Merge branch 'main' into submission-run

.isort.cfg

@@ -1,2 +1,2 @@
 [settings]
-known_third_party = fsspec,h5py,pandas,pytest,s3fs,seaborn,tifffile,zarr
+known_third_party = fsspec,h5py,matplotlib,numpy,pandas,ptitprince,pytest,requests,s3fs,seaborn,tifffile,zarr

.pre-commit-config.yaml

@@ -13,10 +13,10 @@ repos:
         args: ["--profile", "black"]
 
   - repo: https://github.com/ambv/black
-    rev: 19.10b0
+    rev: 20.8b1
     hooks:
       - id: black
-        args: [--target-version=py36]
+        args: [--target-version=py37]
 
   - repo: https://github.com/asottile/pyupgrade
     rev: v2.7.2
@@ -45,7 +45,7 @@ repos:
           - --autofix
 
   - repo: https://gitlab.com/pycqa/flake8
-    rev: 3.8.3
+    rev: 3.9.0
     hooks:
       - id: flake8
         additional_dependencies: [
@@ -58,7 +58,7 @@ repos:
           # default black line length is 88
           "--max-line-length=88",
           # Conflicts with black: E203 whitespace before ':'
-          "--ignore=E203",
+          "--ignore=E203,W503",
           "--rst-roles=class,func,ref,module,const",
         ]
 

README.md

@@ -1,4 +1,8 @@
-# How to use this repository
+[![DOI](https://zenodo.org/badge/329595844.svg)](https://zenodo.org/badge/latestdoi/329595844)
+[![Binder](https://mybinder.org/badge_logo.svg)](https://mybinder.org/v2/gh/ome/ngff-latency-benchmark/master?filepath=notebooks)
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ome/ngff-latency-benchmark/)
+
+# OME-NGFF: latency benchmark
 
 To get started, clone this repository locally:
 ```
@@ -38,3 +42,4 @@ docker-compose run --rm benchmark -sv
 
 This will store both the benchmarking results ('benchmark_data.json') as well as a plotted graph
 ('benchmark_plot.png') in the directory along with the input data.
+

benchmark/Dockerfile

@@ -13,5 +13,6 @@ COPY plot_results.py /benchmark/plot_results.py
 RUN conda run -n benchmark pip install pytest-profiling
 RUN conda run -n benchmark pip install seaborn
 RUN conda run -n benchmark pip install pytest-csv
+RUN conda run -n benchmark pip install ptitprince
 
 ENTRYPOINT ["conda", "run", "--no-capture-output", "-n", "benchmark", "bash", "/benchmark/benchmark.sh"]

benchmark/benchmark.sh

@@ -2,7 +2,6 @@
 
 DIR=${DIR:-./data}
 export BENCHMARK_DATA=${DIR}
-export BENCHMARK_PLOT=${DIR}/benchmark_plot.png
 
 set -e
 set -u
@@ -15,6 +14,5 @@ do
     pytest benchmark.py "$@" --csv=${BENCHMARK_DATA}/${i}_benchmark_data.csv \
         --csv-columns status,duration,properties_as_columns,markers_args_as_columns
     head ${BENCHMARK_DATA}/${i}_benchmark_data.csv
+    python plot_results.py ${BENCHMARK_DATA}/${i}_benchmark_data.csv ${BENCHMARK_DATA}/${i}_benchmark_plot.png
 done
-
-python plot_results.py

benchmark/plot_results.py

@@ -1,73 +1,45 @@
-import json
-import os
-import re
+#!/usr/bin/env python
+import argparse
 
+import matplotlib.pyplot as plt
 import pandas as pd
-import seaborn as sns
-
-data_path = os.environ.get("BENCHMARK_DATA", "benchmark_data")
-plot_path = os.environ.get("BENCHMARK_PLOT", "benchmark_plot.png")
-xy = os.environ.get("XY", "unknown")
-
-csv = pd.read_csv(f"{data_path}/0_benchmark_data.csv")
-
-types = ("Overhead", "Zarr", "TIFF", "HDF5")
-sources = ("local", "http", "s3")
-orders = {"type": types, "source": sources}
-
-pal_points = "colorblind"
-pal_violins = "pastel"
-
-g = sns.FacetGrid(
-    csv,
-    col="source",
-    col_order=sources,
-    sharey=False,
-    height=5,
-    aspect=0.6,
-)
-
-g = g.map(
-    sns.boxenplot,
-    "type",
-    "seconds",
-    order=types,
-    width=0.6,
-    k_depth=2,
-    palette=pal_violins,
-    dodge=True,
-    showfliers=False,
-)
-
-g = g.map(
-    sns.stripplot,
-    "type", # was type
-    "seconds",
-    dodge=True,
-    order=types,
-    jitter=0.2,
-    size=3,
-    palette=pal_points,
-)
-
-g.despine(left=True)
-g.set(yscale ='log', ylim=(0.0009, 1))
-
-# Set axis labels & ticks #
-for ax in g.fig.get_axes():
-    label = ax.get_title().replace("source =", "")
-    ax.set_xlabel(label)
-    ax.set_xticklabels(types)
-    ax.set_title("")
-    for col in ax.collections:
-        # Remove outline of violins
-        col.set_edgecolor("white")
-
-# Only show on far left plot
-g.fig.get_axes()[0].set_ylabel("Seconds")
-g.fig.get_axes()[0].spines["left"].set_visible(True)
-
-# Add annotations
-g.fig.get_axes()[0].text(0.0001, 0.001, "off-\nscale")
-
-g.savefig(plot_path, dpi=600)
+import ptitprince as pt
+
+
+def plot_csv(filename: str, pngpath: str):
+    csv = pd.read_csv(filename)
+
+    f, ax = plt.subplots(figsize=(8, 6))
+    ax = pt.RainCloud(
+        x="type",
+        y="seconds",
+        hue="source",
+        data=csv,
+        palette="Set2",
+        # bw = .2,
+        width_viol=0.6,
+        ax=ax,
+        orient="h",
+        alpha=0.65,
+        # dodge = True,
+        jitter=0.03,
+        move=0.2,
+        # pointplot = True,
+    )
+
+    # ax.set(ylim=(0.0002, 5))
+    ax.set_xscale("log")
+    ax.set_xlabel("seconds per chunk")
+    ax.axes.get_yaxis().get_label().set_visible(False)
+    handles, labels = ax.get_legend_handles_labels()
+    plt.legend(handles[0:3], labels[0:3], loc="lower left")
+    plt.tight_layout()
+    f.savefig(pngpath, dpi=600)
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("csv", help="input csv file")
+    parser.add_argument("png", help="output png file")
+    ns = parser.parse_args()
+    plot_csv(ns.csv, ns.png)

binder/environment.yml

@@ -0,0 +1,13 @@
+channels:
+  - conda-forge
+dependencies:
+  - h5py
+  - matplotlib
+  - numpy
+  - pandas
+  - pytest
+  - request
+  - s3fs
+  - seaborn
+  - tifffile
+  - zarr