Zarr benchmark

zarr_benchmark.py

"""
Benchmark for reading and writing zarr datasets
"""
import zarr
import numpy as np
from tqdm import tqdm
from time import time

num = 1000
filename = 'zarr_test'

# Write dataset
za = zarr.open(filename, 'w')
t0 = time()
for idx in tqdm(range(num)):
    data = np.expand_dims(np.random.random((512, 512)), axis=0)
    if idx == 0:
        za.array(name='blob', data=data)
    else:
        za.blob.append(data)
t1 = time()
print(f'Write duration: {(t1-t0):.2f} s')
print(f'Write throughput: {num/(t1-t0):.2f} frames/s')

# Read dataset
za = zarr.open(filename, 'r')
t0 = time()
for idx in tqdm(range(num)):
    data = za.blob[idx]
t1 = time()
print(f'Read duration: {(t1-t0):.2f} s')
print(f'Read throughput: {num/(t1-t0):.2f} frames/s')