Merge branch 'main' into parallel-concatenate

.github/workflows/benchmarks_run.yml

@@ -71,9 +71,9 @@ jobs:
  with:
  fetch-depth: 0
 
- - name: Install ASV & Nox
+ - name: Install Nox
  run: |
- pip install asv nox
+ pip install nox
 
  - name: Cache environment directories
  id: cache-env-dir
@@ -112,7 +112,7 @@ jobs:
  GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
  PR_NUMBER: ${{ github.event.number }}
  run: |
- python benchmarks/bm_runner.py branch origin/${{ github.base_ref }}
+ nox -s benchmarks -- branch origin/${{ github.base_ref }}
 
  - name: Run overnight benchmarks
  id: overnight
@@ -128,7 +128,7 @@ jobs:
  
  if [ "$first_commit" != "" ]
  then
- python benchmarks/bm_runner.py overnight $first_commit
+ nox -s benchmarks -- overnight $first_commit
  fi
 
  - name: Warn of failure

.github/workflows/ci-manifest.yml

@@ -23,4 +23,4 @@ concurrency:
 jobs:
  manifest:
  name: "check-manifest"
- uses: scitools/workflows/.github/workflows/ci-manifest.yml@2024.05.0
+ uses: scitools/workflows/.github/workflows/ci-manifest.yml@2024.06.0

.github/workflows/refresh-lockfiles.yml

@@ -14,5 +14,5 @@ on:
 
 jobs:
  refresh_lockfiles:
- uses: scitools/workflows/.github/workflows/refresh-lockfiles.yml@2024.05.0
+ uses: scitools/workflows/.github/workflows/refresh-lockfiles.yml@2024.06.0
  secrets: inherit

.pre-commit-config.yaml

@@ -29,7 +29,7 @@ repos:
  - id: no-commit-to-branch
 
 - repo: https://github.com/astral-sh/ruff-pre-commit
- rev: "v0.4.4"
+ rev: "v0.4.7"
  hooks:
  - id: ruff
  types: [file, python]
@@ -38,7 +38,7 @@ repos:
  types: [file, python]
 
 - repo: https://github.com/codespell-project/codespell
- rev: "v2.2.6"
+ rev: "v2.3.0"
  hooks:
  - id: codespell
  types_or: [asciidoc, python, markdown, rst]

benchmarks/README.md

@@ -20,13 +20,13 @@ the PR's base branch, thus showing performance differences introduced
 by the PR. (This run is managed by 
 [the aforementioned GitHub Action](../.github/workflows/benchmark.yml)).
 
-`asv ...` commands must be run from this directory. You will need to have ASV
-installed, as well as Nox (see
-[Benchmark environments](#benchmark-environments)).
-
-The benchmark runner ([bm_runner.py](./bm_runner.py)) provides conveniences for
+To run locally: the **benchmark runner** provides conveniences for
 common benchmark setup and run tasks, including replicating the automated 
-overnight run locally. See `python bm_runner.py --help` for detail.
+overnight run locally. This is accessed via the Nox `benchmarks` session - see
+`nox -s benchmarks -- --help` for detail (_see also: 
+[bm_runner.py](./bm_runner.py)_). Alternatively you can directly run `asv ...`
+commands from this directory (you will still need Nox installed - see
+[Benchmark environments](#benchmark-environments)).
 
 A significant portion of benchmark run time is environment management. Run-time
 can be reduced by placing the benchmark environment on the same file system as
@@ -43,11 +43,17 @@ if it is not already. You can achieve this by either:
 
 * `OVERRIDE_TEST_DATA_REPOSITORY` - required - some benchmarks use
 `iris-test-data` content, and your local `site.cfg` is not available for
-benchmark scripts.
+benchmark scripts. The benchmark runner defers to any value already set in
+the shell, but will otherwise download `iris-test-data` and set the variable
+accordingly.
 * `DATA_GEN_PYTHON` - required - path to a Python executable that can be
 used to generate benchmark test objects/files; see
 [Data generation](#data-generation). The benchmark runner sets this 
-automatically, but will defer to any value already set in the shell.
+automatically, but will defer to any value already set in the shell. Note that
+[Mule](https://github.com/metomi/mule) will be automatically installed into 
+this environment, and sometimes 
+[iris-test-data](https://github.com/SciTools/iris-test-data) (see 
+`OVERRIDE_TEST_DATA_REPOSITORY`).
 * `BENCHMARK_DATA` - optional - path to a directory for benchmark synthetic
 test data, which the benchmark scripts will create if it doesn't already
 exist. Defaults to `<root>/benchmarks/.data/` if not set. Note that some of

benchmarks/benchmarks/__init__.py

@@ -5,7 +5,9 @@
 """Common code for benchmarks."""
 
 from os import environ
-import resource
+import tracemalloc
+
+import numpy as np
 
 
 def disable_repeat_between_setup(benchmark_object):
@@ -61,27 +63,34 @@ class TrackAddedMemoryAllocation:
  AVD's detection threshold and be treated as 'signal'. Results
  smaller than this value will therefore be returned as equal to this
  value, ensuring fractionally small noise / no noise at all.
+ Defaults to 1.0
+
+ RESULT_ROUND_DP : int
+ Number of decimal places of rounding on result values (in Mb).
+ Defaults to 1
 
  """
 
- RESULT_MINIMUM_MB = 5.0
-
- @staticmethod
- def process_resident_memory_mb():
- return resource.getrusage(resource.RUSAGE_SELF).ru_maxrss / 1024.0
+ RESULT_MINIMUM_MB = 0.2
+ RESULT_ROUND_DP = 1 # I.E. to nearest 0.1 Mb
 
  def __enter__(self):
- self.mb_before = self.process_resident_memory_mb()
+ tracemalloc.start()
  return self
 
  def __exit__(self, *_):
- self.mb_after = self.process_resident_memory_mb()
+ _, peak_mem_bytes = tracemalloc.get_traced_memory()
+ tracemalloc.stop()
+ # Save peak-memory allocation, scaled from bytes to Mb.
+ self._peak_mb = peak_mem_bytes * (2.0**-20)
 
  def addedmem_mb(self):
  """Return measured memory growth, in Mb."""
- result = self.mb_after - self.mb_before
+ result = self._peak_mb
  # Small results are too vulnerable to noise being interpreted as signal.
  result = max(self.RESULT_MINIMUM_MB, result)
+ # Rounding makes results easier to read.
+ result = np.round(result, self.RESULT_ROUND_DP)
  return result
 
  @staticmethod
@@ -105,6 +114,33 @@ def _wrapper(*args, **kwargs):
  decorated_func.unit = "Mb"
  return _wrapper
 
+ @staticmethod
+ def decorator_repeating(repeats=3):
+ """Benchmark to track growth in resident memory during execution.
+
+ Tracks memory for repeated calls of decorated function.
+
+ Intended for use on ASV ``track_`` benchmarks. Applies the
+ :class:`TrackAddedMemoryAllocation` context manager to the benchmark
+ code, sets the benchmark ``unit`` attribute to ``Mb``.
+
+ """
+
+ def decorator(decorated_func):
+ def _wrapper(*args, **kwargs):
+ assert decorated_func.__name__[:6] == "track_"
+ # Run the decorated benchmark within the added memory context
+ # manager.
+ with TrackAddedMemoryAllocation() as mb:
+ for _ in range(repeats):
+ decorated_func(*args, **kwargs)
+ return mb.addedmem_mb()
+
+ decorated_func.unit = "Mb"
+ return _wrapper
+
+ return decorator
+
 
 def on_demand_benchmark(benchmark_object):
  """Disable these benchmark(s) unless ON_DEMAND_BENCHARKS env var is set.