Force garbage collection in benchmarking code

gcpy/benchmark.py gcpy/budget_ox.py gcpy/budget_tt.py gcpy/ste_flux.py - At the end of each method, manually delete the larger objects (such as xarray Datasets) and call gc.collect() to force garbage collection. This should hopefully prevent problems with the benchmark scripts halting due to exceeding requested memory (which seems to happen if all 1-year benchmark artifacts are requested). Signed-off-by: Bob Yantosca <[email protected]>
geoschem · Oct 4, 2022 · 71cdba9 · 71cdba9
1 parent f0003b9
commit 71cdba9
Show file tree

Hide file tree

Showing 4 changed files with 112 additions and 40 deletions.
diff --git a/gcpy/benchmark.py b/gcpy/benchmark.py
@@ -14,12 +14,13 @@
 import sparselt.xr
 from joblib import Parallel, delayed
 from tabulate import tabulate
-import gcpy.util as util
+from gcpy import util
 from gcpy.plot import compare_single_level, compare_zonal_mean
 from gcpy.regrid import create_regridders
 from gcpy.grid import get_troposphere_mask
 from gcpy.units import convert_units
 import gcpy.constants as gcon
+import gc
 
 # Save warnings format to undo overwriting built into PyPDF2
 warning_format = warnings.showwarning
@@ -1203,13 +1204,20 @@ def createplots(filecat):
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-    refmetds = xr.Dataset()
-    devmetds = xr.Dataset()
-    second_ref = xr.Dataset()
-    second_dev = xr.Dataset()
-
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    #refmetds = xr.Dataset()
+    #devmetds = xr.Dataset()
+    #second_ref = xr.Dataset()
+    #second_dev = xr.Dataset()
+    del refds
+    del devds
+    del refmetds
+    del devmetds
+    del second_ref
+    del second_dev
+    gc.collect()
+
 
 def make_benchmark_emis_plots(
         ref,
@@ -1617,8 +1625,11 @@ def createfile_bench_cat(filecat):
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    del refds
+    del devds
+    gc.collect()
 
 
 def make_benchmark_emis_tables(
@@ -1792,11 +1803,16 @@ def make_benchmark_emis_tables(
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-    refmetds = xr.Dataset()
-    devmetds = xr.Dataset()
-
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    #refmetds = xr.Dataset()
+    #devmetds = xr.Dataset()
+    del refds
+    del devds
+    del refmetds
+    del devmetds
+    gc.collect()
+
 
 def make_benchmark_jvalue_plots(
         ref,
@@ -2190,9 +2206,12 @@ def make_benchmark_jvalue_plots(
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    del refds
+    del devds
+    gc.collect()
+
 
 def make_benchmark_aod_plots(
         ref,
@@ -2507,8 +2526,11 @@ def make_benchmark_aod_plots(
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    del refds
+    del devds
+    gc.collect()
 
 
 def make_benchmark_mass_tables(
@@ -2760,9 +2782,12 @@ def make_benchmark_mass_tables(
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    del refds
+    del devds
+    gc.collect()
+
 
 def make_benchmark_oh_metrics(
         ref,
@@ -2987,10 +3012,15 @@ def print_metrics_to_file(f, title1, title2, ref, dev, diff, pctdiff):
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-    refmetds = xr.Dataset()
-    devmetds = xr.Dataset()
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    #refmetds = xr.Dataset()
+    #devmetds = xr.Dataset()
+    del refds
+    del devds
+    del refmetds
+    del devmetds
+    gc.collect()
 
 
 def make_benchmark_wetdep_plots(
@@ -3272,11 +3302,16 @@ def make_benchmark_wetdep_plots(
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    refds = xr.Dataset()
-    devds = xr.Dataset()
-    refmetds = xr.Dataset()
-    devmetds = xr.Dataset()
-
+    #refds = xr.Dataset()
+    #devds = xr.Dataset()
+    #refmetds = xr.Dataset()
+    #devmetds = xr.Dataset()
+    del refds
+    del devds
+    del refmetds
+    del devmetds
+    gc.collect()
+
 
 def make_benchmark_aerosol_tables(
         devdir,
@@ -3558,10 +3593,14 @@ def print_aerosol_metrics(data, species_list, filename, title, label):
     # -------------------------------------------
     # Clean up
     # -------------------------------------------
-    ds_aer = xr.Dataset()
-    ds_spc = xr.Dataset()
-    ds_met = xr.Dataset()
-
+    #ds_aer = xr.Dataset()
+    #ds_spc = xr.Dataset()
+    #ds_met = xr.Dataset()
+    del ds_aer
+    del ds_spc
+    del ds_met
+    gc.collect()
+
 
 def make_benchmark_operations_budget(
         refstr,
@@ -4089,10 +4128,14 @@ def make_benchmark_operations_budget(
     # ------------------------------------------
     # Clean up
     # ------------------------------------------
-    df = pd.DataFrame()
-    ref_ds = xr.Dataset()
-    dev_ds = xr.Dataset()
-
+    #df = pd.DataFrame()
+    #ref_ds = xr.Dataset()
+    #dev_ds = xr.Dataset()
+    del df
+    del ref_ds
+    del dev_ds
+    gc.collect()
+
 
 def make_benchmark_mass_conservation_table(
         datafiles,
@@ -4239,6 +4282,8 @@ def make_benchmark_mass_conservation_table(
         print(' Abs diff =  {:>16.3f} g'.format(absdiff), file=f)
         print(' Pct diff =  {:>16.10f} %'.format(pctdiff), file=f)
 
+    gc.collect()
+
 
 def get_species_database_dir(config):
     """

diff --git a/gcpy/budget_ox.py b/gcpy/budget_ox.py
@@ -15,6 +15,7 @@
 import gcpy.constants as constants
 from gcpy.grid import get_troposphere_mask
 import gcpy.util as util
+import gc
 
 # Suppress harmless run-time warnings (mostly about underflow in division)
 warnings.filterwarnings("ignore", category=RuntimeWarning)
@@ -753,3 +754,14 @@ def global_ox_budget(
         drydep,
         metrics
     )
+
+    # ==================================================================
+    # Force garbage collection
+    # ==================================================================
+    del globvars
+    del mass
+    del prodloss
+    del wetdep
+    del drydep
+    del metrics
+    gc.collect()
diff --git a/gcpy/budget_tt.py b/gcpy/budget_tt.py
@@ -18,6 +18,7 @@
 import gcpy.constants as constants
 from gcpy.grid import get_troposphere_mask
 import gcpy.util as util
+import gc
 
 # Suppress harmless run-time warnings (mostly about underflow in division)
 warnings.filterwarnings("ignore", category=RuntimeWarning)
@@ -944,3 +945,11 @@ def transport_tracers_budgets(
             data,
             key
         )
+
+    # ==================================================================
+    # Force garbage collection
+    # ==================================================================
+    del globvars
+    del data
+    del ds
+    gc.collect()
diff --git a/gcpy/ste_flux.py b/gcpy/ste_flux.py
@@ -15,6 +15,7 @@
 import pandas as pd
 import xarray as xr
 import gcpy.constants as physconsts
+import gc
 
 # Suppress harmless run-time warnings (mostly about underflow in division)
 warnings.filterwarnings("ignore", category=RuntimeWarning)
@@ -325,3 +326,8 @@ def make_benchmark_ste_table(devstr, files, year,
 
     # Print the STE fluxes
     print_ste(globvars, df)
+
+    # Force garbage collection
+    del globvars
+    del df
+    gc.collect()