FAH-News

[scsimodo]

Was ist ein SMP Client ???

Client für CPUs mit mehreren Cores

[duftkerze]

smp= Symmetrisches Multiprozessorsystem

d.h. paralleles abarbeiten von instruktionen .

in diesem fall heist es, es werden mehrere CPUs von einem clienten unterstützt.bisher pro client eine cpu.

der aktuelle GPU-client ist schon ein smp. (CPU+GPU)

hoffe geholfen zu haben

gruss dufti

[arminius]

Hier ein Artikel zum Thema: http://www.wired.com/news/technology/co ... n_index_12
a.

[duftkerze]

jo wollte mal fragen ob jemand von euch den smp clienten tested.


Ich tu es gerade....

habe testweise suse 10.1 auf meinem microdrive installiert.

über nacht hat er bereits 50 % abgearbeitet mit 4 nodes auf meinem x2
es werden 5millionen steps bearbeitet und es gibt über 500 credits dafür.
den erste versuch habe ich mit ubuntu gemacht. da hatte ich ein paar probs.
z.bsp. kein i-net kontakt mit dem clienten obwohl der browser funktionierte.
und nur 90% prozent auslastung bei beiden cores.

[scsimodo]

jo wollte mal fragen ob jemand von euch den smp clienten tested.

Hab ihn mal auf meinem Mac Mini installiert und nach ein bisschen Kampf auch zu laufen bekommen. Hab's dann mal gestartet, lief auch soweit ganz ordentlich. Allerdings schreibt er mir auch nach 2 Stunden keine Checkpoints, sondern fängt wieder von vorne an. Also unbrauchbar bei diesen Laufzeiten. Wollte die WU wieder zurückschicken und nichts mehr holen, das scheint allerdings ein Ding der Unmöglichkeit zu sein, die FAH-Homepage sagt darüber nichts. Also müssen die halt warten bis die WU verfällt, selber Schuld...

[duftkerze]

hier mal ne übersicht wie es aussieht ,wenn ne wu beendet wird auf meinem
x2 4800.........

[03:06:36] Completed 4900000 out of 5000000 steps (98 %)
[03:15:38] Writing local files
[03:15:38] Completed 4950000 out of 5000000 steps (99 %)



M E G A - F L O P S A C C O U N T I N G

Parallel run - timing based on wallclock.
RF=Reaction-Field FE=Free Energy SCFE=Soft-Core/Free Energy
T=Tabulated W3=SPC/TIP3p W4=TIP4p (single or pairs)
NF=No Forces

Computing: M-Number M-Flops % of Flops
-----------------------------------------------------------------------
Coulomb 29403.424265 793892.455155 1.5
Coulomb [W3] 122.461673 9796.933840 0.0
Coulomb + LJ 52254.462108 1985669.560104 3.8
Coulomb + LJ [W3] 14561.294024 1325077.756184 2.5
Coulomb + LJ [W3-W3] 169820.185463 41605945.438435 79.0
Outer nonbonded loop 39999.567054 399995.670540 0.8
1,4 nonbonded interactions 2182.401408 196416.126720 0.4
NS-Pairs 62748.693133 1317722.555793 2.5
Reset In Box 1560.240066 14042.160594 0.0
Shift-X 62409.240264 374455.441584 0.7
CG-CoM 575.363712 16685.547648 0.0
Sum Forces 46806.930198 46806.930198 0.1
Angles 1523.650983 248355.110229 0.5
Propers 176.700114 40464.326106 0.1
RB-Dihedrals 1748.401128 431855.078616 0.8
Virial 15769.710174 283854.783132 0.5
Ext.ens. Update 15602.310066 842524.743564 1.6
Stop-CM 1560.230000 15602.300000 0.0
Calc-Ekin 15602.320132 421262.643564 0.8
Shake 8021.377035 240641.311050 0.5
Constraint-V 15602.310066 93613.860396 0.2
Shake-Init 844.750545 8447.505450 0.0
Constraint-Vir 15613.160073 374715.841752 0.7
Settle 4922.803176 1590065.425848 3.0
-----------------------------------------------------------------------
Total 52677909.506502 100.0
-----------------------------------------------------------------------

NODE (s) Real (s) (%)
Time: 16851.000 16851.000 100.0
4h40:51
(Mnbf/s) (GFlops) (ns/day) (hour/ns)
Performance: 15.795 3.126 51.273 0.468
[03:24:39] Writing local files
[03:24:39] Completed 5000000 out of 5000000 steps (100 %)
[03:24:39] Writing final coordinates.
[03:24:39] Past main M.D. loop
[03:24:39] Will end MPI now
[03:25:39]
[03:25:39] Finished Work Unit:
[03:25:39] - Reading up to 241704 from "work/wudata_02.arc": Read 241704
[03:25:39] - Reading up to 7300024 from "work/wudata_02.xtc": Read 7300024
[03:25:40] goefile size: 0
[03:25:40] logfile size: 134832
[03:25:40] Leaving Run
[03:25:40] - Writing 7808852 bytes of core data to disk...
[03:25:40] ... Done.
[03:25:42] - Shutting down core
[03:25:42]
[03:25:42] Folding@home Core Shutdown: FINISHED_UNIT
[0]0:Return code = 100
[0]1:Return code = 0, signaled with Quit
[0]2:Return code = 0, signaled with Quit
[0]3:Return code = 0, signaled with Quit
[03:25:47] CoreStatus = 64 (100)
[03:25:47] Sending work to server


[03:25:47] + Attempting to send results
[03:28:24] + Results successfully sent
[03:28:24] Thank you for your contribution to Folding@Home.
[03:28:24] + Number of Units Completed: 2

[0]0:Return code = 0, signaled with Quit
[0]1:Return code = 0, signaled with Quit
[0]3:Return code = 0, signaled with Quit
[03:32:29] - Preparing to get new work unit...
[03:32:29] + Attempting to get work packet
[03:32:29] - Connecting to assignment server
[03:32:30] - Successful: assigned to (171.64.65.56).
[03:32:30] + News From Folding@Home: Welcome to Folding@Home
[03:32:30] Loaded queue successfully.
[03:32:33] + Closed connections
[03:32:33]
[03:32:33] + Processing work unit
[03:32:33] Core required: FahCore_a1.exe
[03:32:33] Core found.
[03:32:33] Working on Unit 03 [November 26 03:32:33]
[03:32:33] + Working ...
[03:32:33]
[03:32:33] *------------------------------*
[03:32:33] Folding@Home Gromacs SMP Core
[03:32:33] Version 1.72 (November 12, 2006)
[03:32:33]
[03:32:33] Preparing to commence simulation
[03:32:33] - Ensuring status. Please wait.
[03:32:33] - Starting from initial work packet
[03:32:33]
[03:32:33] Project: 3007 (Run 4, Clone 7, Gen 6)
[03:32:33]
[03:32:33] Assembly optimizations on if available.
[03:32:33] Entering M.D.
[03:32:50] itial work pa- Starting from initial work packet
[03:32:50]
[03:32:50] Project: 3007 (Run 4, Clone 7, Gen 6)
[03:32:50]
[03:32:50] Entering M.D.
NNODES=4, MYRANK=1, HOSTNAME=x24800
NNODES=4, MYRANK=2, HOSTNAME=x24800
NNODES=4, MYRANK=3, HOSTNAME=x24800
NNODES=4, MYRANK=0, HOSTNAME=x24800
NODEID=1 argc=15
NODEID=2 argc=15
NODEID=3 argc=15
NODEID=0 argc=15
Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
Copyright (c) 1991-2000, University of Groningen, The Netherlands.
Copyright (c) 2001-2004, The GROMACS development team,
check out http://www.gromacs.org for more information.

This inclusion of Gromacs code in the Folding@Home Core is under
a special license (see http://folding.stanford.edu/gromacs.html)
specially granted to Stanford by the copyright holders. If you
are interested in using Gromacs, visit www.gromacs.org where
you can download a free version of Gromacs under
the terms of the GNU General Public License (GPL) as published
by the Free Software Foundation; either version 2 of the License,
or (at your option) any later version.

[03:32:56] mdrunner cpfilename:
starting mdrun 'SMP-msv-03'
5000000 steps, 10000.0 ps.

[03:32:57] mpleted 0 out of 5000000 steps (0 %)
[03:32:57] s
[03:32:57] Extra SSE boost OK.
[03:41:56] mpleted 50000 out of 5000000 steps (1 %)
[03:50:57] Writing local files
[03:50:57] Completed 100000 out of 5000000 steps (2 %)
[03:59:58] Writing local files
[03:59:58] Completed 150000 out of 5000000 steps (3 %)

[duftkerze]

wenn ich auf meinem neuen dual core 2 suse linux 10.1 zum laufen bekomme
teste ich den smp clienten auch mal darauf..
bis jetzt schaffe ich nicht mal bis zur installation,
nach dem einlesen des kernels findet er die dvd nicht mehr.

[duftkerze]

[02:54:09] Writing local files
[02:54:09] Completed 1450000 out of 5000000 steps (29 %)
[03:01:59] Warning: long 1-4 interactions
-------------------------------------------------------
Program Core_A0.exe, VERSION 3.3
Source code file: nsgrid.c, line: 235

Range checking error:
Explanation: During neighborsearching, we assign each particle to a grid
based on its coordinates. If your system contains collisions or parameter
errors that give particles very high velocities you might end up with some
coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot
put these on a grid, so this is usually where we detect those errors.
Make sure your system is properly energy-minimized and that the potential
energy seems reasonable before trying again.

Variable ci has value -2147483648. It should have been within [ 0 .. 1320 ]
Please report this to the mailing list (gmx-users@gromacs.org)
-------------------------------------------------------

Thanx for Using GROMACS - Have a Nice Day

Error on node 1, will try to stop all the nodes
Halting parallel program Core_A0.exe on CPU 1 out of 4

gcq#0: Thanx for Using GROMACS - Have a Nice Day

[cli_1]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, -1) - process 1
-------------------------------------------------------
Program Core_A0.exe, VERSION 3.3
Source code file: nsgrid.c, line: 235

Range checking error:
Explanation: During neighborsearching, we assign each particle to a grid
based on its coordinates. If your system contains collisions or parameter
errors that give particles very high velocities you might end up with some
coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot
put these on a grid, so this is usually where we detect those errors.
Make sure your system is properly energy-minimized and that the potential
energy seems reasonable before trying again.

Variable ci has value -2147483648. It should have been within [ 0 .. 1320 ]
Please report this to the mailing list (gmx-users@gromacs.org)
-------------------------------------------------------

Thanx for Using GROMACS - Have a Nice Day

Error on node 2, will try to stop all the nodes
Halting parallel program Core_A0.exe on CPU 2 out of 4

gcq#0: Thanx for Using GROMACS - Have a Nice Day

[cli_2]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, -1) - process 2
[03:02:45]
[03:02:45] Folding@home Core Shutdown: INTERRUPTED
-------------------------------------------------------
Program Core_A0.exe, VERSION 3.3
Source code file: nsgrid.c, line: 235

Range checking error:
Explanation: During neighborsearching, we assign each particle to a grid
based on its coordinates. If your system contains collisions or parameter
errors that give particles very high velocities you might end up with some
coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot
put these on a grid, so this is usually where we detect those errors.
Make sure your system is properly energy-minimized and that the potential
energy seems reasonable before trying again.

Variable ci has value -2147483648. It should have been within [ 0 .. 1320 ]
Please report this to the mailing list (gmx-users@gromacs.org)
-------------------------------------------------------

Thanx for Using GROMACS - Have a Nice Day

Error on node 0, will try to stop all the nodes
Halting parallel program Core_A0.exe on CPU 0 out of 4

gcq#0: Thanx for Using GROMACS - Have a Nice D-------------------------------------------------------
Program Core_A0.exe, VERSION 3.3
Source code file: nsgrid.c, line: 235

Range checking error:
Explanation: During neighborsearching, we assign each particle to a grid
based on its coordinates. If your system contains collisions or parameter
errors that give particles very high velocities you might end up with some
coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot
put these on a grid, so this is usually where we detect those errors.
Make sure your system is properly energy-minimized and that the potential
energy seems reasonable before trying again.

Variable ci has value -2147483648. It should have been within [ 0 .. 1320 ]
Please report this to the mailing list (gmx-users@gromacs.org)
-------------------------------------------------------

Thanx for Using GROMACS - Have a Nice Day

Error on node 3, will try to stop all the nodes
Halting parallel program Core_A0.exe on CPU 3 out of 4

gcq#0: Thanx for Using GROMACS - Have a Nice Day

[cli_0]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, -1) - process 0
ay

[cli_3]: aborting job:
application called MPI_Abort(MPI_COMM_WORLD, -1) - process 3
[03:02:50] CoreStatus = FF (255)
[03:02:50] Client-core communications error: ERROR 0xff
[03:02:50] Deleting current work unit & continuing...
[0]0:Return code = 0, signaled with Quit
[0]2:Return code = 0, signaled with Quit
[0]3:Return code = 0, signaled with Quit
[03:07:12] - Preparing to get new work unit...
[03:07:12] + Attempting to get work packet
[03:07:12] - Connecting to assignment server
[03:07:13] - Successful: assigned to (0.0.0.0).
[03:07:13] + News From Folding@Home: Welcome to Folding@Home
[03:07:13] Work Unit has an invalid address.
[03:07:13] - Error: Attempt #1 to get work failed, and no other work to do.
Waiting before retry.

[duftkerze]

es gibt einen neuen gpu-client mit der unterstützung von x1600,x1800
und x1900.
also ran an die grafikarten
ebenfalls gibt es einen neuen smp-client.bin schon beim testen

[SpeedKing]

es gibt einen neuen gpu-client mit der unterstützung von x1600,x1800 und x1900

Also doch! Hätte mich doch gewundert!

[duftkerze]

ich habe die Ehre die grösste WU bei Fah zu berechnen.

es gibt 1760 credits dafür

nach 6 stunden hatte ich 19 prozent geschafft....

[Rincewind]

Na dann viel Spaß! dann hast du ja nur noch ca 24h vor dir *g*