Merge PR #319 (Add benchmark scripts to scrape timing info)

This merge brings PR #310 (Add benchmark scripts to scrape timing information from log files, by @yantosca) into the GCPy 1.5.0 development stream. This PR adds the functionality to produce timing tables of Ref & Dev models from benchmark logfile output. Signed-off-by: Bob Yantosca <[email protected]>
geoschem · May 13, 2024 · 4be43b5 · 4be43b5
2 parents f66e81e + e0bd4a7
commit 4be43b5
Show file tree

Hide file tree

Showing 13 changed files with 1,099 additions and 205 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -20,10 +20,15 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Function `rename_speciesconc_to_speciesconcvv` in `gcpy/benchmark/modules/benchmark_utils.py`
 - Function `copy_file_to_dir` in `gcpy/util.py`.  This is a wrapper for `shutil.copyfile`.
 - GitHub Action config file `.github/workflows/stale.yml`, which replaces StaleBot
-- Example script `gcpy/examples/hemco/make_mask_file.py`
 - Added `gcpy/community/format_hemco_data.py` from @hannahnesser
 - Added `gcpy/examples/hemco/format_hemco_demo.py` from @hannahnesser
 - Added HCl to `gcpy/benchmark/modules/emission_species.yml` and GTChlorine to `gcpy/benchmark/modules/emission_inventories.yml` for GEOS-Chem 14.4.0
+- Example script `gcpy/examples/working_with_files/make_mask_file.py`
+- Convenience function `replace_whitespace` in `gcpy/util.py`
+- Benchmark script `gcpy/benchmark/modules/benchmark_scrape_gcclassic_timers.py`
+- Benchmark script `gcpy/benchmark/modules/benchmark_scrape_gchp_timers.py`
+- 1-year benchmark scripts now produce GCC vs GCC and GCHP vs GCHP timing tables
+- Functions `gcc_vs_gcc_dirs`, `gchp_vs_gcc_dirs`, `gchp_vs_gchp_dirs`, and `get_log_filepaths` in `gcpy/benchmark/modules/benchmark_utils.py`
 
 ### Changed
 - Bump pip from 23.2.1 to 23.3 (dependabot suggested this)
@@ -45,6 +50,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Create radionuclide, STE flux, and mass conservation tables for Ref and Dev versions in TransportTracers benchmarks
 - Use new function `copy_file_to_dir` to copy the benchmark script and configuration file to the benchmark results folders
 - Updated GitHub stalebot config file `stale.yml` with new issue/PR labels that should not go stale
+- Updated benchmark driver scripts and config files to print GCClassic & GCHP timing information 
 
 ### Fixed
 - CS inquiry functions in `gcpy/cstools.py` now work properly for `xr.Dataset` and `xr.DataArray` objects

diff --git a/gcpy/benchmark/cloud/template.1hr_benchmark.yml b/gcpy/benchmark/cloud/template.1hr_benchmark.yml
@@ -47,13 +47,17 @@ data:
       dir: ref-gcc/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-07-01T01:00:00"
     gchp:
       version: ${GEOSCHEM_BENCHMARK_REF_PRIMARY_KEY}
       dir: ref-gchp/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-07-01T01:00:00"
       is_pre_14.0: False
@@ -64,13 +68,17 @@ data:
       dir: dev-gcc/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-07-01T01:00:00"
     gchp:
       version: ${GEOSCHEM_BENCHMARK_DEV_PRIMARY_KEY}
       dir: dev-gchp/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-07-01T01:00:00"
       is_pre_14.0: False
@@ -117,6 +125,7 @@ options:
     ops_budget_table: False
     OH_metrics: True
     ste_table: True # GCC only
+    timing_table: True
     summary_table: True
     plot_options:
       by_spc_cat: True

diff --git a/gcpy/benchmark/cloud/template.1mo_benchmark.yml b/gcpy/benchmark/cloud/template.1mo_benchmark.yml
@@ -47,13 +47,17 @@ data:
       dir: ref-gcc/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
     gchp:
       version: ${GEOSCHEM_BENCHMARK_REF_PRIMARY_KEY}
       dir: ref-gchp/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
       is_pre_14.0: False
@@ -64,13 +68,17 @@ data:
       dir: dev-gcc/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
     gchp:
       version: ${GEOSCHEM_BENCHMARK_DEV_PRIMARY_KEY}
       dir: dev-gchp/run-directory
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: runlog.txt
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
       is_pre_14.0: False
@@ -117,6 +125,7 @@ options:
     ops_budget_table: False
     OH_metrics: True
     ste_table: True # GCC only
+    timing_table: True
     summary_table: True
     plot_options:
       by_spc_cat: True

diff --git a/gcpy/benchmark/config/1mo_benchmark.yml b/gcpy/benchmark/config/1mo_benchmark.yml
@@ -47,13 +47,17 @@ data:
       dir: GCC_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: "GC.log"
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
     gchp:
       version: GCHP_ref
       dir: GCHP_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: "gchp.%Y%m%d_0000z.log"      
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
       is_pre_14.0: False
@@ -64,13 +68,17 @@ data:
       dir: GCC_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: .
+      logs_template: "GC.log"      
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
     gchp:
       version: GCHP_dev
       dir: GCHP_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "gchp.%Y%m%d_0000z.log"
       bmk_start: "2019-07-01T00:00:00"
       bmk_end: "2019-08-01T00:00:00"
       is_pre_14.0: False
@@ -117,6 +125,7 @@ options:
     ops_budget_table: False
     OH_metrics: True
     ste_table: True # GCC only
+    timing_table: True
     summary_table: True
     plot_options:
       by_spc_cat: True

diff --git a/gcpy/benchmark/config/1yr_fullchem_benchmark.yml b/gcpy/benchmark/config/1yr_fullchem_benchmark.yml
@@ -59,13 +59,17 @@ data:
       dir: GCC_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "log.%Y%m%d"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
     gchp:
       version: GCHP_ref
       dir: GCHP_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "gchp.%Y%m%d_0000z.log"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
       is_pre_14.0: False
@@ -76,13 +80,17 @@ data:
       dir: GCC_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "log.%Y%m%d"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
     gchp:
       version: GCHP_dev
       dir: GCHP_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "gchp.%Y%m%d_0000z.log"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
       is_pre_14.0: False
@@ -130,6 +138,7 @@ options:
     Ox_budget_table: True
     ste_table: True # GCC only
     OH_metrics: True
+    timing_table: True
     plot_models_vs_obs: True
     plot_options:
       by_spc_cat: True

diff --git a/gcpy/benchmark/config/1yr_tt_benchmark.yml b/gcpy/benchmark/config/1yr_tt_benchmark.yml
@@ -48,13 +48,17 @@ data:
       dir: GCC_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "log.%Y%m%d"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
     gchp:
       version: GCHP_ref
       dir: GCHP_ref
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "gchp.%Y%m%d_0000z.log"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
       is_pre_14.0: False
@@ -65,13 +69,17 @@ data:
       dir: GCC_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "log.%Y%m%d"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
     gchp:
       version: GCHP_dev
       dir: GCHP_dev
       outputs_subdir: OutputDir
       restarts_subdir: Restarts
+      logs_subdir: Logs
+      logs_template: "gchp.%Y%m%d_0000z.log"
       bmk_start: "2019-01-01T00:00:00"
       bmk_end: "2020-01-01T00:00:00"
       is_pre_14.0: False
@@ -115,6 +123,7 @@ options:
     mass_table: True
     ste_table: True
     cons_table: True
+    timing_table: False
   #
   # n_cores: Specify the number of cores to use.
   # -1: Use $OMP_NUM_THREADS         cores