13:17:33] + Working ...
[13:17:39] 
[13:17:39] *------------------------------*
[13:17:39] Folding@Home Gromacs Core
[13:17:39] Version 1.70 (October 24, 2004)
[13:17:39] 
[13:17:39] Preparing to commence simulation
[13:17:39] - Ensuring status. Please wait.
[13:17:57] - Looking at optimizations...
[13:17:57] - Working with standard loops on this execution.
[13:17:57] - Previous termination of core was improper.
[13:17:57] - Files status OK
[13:18:10] - Expanded 2911412 -> 16166417 (decompressed 555.2 percent)
[13:18:11] 
[13:18:11] Project: 1140 (Run 54, Clone 17, Gen 0)
[13:18:11] 
[13:18:13] Entering M.D.
[13:18:39] (Starting from checkpoint)
[13:18:40] Protein: p1140_RIBO_FSpeptide_EXT_nospring
[13:18:40] 
[13:18:40] Writing local files
[13:18:42] Completed 201737 out of 250000 steps  (81)
[14:21:20] Writing local files
[14:21:21] Completed 202500 out of 250000 steps  (81)
[16:43:21] Writing local files
[16:43:21] Completed 205000 out of 250000 steps  (82)
[18:43:22] Writing local files
[18:43:23] Completed 207500 out of 250000 steps  (83)
[20:44:47] Writing local files
[20:44:48] Completed 210000 out of 250000 steps  (84)
[22:43:17] Writing local files
[22:43:17] Completed 212500 out of 250000 steps  (85)
[00:40:58] Writing local files
[00:40:58] Completed 215000 out of 250000 steps  (86)
[02:42:12] Writing local files
[02:42:12] Completed 217500 out of 250000 steps  (87)
[05:34:53] Writing local files
[05:34:54] Completed 220000 out of 250000 steps  (88)
[10:02:59] Writing local files
[10:03:01] Completed 222500 out of 250000 steps  (89)
[13:25:17] Writing local files
[13:25:18] Completed 225000 out of 250000 steps  (90)
[15:49:26] Writing local files
[15:49:27] Completed 227500 out of 250000 steps  (91)
[18:01:30] Writing local files
[18:01:31] Completed 230000 out of 250000 steps  (92)
[20:08:36] Writing local files
[20:08:37] Completed 232500 out of 250000 steps  (93)
[23:07:19] Writing local files
[23:07:21] Completed 235000 out of 250000 steps  (94)
[02:47:53] Writing local files
[02:47:56] Completed 237500 out of 250000 steps  (95)
[05:19:00] Writing local files
[05:19:00] Completed 240000 out of 250000 steps  (96)
[07:22:01] Writing local files
[07:22:01] Completed 242500 out of 250000 steps  (97)
[09:22:47] Writing local files
[09:22:48] Completed 245000 out of 250000 steps  (98)
[11:23:36] Writing local files
[11:23:36] Completed 247500 out of 250000 steps  (99)
[14:16:56] Writing local files
[14:16:58] Completed 250000 out of 250000 steps  (100)
[14:16:59] Writing final coordinates.
[14:17:26] Past main M.D. loop
[14:18:28] 
[14:18:28] Finished Work Unit:
[14:18:28] - Reading up to 2032080 from "work/wudata_08.arc": Read 2032080
[14:18:31] - Reading up to 1871296 from "work/wudata_08.xtc": Read 1871296
[14:18:34] goefile size: 0
[14:18:35] logfile size: 100705
[14:18:35] Leaving Run
[14:18:37] - Writing 4011813 bytes of core data to disk...
[14:18:40]   ... Done.
[14:18:43] - Shutting down core
[14:18:43] 
[14:18:43] Folding@home Core Shutdown: FINISHED_UNIT
[14:18:51] CoreStatus = 64 (100)
[14:18:51] Sending work to server


[14:18:51] + Attempting to send results
[14:20:38] + Results successfully sent
[14:20:38] Thank you for your contribution to Folding@Home.
[14:20:38] + Number of Units Completed: 29

Whoa! Big Point WUs

Member

Unoriginal Macho Moderator

Captain Random Senior Evil

Member

Contributing Member

Member

Contributing Member

Member

Member

Captain Random Senior Evil

Member

Member

Captain Random Senior Evil

Member

Member

Captain Random Senior Evil

Member

Member

Captain Random Senior Evil

Member

Similar threads