dt.truncate is 3-4x slower in polars compared to pandas

[Chuck321123]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

import pandas as pd
import polars as pl

num_rows = 1000000
utc_time = pd.date_range(start='2023-01-01', periods=num_rows, freq='min')

# Create the DataFrame
df = pd.DataFrame({
    'UTC_Time': utc_time
})

df['UTC_Time'] = df['UTC_Time'].sort_values()

# Display the first few rows
print(df.head())

Interval = 2

%timeit df["UTC_Time"].dt.floor(f'{Interval}h')

df = pl.DataFrame(df)

%timeit df.with_columns(pl.col("UTC_Time").dt.truncate(f"{Interval}h"))

Log output

8.7 ms ± 1.38 ms per loop (mean ± std. dev. of 7 runs, 100 loops each)
32.1 ms ± 915 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)

Issue description

So in my testings I get that truncate is way slower than the pandas equivalent. Do you guys also get the same results? In that case, it would be nice if someone could find a way to speed up this function.

Expected behavior

That dt.truncate becomes faster than the pandas equivalent

Installed versions

--------Version info---------
Polars:               0.20.30
Index type:           UInt32
Platform:             Windows-11-10.0.22631-SP0
Python:               3.12.2 | packaged by Anaconda, Inc. | (main, Feb 27 2024, 17:28:07) [MSC v.1916 64 bit (AMD64)]

----Optional dependencies----
adbc_driver_manager:  <not installed>
cloudpickle:          3.0.0
connectorx:           <not installed>
deltalake:            <not installed>
fastexcel:            <not installed>
fsspec:               <not installed>
gevent:               <not installed>
hvplot:               <not installed>
matplotlib:           3.8.3
nest_asyncio:         1.6.0
numpy:                1.26.4
openpyxl:             3.1.2
pandas:               2.2.1
pyarrow:              15.0.2
pydantic:             <not installed>
pyiceberg:            <not installed>
pyxlsb:               <not installed>
sqlalchemy:           <not installed>
torch:                <not installed>
xlsx2csv:             <not installed>
xlsxwriter:           <not installed>```

</details>

[cmdlineluser]

I see a similar difference.

This particular operation seems to run faster on the streaming engine.

(df.lazy()
   .with_columns(pl.col("UTC_Time").dt.truncate(f"{Interval}h"))
   .collect(streaming=True)
)

[MarcoGorelli]

thanks for the report! for a quick look at the code, nothing jumps out at me (like it did in the offset_by one) - time to investigate

[MarcoGorelli]

I might have a fix, just testing it out to be sure