ipydatagrid is slow compared to ipyaggrid

[mhouston2]

I have been a user of ipyaggrid for a bit and I like it a lot. Recently it started breaking when using newer versions of jupyter lab so I've looked into trying out ipydatagrid. However, ipydatagrid is running very slowly. I have a 6x52 sized dataframe with nothing fancy just text/numbers that is embedded within HBox and VBox widgets. There is almost a 1 second lag when I click a cell and it gets selected. This is all I have:

grid = DataGrid(df, layout={"height": '600px'}, selection_mode="cell") grid.auto_fit_columns = True grid.auto_fit_params = {"area": "body", "padding": 60}

Any thoughts?

[ibdafna]

Hi @mhouston2 and welcome to the ipydatagrid community!

Can you please share a full reproducible snippet including your pandas dataframe and the browser you're using?

Many thanks

[mhouston2]

Hi. I have a little more information. When preparing the reproducible snippet into a separate notebook for you I noticed it runs much faster in a new clean notebook. The notebook I had been running the code in previously had about 870 cells of code (I cleared all the output though so it's only regular code that remains.). I am running Firefox 95.0 on Linux. Would you still like to see the snippet? I haven't had this problem with grids before in the large notebook although it does take time to load the notebook initially.

[martinRenou]

Is this using JupyterLab or Jupyter Notebook?

The notebook I had been running the code in previously had about 870 cells of code

That seems like a lot of cells IMO. Splitting your Notebook into multiple Notebooks and Python files might help on this.

If you are using JupyterLab, there are some performance improvements concerning the Notebook rendering that should come soon. This might improve the use case of ~1000 cells Notebook.

[ibdafna]

Closing this but feel free to reopen if anything comes up!

[ankostis]

Please reopen, for 200.000x3 cells, it takes x3 (9.5sec vs 3.0sec) more than ipyaggrid:

>>> import pandas as pd
>>> import numpy as np
>>> import jupyter
>>> jupyter.__version__
'1.0.0'

>>> df=pd.DataFrame(np.random.rand(200_000,3), columns=list("ABC"))

>>> import ipydatagrid
>>> ipydatagrid.__version__
'1.1.15'

>>> %time dg = ipydatagrid.DataGrid(df)
CPU times: user 9.36 s, sys: 119 ms, total: 9.48 s
Wall time: 9.48 s
>>> %time display(dg)
CPU times: user 0 ns, sys: 3.04 ms, total: 3.04 ms
Wall time: 1.86 ms

>>> import ipyaggrid
>>> ipyaggrid.__version__
'0.3.2'

>>> dd, grid_options = ipyaggrid.util.Util.prepare_singleindex_df(df, {})
>>> %time ag = ipyaggrid.Grid(grid_data=df grid_options=grid_options)
CPU times: user 2.59 s, sys: 40 ms, total: 2.63 s
Wall time: 2.63 s
>>> %time display(ag)
CPU times: user 0 ns, sys: 2.08 ms, total: 2.08 ms
Wall time: 1.73 ms

[ankostis]

With more columns, 200,000 x 30, it takes x2 more (60sec vs 35sec).