| Home | Reviews | Forums | Downloads | $ EXTREME Deals $ | RealTime Pricing | Free Magazines | Gear | Folding Stats | Newsletter | Contact Us |
|
|
|||
|
||||
|
|||||||
| Register | Forum Rules | FAQ | Search | Today's Posts | Mark Forums Read |
| Welcome Guest Visitor! Please Register, It's Free and Fun To Participate! | |
|
The EXTREME Overclocking Forums are a place for people to learn how to overclock and tweak their PC's components like the CPU, memory (RAM), or video card in order to gain the maximum performance out of their system. There are lots of discussions about new processors, graphics cards, cooling products, power supplies, cases, and so much more!
You are currently viewing our boards as a "guest" which gives you limited access to view most discussions. You need to register before you can post: click the register link to proceed. Before you register, please read the forum rules. By joining our free community you will have access to post topics, communicate privately with other members (PM), respond to polls, upload your own pictures, and access many other special features. Registration is fast, simple, and absolutely free! To start viewing messages, select the forum that you want to visit from the selection below. After you have registered and read the forum rules, you can check out the FAQ for more information on using the forum. We hope you enjoy your stay here! Note To Spammers: We do not allow unsolicited advertising! Spam is usually reported & deleted within minutes of it being posted, so don't waste your time (or ours)! |
|
| Please Register to Post a Reply |
|
|
Thread Tools |
04-01-2009, 06:21 AM
|
#1 | ||||
|
In The Fold
Senior Member
|
VM SMP Client Fails to Get New WU
Code:
[10:40:44] Completed 245000 out of 250000 steps (98%)
[10:51:20] Completed 247500 out of 250000 steps (99%)
[11:01:58] Completed 250000 out of 250000 steps (100%)
Writing final coordinates.
Average load imbalance: 4.3 %
Part of the total run time spent waiting due to load imbalance: 3.0 %
Steps where the load balancing was limited by -rdd, -rcon and/or -dds: Z 0 %
Parallel run - timing based on wallclock.
NODE (s) Real (s) (%)
Time: 63691.000 63691.000 100.0
17h41:31
(Mnbf/s) (GFlops) (ns/day) (hour/ns)
Performance: 56.107 8.545 26.290 0.913
gcq#0: Thanx for Using GROMACS - Have a Nice Day
[11:02:02] DynamicWrapper: Finished Work Unit: sleep=1000
[11:02:03]
[11:02:03] Finished Work Unit:
[11:02:03] - Reading up to 21148416 from "work/wudata_03.trr": Read 21148416
[11:02:05] trr file hash check passed.
[11:02:05] - Reading up to 4523360 from "work/wudata_03.xtc": Read 4523360
[11:02:05] xtc file hash check passed.
[11:02:05] edr file hash check passed.
[11:02:05] logfile size: 175881
[11:02:05] Leaving Run
[11:02:07] Done with run, master node
[11:02:07] - Writing 26029561 bytes of core data to disk...
[11:02:11] ... Done.
[11:02:21] - Shutting down core
[11:02:21]
[11:02:21] Folding@home Core Shutdown: FINISHED_UNIT
Error encountered before initializing MPICH
[11:05:30] CoreStatus = 64 (100)
[11:05:30] Unit 3 finished with 75 percent of time to deadline remaining.
[11:05:30] Updated performance fraction: 0.736821
[11:05:30] Sending work to server
[11:05:30] Project: 2672 (Run 0, Clone 91, Gen 100)
[11:05:30] + Attempting to send results [April 1 11:05:30 UTC]
[11:05:30] - Reading file work/wuresults_03.dat from core
[11:05:35] (Read 26029561 bytes from disk)
[11:05:35] Connecting to http://171.64.65.56:8080/
This is where I killed the client and restarted fah.
After looking at the log I see that the finished
wu's data has not been uploaded to FAH.
[12:16:41] ***** Got an Activate signal (2)
[12:16:41] Killing all core threads
Folding@Home Client Shutdown.
jim@jim-VM1:~/folding/fah$ ./fah6
Note: Please read the license agreement (fah6 -license). Further
use of this software requires that you have read and accepted this agreement.
2 cores detected
--- Opening Log file [April 1 12:22:14 UTC]
# Linux SMP Console Edition ###################################################
###############################################################################
Folding@Home Client Version 6.24beta
http://folding.stanford.edu
###############################################################################
###############################################################################
Launch directory: /home/jim/folding/fah
Executable: ./fah6
Arguments: -smp -forceasm -verbosity 9
[12:22:14] - Ask before connecting: No
[12:22:14] - User name: bettik (Team 11314)
[12:22:14] - User ID: 2B8826C63C8A8041
[12:22:14] - Machine ID: 1
[12:22:14]
[12:22:14] Loaded queue successfully.
[12:22:14] - Autosending finished units... [April 1 12:22:14 UTC]
[12:22:14] Trying to send all finished work units
[12:22:14] Project: 2672 (Run 0, Clone 91, Gen 100)
[12:22:14] + Attempting to send results [April 1 12:22:14 UTC]
[12:22:14] - Reading file work/wuresults_03.dat from core
[12:22:14] - Preparing to get new work unit...
[12:22:14] (Read 26029561 bytes from disk)
[12:22:14] Connecting to http://171.64.65.56:8080/
[12:22:14] + Attempting to get work packet
[12:22:14] - Will indicate memory of 626 MB
[12:22:14] - Connecting to assignment server
[12:22:14] Connecting to http://assign.stanford.edu:8080/
[12:22:16] Posted data.
[12:22:16] Initial: 40AB; - Successful: assigned to (171.64.65.56).
[12:22:16] + News From Folding@Home: Welcome to Folding@Home
[12:22:16] Loaded queue successfully.
[12:22:16] Connecting to http://171.64.65.56:8080/
[12:22:22] Posted data.
[12:22:22] Initial: 0000; - Receiving payload (expected size: 4836862)
[12:23:01] - Downloaded at ~121 kB/s
[12:23:01] - Averaged speed for that direction ~159 kB/s
[12:23:01] + Received work.
[12:23:01] + Closed connections
[12:23:01]
[12:23:01] + Processing work unit
[12:23:01] At least 4 processors must be requested.Core required: FahCore_a2.exe
[12:23:01] Core found.
[12:23:01] Working on queue slot 04 [April 1 12:23:01 UTC]
[12:23:01] + Working ...
[12:23:01] - Calling './mpiexec -np 4 -host 127.0.0.1 ./FahCore_a2.exe -dir work/ -suffix 04 -checkpoint 15 -forceasm -verbose -lifeline 25095 -version 624'
[12:23:01]
[12:23:01] *------------------------------*
[12:23:01] Folding@Home Gromacs SMP Core
[12:23:01] Version 2.04 (Thu Jan 29 16:43:57 PST 2009)
[12:23:01]
[12:23:01] Preparing to commence simulation
[12:23:01] - Ensuring status. Please wait.
[12:23:02] Called DecompressByteArray: compressed_data_size=4836350 data_size=23979541, decompressed_data_size=23979541 diff=0
[12:23:02] - Digital signature verified
[12:23:02]
[12:23:02] Project: 2669 (Run 11, Clone 87, Gen 77)
[12:23:02]
[12:23:02] Assembly optimizations on if available.
[12:23:02] Entering M.D.
[12:23:13] on if available.
[12:23:13] Entering M.D.
NNODES=4, MYRANK=1, HOSTNAME=jim-VM1
NNODES=4, MYRANK=2, HOSTNAME=jim-VM1
NNODES=4, MYRANK=3, HOSTNAME=jim-VM1
NNODES=4, MYRANK=0, HOSTNAME=jim-VM1
NODEID=0 argc=20
NODEID=1 argc=20
NODEID=2 argc=20
NODEID=3 argc=20
:-) G R O M A C S (-:
Groningen Machine for Chemical Simulation
:-) VERSION 4.0.3_pre (-:
Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
Copyright (c) 1991-2000, University of Groningen, The Netherlands.
Copyright (c) 2001-2008, The GROMACS development team,
check out http://www.gromacs.org for more information.
:-) mdrun (-:
Reading file work/wudata_04.tpr, VERSION 3.3.99_development_20070618 (single precision)
Note: tpx file_version 48, software version 58
NOTE: The tpr file used for this simulation is in an old format, for less memory usage and possibly more performance create a new tpr file with an up to date version of grompp
Prior to starting the new wu the data from the previous job is now sent,
then the new job starts.
Making 1D domain decomposition 1 x 1 x 4
starting mdrun '22872 system'
19500002 steps, 39000.0 ps (continuing from step 19250002, 38500.0 ps).
[12:26:19] Posted data.
[12:26:20] Initial: 0000; - Uploaded at ~100 kB/s
[12:26:27] - Averaged speed for that direction ~121 kB/s
[12:26:27] + Results successfully sent
[12:26:27] Thank you for your contribution to Folding@Home.
[12:26:27] + Number of Units Completed: 13
[12:26:38] + Sent 1 of 1 completed units to the server
[12:26:38] - Autosend completed
[12:34:14] ps (1%)
[12:45:15] Completed 5008 out of 250000 steps (2%)
Last edited by bettik : 04-01-2009 at 08:42 AM. Reason: Additional Info |
||||
|
|
|
04-01-2009, 01:27 PM
|
#2 | ||||
|
Folder
Senior Member
|
Reinstall not a bad idea...
I still can't keep my vm smp from locking my system...... grrrrrrrr |
||||
|
|
|
04-01-2009, 01:35 PM
|
#3 | ||||
|
Extreme Overclocker
Senior Member
|
Quote:
 )
|
||||
|
|
|
04-01-2009, 05:00 PM
|
#4 | ||||
|
In The Fold
Senior Member
|
Quote:
Furthermore, after the restart and prior to working on the new wu the client did indeed upload all of the data from the prior wu. That is clear in the log if you read it. So how did a2 wu's get "stolen from legit quads". If anything I wrote was unclear all you had to do was ask but instead you have once again jumped to a conclusion, something I might add that you are becoming famous for here. As far as you not preaching...I don't think that's possible. |
||||
|
|
|
04-01-2009, 05:16 PM
|
#5 | ||||
|
Extreme Overclocker
Senior Member
|
You misunderstood, as did I btw as I didn't even notice the two hour delay between ending mpi loop and the sigterm signal. I thought it was a couple of minutes, exuse my wrong interpretation
 The comment on a2 wu's and their intended usage was just that, want to read the comment posted at FF from people who bought 8 core machines just for those a2 wu's you're running on two cores because you get more ppd with them? I know where I'm famous for, it's not for jumping to an conclusion it's for speaking the truth to people who don't like to be told such, and if you disagree please point me exactly where I made those wrong conclusions ( which offcourse I done at some time, just as everyone else. But I more then make up for that with coming to the right conclusions which you will notice when you try to bring facts into that dicussion instead of personal opinion ). Meh you remind me of VooDoo, maybe it's not even worth my time, on ff I repleid since I care more about what people think of me then here. If people here don't like me because they want to beleive the slanter of a vocal few who got ****ed of by me, fine doesn't bother me  Now, if anything I wrote was unclear, feel free to ask for clarification, but if you do be prepared to back everything up with links and qoutes I can rebute/reply to or I won't even bother.
Last edited by _MtM_ : 04-01-2009 at 05:23 PM. |
||||
|
|
|
04-01-2009, 06:59 PM
|
#6 | ||||
|
In The Fold
Senior Member
|
Quote:
Quote:
"But you said you would rather run vm's, and I agree those are a better way off fully utilizing a quad so if you want to run those I'll tell you one thing, it's a 3 step process. 1. Download and install vmplayer -> http://www.vmware.com/products/player/ 2. Download the NF folding applicance -> http://reilly.homeip.net/folding/vm.html 3. Start the vmplayer, load the appliance. Note the ip it get's. Point your browser to the IP and configure the client as you please. It's really simple this way. Edit: I should correct myself. Better use a quad isn't really true. Yes you get more points, but you will have slower turnaround times and PG has stressed that return time is of the essence. Each wu needs to be folded before a new one can be given out, the project is serial in nature and therefore, doing one project in 20 hours is better then doing two projects in 30 hours even if it gives less points. Edit2: Now I'm doubting myself as I use workstation not player, not sure if player allows two instances to run at the same time but I think it does. With an a2 core, you'll be surprised 9k Is something I can do with 2 vm's and a 9600gt " Guess who said that? YOU! That is a quote from you to me when I asked back in February how the best way was to set up my rig. Since then I have completed 703 WU's, left my computer on 24/7 to do so, paid higher electric bills and probably taken some life off my rig for the same cause you seem to have anointed yourself the lead champion of. If there is something in your quote above that told me to configure a different way, it wasn't clear to me. I was a noob with this and you knew it. That is why I asked for help. Quote:
If you feel lectured I'm sorry. I would ask, how does it feel? You once lectured me on the virtues of knowledge. Sheez! Last edited by bettik : 04-01-2009 at 07:19 PM. |
||||
|
|
|
04-02-2009, 02:49 AM
|
#7 | ||||
|
Extreme Overclocker
Senior Member
|
Wow is all I can say.
Did you notice I corrected myself in that same post though? Yes I might have been bad as well, mostly because I take things to personal so if I get the impression from people that even they don't care about what's best for science or epeen ( read - people who ought the be caring, and lecturing as they have a responcibility to inform people imo ). So, in an emotional reaction I say, screw it why not use vm's, it's not like people really care. But I always correct that initial reaction. Yeah I feel lectured, and I should, don't think I'm to thick to realize my own mistakes. I don't ever want to pretend to be the 'champion' of anything, if you want a champion look for AtlasFolder, or closer to home Jevans and DriveEuro and mjbservices to name a few from memory alone. Or some people who don't seem to post here anymore but where instrumental to get me more interested in this whole project, people like femfolder. I'm sorry Bettik, think that VooDoo had more influence on me then I realised and it got carried over ( that happens with me, I got borderline features among allot of things and I thought I was doing pretty good but I realise that I was wrong ). I'm going to end this with just saying something I hope you'll understand and appriciate, thanks for the well timed lecture I needed it. Edit: One thing though, your client didn't fail to get a new unit, it failed and hung at 100% as is a known problem http://foldingforum.org/viewtopic.php?f=8&t=3989 ( just a random thread about it as there are more ). Seems it's one of those problems which aren't directly client related but have a strong dependancy on the os and external influences mostly network related  Additional Comment: If this is where you think 'I lectured you on the virtues of knowledge' I think you didn't pick of the sarcasm -> http://forums.extremeoverclocking.co...d.php?t=315687 Last 3 posts.. Last edited by _MtM_ : 04-02-2009 at 02:49 AM. Reason: Automerged Doublepost |
||||
|
|
|
04-02-2009, 06:34 AM
|
#8 | ||||
|
In The Fold
Senior Member
|
_MtM_ Thank you for the sincerity of your post. I am not comfortable with these confrontations so I hope we can put this behind us.
Now back the original question. Following up on this new information regarding a2 units running on a vm, are you suggesting that I not use the linux smp client in a vm? I'm open to this, but help me understand. I realize that a single vm is supposed to utilize only 2 cores but something is not making sense to me here. Attached is a screen shot of the windows task manager. I currently have only ONE vm running an a2 wu and tm shows cpu usage at 50% and the process is spread over all 4 cores. So I guess this equates to 2 cores running at 100%. So are you saying that if the same client was running under native linux I would have 100% utilization across all four cores? Or, am I missing something important? Let me know what you think. |
||||
|
|
|
04-02-2009, 06:54 AM
|
#9 | ||||
|
Extreme Overclocker
Senior Member
|
I would be happy to leave it behind us, I hope that I learned from your correction as well. Saddens me to realize I'm not convinced though, as I been going through this cycle for awhile now.
Yes vm only exposes 2 cores, and one of the main advantage of a2 cores is their ability to scale with any amount of cores. That's what convinced me to quit using vm's, I realise you will get more cpu saturation if you use them but it's not really in the best intrest of the project. I use my machine for allot more then folding or it would have been on native linux a long time ago. Yes, on native linux if you get an a2 core even if you have an 8 core or more box it will use all off them, it's designed that way. By doing such, you're going to decrease the turnaround time which is key with hpc clients and their work units, so yes it would be better for the science. Not for ppd, as multiple instances will give more ppd, and not for cpu saturation since not all threads will peg a core to it's maximum, but it will make sure your unit is turned around sooner and this has been said more then once to be the key benefit of hpc clients. I even removed the vm from my dual core e6600 for that same reason. Yes, windows smp would also only be able to use 2 cores, no diffrence there, but they don't advice/ask people to only run them on quad core or more machines for nothing that's one minor reason. The other, major one, is that I noticed that a2 wu's are not really abundant, and I saw allot of people who invested in heavy cpu setups with dual quads just for the a2 wu's not being able to get any work for them, while I was able to get them and actually slow down the chain of wu's because I was running a dual core to process those a2 wu's. |
||||
|
|
|
04-02-2009, 07:15 AM
|
#10 | ||||
|
In The Fold
Senior Member
|
Ok. So if I understand your advice correctly, and from a stictly "whats best for the cause" perspective no one should be using hpc clients in a vm? Hmm. Seems I've come full circle since February.
Now, assuming I move back to using the windows smp console client, would your advise be to run one instance of the client or two with the affinity changer service? I hope the question makes sense. edit: one thing I still don't get. Why does task manager show all four cores being used? |
||||
|
|
|
04-02-2009, 08:22 AM
|
#11 | ||||
|
Extreme Overclocker
Senior Member
|
Correct, running linux in a vm is not recommended for the reasons given. The same goes for dual windows smp. The proper advice has always been 'one thread per core', where with gpu clients it's debatable since the cpu utilisation is low especially on Nvidia.
Taskmanager shows correct info, it's the windows scheduler which devides the load over all cores, the load being 50% of your quad since the vm only exposes two cores. |
||||
|
|
Sponsored Links:
|
| Please Register to Post a Reply |
| Thread Tools | |
|
|
