[SCore-users] (no subject)
Shinji Sumimoto
s-sumi at flab.fujitsu.co.jp
Mon Jul 29 18:29:08 JST 2002
From: neural_shock at e-mail.ru
Subject: Re: [SCore-users] (no subject)
Date: Mon, 29 Jul 2002 13:04:18 +0400
Message-ID: <3d450512.44db.0 at e-mail.ru>
neural_shock> >neural_shock> >neural_shock> hello.
neural_shock> >neural_shock> >neural_shock>
neural_shock> >neural_shock> >neural_shock> 1. yes it works fine.
neural_shock> >neural_shock> >neural_shock> 2. i did not run lu class w. there are too few
neural_shock> memory on my test
neural_shock> >neural_shock> machines even
neural_shock> >neural_shock> >neural_shock> for class b of this programm ( as far as i think.
neural_shock> SCore's FEP
neural_shock> >neural_shock> does not print
neural_shock> >neural_shock> >neural_shock> "memory could be exhausted" but nodes begins swap
neural_shock> pages too heavy
neural_shock> >neural_shock> when i run
neural_shock> >neural_shock> >neural_shock> lu.B.2 ).
neural_shock> >neural_shock> >neural_shock>
neural_shock> >neural_shock> >neural_shock> with respect, mike.
neural_shock> >neural_shock> >
neural_shock> >neural_shock> >Are these true?
neural_shock> >neural_shock> >
neural_shock> >neural_shock> >Class W is smaller than Class A.
neural_shock> >neural_shock> >
neural_shock> >neural_shock> >Class S < W < A < B
neural_shock> >neural_shock> >
neural_shock> >neural_shock>
neural_shock> >neural_shock> 嫖s. you are right i was wrong. and i have tested it already. results are the
neural_shock> >neural_shock> same.
neural_shock> >
neural_shock> >So, how about class S?
neural_shock> >
neural_shock> >Shinji.
neural_shock> >------
neural_shock> >Shinji Sumimoto, Fujitsu Labs
neural_shock>
neural_shock> S is too small. there are no rpoblems. but as i think it is simply because checkpointing is already done when system shuts down scored.
Sorry.
How many memory do your nodes have?
Maybe 10MB free memory is needed for lu.W.2.
In my environment, lu.A.2 and lu.W.2 works fine.
====================================================
[s-sumi bin]$ scrun -nodes=2,checkpoint=5s,scored=server ./lu.A.2
SCore-D 5.0.0 connected (jid=2).
<0:0> SCORE: 2 nodes (2x1) ready.
NAS Parallel Benchmarks 2.2 -- LU Benchmark
Size: 64x 64x 64
Iterations: 250
Number of processes: 2
Time step 1
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
Time step 20
SCORE: Checkpointing ... FEP:WARNING SCore-D unexpectedly terminated.
FEP: [29/Jul/2002 18:31:19] Waiting for SCore-D restarted ...
FEP: [29/Jul/2002 18:31:24] SCore-D restarted.
SCore-D 5.0.0 connected (jid=2).
SCORE: Execution restarted from checkpoint.
Time step 20
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
SCORE: Checkpointing ... done.
====================================================
Shinji.
------
Shinji Sumimoto, Fujitsu Labs
More information about the SCore-users
mailing list