[MITgcm-support] Cost of timestep versus checkpoint

Narayanan, Sri Hari Krishna snarayan at mcs.anl.gov
Thu Nov 27 08:43:26 EST 2014


Thank you. This is very helpful. I'll take a look at STDOUT.0000.
Happy Thanksgiving,
Krishna

________________________________
From: Dimitris Menemenlis [dmenemenlis at gmail.com]
Sent: Tuesday, November 25, 2014 9:57 AM
To: MITgcm Support
Subject: Re: [MITgcm-support] Cost of timestep versus checkpoint

if you look at end of the STDOUT.0000 file,
this information is there for your particular platform/set-up, e.g.,

(PID.TID 0000.0001)   Seconds in section "ALL                    [THE_MODEL_MAIN]":
(PID.TID 0000.0001)           User time:   1.1000000000000001
(PID.TID 0000.0001)         System time:   4.0000000000000001E-002
(PID.TID 0000.0001)     Wall clock time:   1.1838099956512451
(PID.TID 0000.0001)          No. starts:           1
(PID.TID 0000.0001)           No. stops:           1
...
(PID.TID 0000.0001)   Seconds in section "SEAICE_MODEL    [DO_OCEANIC_PHYS]":
(PID.TID 0000.0001)           User time:   7.0000000000000062E-002
(PID.TID 0000.0001)         System time:   0.0000000000000000
(PID.TID 0000.0001)     Wall clock time:   6.4920902252197266E-002
(PID.TID 0000.0001)          No. starts:           9
(PID.TID 0000.0001)           No. stops:           9
...
(PID.TID 0000.0001)   Seconds in section "DO_WRITE_PICKUP     [FORWARD_STEP]":
(PID.TID 0000.0001)           User time:   0.0000000000000000
(PID.TID 0000.0001)         System time:   0.0000000000000000
(PID.TID 0000.0001)     Wall clock time:   1.0550975799560547E-002
(PID.TID 0000.0001)          No. starts:           9
(PID.TID 0000.0001)           No. stops:           9


On Nov 25, 2014, at 7:53 AM, Matthew Mazloff <mmazloff at ucsd.edu<mailto:mmazloff at ucsd.edu>> wrote:

Hi Krishna

This ratio would be platform and setup dependent.

Matt


On Nov 25, 2014, at 6:34 AM, "Narayanan, Sri Hari Krishna" <snarayan at mcs.anl.gov<mailto:snarayan at mcs.anl.gov>> wrote:

Hi,

Has anyone ever measured the cost of writing a checkpoint to disk versus the cost of performing a single timestep for any MITgcm package?

Please let me know if I can clarify the question somehow.

Thanks
Krishna


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20141127/dfc63081/attachment.htm>


More information about the MITgcm-support mailing list