df.take is much slower against pandas

[YarShev]

On a machine with 192 CPUs.

# import pandas as pd
import modin.pandas as pd
import numpy as np
import time

df = pd.DataFrame(data=np.random.randint(99999, 99999999, size=(100000000,4)),
                columns=['C1','C2','C3','C4'])

to_take = np.random.randint(0, 100000000, size=80000000)
t0 = time.time()
df.take(to_take, axis=0)
t1 = time.time()
print('time to take: ', t1 - t0)
# time for take:  44.71011233329773 in Modin
# time for take:  3.082368850708008 in pandas

[YarShev]

cc @dchigarev

[YarShev]

# import pandas as pd
import modin.pandas as pd
import numpy as np
import time

df = pd.DataFrame(data=np.random.randint(99999, 99999999, size=(100000000,1)),
                columns=['C1']).squeeze(axis=1)

to_take = np.random.randint(0, 100000000, size=80000000)
t0 = time.time()
df.take(to_take, axis=0)
t1 = time.time()
print('time to take: ', t1 - t0)
# time for take:  37.6995530128479 in Modin
# time for take:  1.492713212966919 in pandas