Difference between revisions of "Speeding up GEOS-Chem"

From Geos-chem
Jump to: navigation, search
 
Line 10: Line 10:
  
  
On this page we will describe some recommended methods for speeding up GEOS-Chem simulations.
+
This content has been moved to the [https://geos-chem.readthedocs.io/en/latest/gcc-guide/05-run/run-speedup.html '''Speed up a slow simulation''' section of <tt>geos-chem.readthedocs.io</tt>]
 
+
== Overview ==
+
 
+
[[Guide to GEOS-Chem performance|GEOS-Chem performance]] is continuously monitored by the [[GEOS-Chem Support Team]] and user community via official benchmarks and unofficial timing tests. It has been shown that running GEOS-Chem with [[Centralized_chemistry_time_step#Optimal_Configuration|the recommended timesteps]] from Philip et al. (2016) can increase run times by approximately a factor of 2.  To speed up GEOS-Chem simulations, users may choose to use any of the following options.
+
 
+
=== Run with coarser timesteps for faster turnaround and lower accuracy ===
+
 
+
Using these transport timesteps (as was done prior to [[GEOS-Chem v11-01|v11-01]]) should speed up your simulation by a factor of 2:
+
*30 min = 1800 sec (4&deg; x 5&deg;)
+
*15 min = 900 sec (2&deg; x 2.5&deg;)
+
 
+
=== Select one of the mechanisms in which detailed stratospheric chemistry is omitted ===
+
 
+
These mechanisms are: Tropchem, complexSOA, complexSOA_SVPOA.
+
 
+
=== Turn off diagnostics that you don't need ===
+
 
+
Several diagnostics are turned on by default in [[GEOS-Chem_input_files#The_input.geos_file<tt>input.geos</tt>]] (for bpch diagnostics) and [[GEOS-Chem_input_files#The_HISTORY.rc_file|<tt>HISTORY.rc</tt>]] (for netCDF diagnostics). Please check those files and turn off any diagnostics that you do not need to reduce time spent in file I/O.
+
 
+
=== Make sure debug flags are turned off ===
+
 
+
If you previously compiled GEOS-Chem with [[Debugging_GEOS-Chem#Recompile_GEOS-Chem_with_debug_options_turned_on|debug compiler flags]] turned on, then you should do <tt>make realclean</tt> and recompile prior to submitting your production runs. The debug options are known to slow down simulations.
+
 
+
== Validation ==
+
 
+
The table below demonstrates the speedups that can be obtained by changing the chemistry mechanisms and/or timesteps:
+
 
+
{| border=1 cellspacing=0 cellpadding=5
+
|-valign="top" bgcolor="#CCCCCC"
+
!width="125px"|Run name, <br>timesteps,<br>and submitter
+
!width="175px"|Machine or Node<br>and Compiler
+
!width="100px"|CPU vendor
+
!width="150px"|CPU model
+
!width="75px"|Speed [MHz]
+
!width="50px"|# of<br>CPUs
+
!width="85px"|CPU time
+
!width="120px"|Wall time
+
!width="75px|CPU / Wall<br>ratio
+
!width="75px"|% of ideal
+
 
+
|-valign="top" bgcolor="#CCFFFF"
+
!colspan="10"|Standard chemistry mechanism
+
 
+
|-valign="top"
+
|v11-01-standard<br>'''C20T10'''<br>Bob Yantosca
+
|regal16.rc.fas.harvard.edu<br>ifort 11.1
+
|GenuineIntel
+
|Intel(R) Xeon(R) CPU E5-2660 @ 2.20GHz
+
|2199.915
+
|8
+
|62554.07 s<br>'''17:22:34'''
+
|9355.80 s<br>'''02:35:59'''
+
|6.6861
+
|83.58
+
 
+
|-valign="top"
+
|v11-01-standard<br>'''C60T30'''<br>Melissa Sulprizio
+
|regal17.rc.fas.harvard.edu<br>ifort 11.1
+
|GenuineIntel
+
|Intel(R) Xeon(R) CPU E5-2660 @ 2.20GHz
+
|2199.993
+
|8
+
|20770.70 s<br>'''08:16:11'''
+
|4560.04 s<br>'''01:16:03'''<br><span style="color:red">2.05X faster than v11-01-standard C20T10</span>
+
|6.5286
+
|81.61
+
 
+
|-valign="top" bgcolor="#CCFFFF"
+
!colspan="10"|Tropchem chemistry mechanism
+
 
+
|-valign="top"
+
|v11-01-tropchem<br>'''C20T10'''<br>Melissa Sulprizio
+
|regal16.rc.fas.harvard.edu<br>ifort 11.1
+
|GenuineIntel
+
|Intel(R) Xeon(R) CPU E5-2660 @ 2.20GHz
+
|2199.993
+
|8
+
|26156.34 s<br>'''07:15:56'''
+
|3881.55 s<br>'''01:04:48'''<br><span style="color:blue">2.41X faster than v11-01-standard C20T10</span>
+
|6.7386
+
|84.23
+
 
+
|-valign="top"
+
|v11-01-tropchem<br>'''C60T30'''<br>Melissa Sulprizio
+
|regal16.rc.fas.harvard.edu<br>ifort 11.1
+
|GenuineIntel
+
|Intel(R) Xeon(R) CPU E5-2660 @ 2.20GHz
+
|2200.993
+
|8
+
|11802.31 s<br>'''03:16:42'''
+
|1923.35 s<br>'''00:32:12'''<br><span style="color:green">2.02X faster than v11-01-tropchem C20T10</span>
+
|6.1363
+
|76.70
+
 
+
|}
+
 
+
As you can see from the results above:
+
 
+
#Switching from the '''C20T10''' configuration to the coarser '''C60T30''' configuration results in a factor-of-2 speedup.
+
#*This is attributed to calling transport and chemistry less frequently.
+
#Switching from '''v11-01-standard''' simulation to '''v11-01-tropchem''' results in an additional factor-of-2 speedup. 
+
#*This is because the standard simulation contains the [[UCX chemistry mechanism|UCX option]], which includes a detailed stratospheric chemistry mechanism. 
+
#*The detailed stratospheric chemistry is of course omitted in the tropchem simulation.
+
  
  
 
----
 
----
 
'''''[[Profiling GEOS-Chem with the TAU performance system|Previous]] | [[Guide to GEOS-Chem performance]]'''''
 
'''''[[Profiling GEOS-Chem with the TAU performance system|Previous]] | [[Guide to GEOS-Chem performance]]'''''

Latest revision as of 18:21, 16 September 2022

Previous | Next | Guide to GEOS-Chem performance

  1. Parallelizing GEOS-Chem
  2. GEOS-Chem 7-day timing tests
  3. GEOS-Chem scalability
  4. GEOS-Chem 1-month benchmark timing results
  5. Profiling GEOS-Chem with the TAU performance system
  6. Speeding up GEOS-Chem


This content has been moved to the Speed up a slow simulation section of geos-chem.readthedocs.io



Previous | Guide to GEOS-Chem performance