html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/pull/5922#issuecomment-963540277,https://api.github.com/repos/pydata/xarray/issues/5922,963540277,IC_kwDOAMm_X845bnU1,14371165,2021-11-08T20:20:55Z,2021-11-08T20:20:55Z,MEMBER,When does https://pandas.pydata.org/speed/xarray/#/ update by the way? I was thinking the pandas and dask groupby might be interesting there.,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1040185743
https://github.com/pydata/xarray/pull/5922#issuecomment-962495898,https://api.github.com/repos/pydata/xarray/issues/5922,962495898,IC_kwDOAMm_X845XoWa,14371165,2021-11-06T19:03:15Z,2021-11-06T19:03:15Z,MEMBER,"@dcherian What do you think of this? Had to reduce the values quite a bit to get it decently fast, hopefully it still shows the relevant parts.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1040185743
https://github.com/pydata/xarray/pull/5922#issuecomment-962477227,https://api.github.com/repos/pydata/xarray/issues/5922,962477227,IC_kwDOAMm_X845Xjyr,14371165,2021-11-06T16:40:59Z,2021-11-06T18:07:47Z,MEMBER,"Notes
* 3df1015 - 28m 9s, original
* 445312d - 25m 19s, ~x0.5 values
* c56dd94 - 9m 50s, ~x0.2 values
* a89f62b - 8m 42s, remove pandas and dask dataframe tests from ci","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1040185743
https://github.com/pydata/xarray/pull/5922#issuecomment-955711837,https://api.github.com/repos/pydata/xarray/issues/5922,955711837,IC_kwDOAMm_X8449wFd,14371165,2021-10-31T14:34:34Z,2021-10-31T14:34:34Z,MEMBER,"I checked the timing in #5796, turns out they we're (of course) as slow... The total time was probably fine though since groupby had so few tests at that point but now with a few more it scales quite a bit.
I think we have to reduce the size of the datasets and maybe increase the resampling points a little. The tests with 5s are major bottlenecks, we should try and get it down to max 100-500ms in my opinion.
We can increase the numbers again if the `numpy_groupies` pr improves the speed significantly.
If you temporarily remove the other asv-files we can get a better feel how long the groupby file takes and it's also a little easier to read the report when iterating like this.
It's also curious how `mean` is faster for some of these, I don't understand that at all.
```
[ 55.91%] ··· groupby.GroupBy.time_agg_large_num_groups ok
[ 55.91%] ··· ======== ========== ==========
-- ndim
-------- ---------------------
method 1 2
======== ========== ==========
sum 577±30ms 880±40ms
mean 649±40ms 955±60ms
======== ========== ==========
[ 56.01%] ··· groupby.GroupBy.time_agg_small_num_groups ok
[ 56.01%] ··· ======== ========== ==========
-- ndim
-------- ---------------------
method 1 2
======== ========== ==========
sum 279±20ms 445±20ms
mean 325±20ms 483±30ms
======== ========== ==========
[ 56.10%] ··· groupby.GroupBy.time_init ok
[ 56.10%] ··· ====== =============
ndim
------ -------------
1 1.08±0.09ms
2 6.45±0.5ms
====== =============
[ 56.20%] ··· groupby.GroupByDask.time_agg_large_num_groups ok
[ 56.20%] ··· ======== ============ ============
-- ndim
-------- -------------------------
method 1 2
======== ============ ============
sum 2.27±0.09s 5.57±0.1s
mean 1.45±0.03s 4.68±0.05s
======== ============ ============
[ 56.30%] ··· groupby.GroupByDask.time_agg_small_num_groups ok
[ 56.30%] ··· ======== ============ ============
-- ndim
-------- -------------------------
method 1 2
======== ============ ============
sum 2.14±0.02s 5.27±0.05s
mean 1.40±0.02s 4.62±0.02s
======== ============ ============
[ 56.40%] ··· groupby.GroupByDask.time_init ok
[ 56.40%] ··· ====== ============
ndim
------ ------------
1 5.46±0.1ms
2 31.1±2ms
====== ============
[ 56.49%] ··· ...oupByDaskDataFrame.time_agg_large_num_groups ok
[ 56.49%] ··· ======== ============= ==========
-- ndim
-------- ------------------------
method 1 2
======== ============= ==========
sum 1.36±0.07ms 829±20ms
mean 1.10±0.07ms 903±20ms
======== ============= ==========
[ 56.59%] ··· ...oupByDaskDataFrame.time_agg_small_num_groups ok
[ 56.59%] ··· ======== ============= ==========
-- ndim
-------- ------------------------
method 1 2
======== ============= ==========
sum 1.32±0.02ms 415±10ms
mean 1.06±0.02ms 468±9ms
======== ============= ==========
[ 56.69%] ··· groupby.GroupByDaskDataFrame.time_init ok
[ 56.69%] ··· ====== =============
ndim
------ -------------
1 36.8±2μs
2 5.98±0.09ms
====== =============
[ 56.78%] ··· ...y.GroupByDataFrame.time_agg_large_num_groups ok
[ 56.78%] ··· ======== ============= ==========
-- ndim
-------- ------------------------
method 1 2
======== ============= ==========
sum 1.39±0.02ms 855±20ms
mean 1.14±0.07ms 907±10ms
======== ============= ==========
[ 56.88%] ··· ...y.GroupByDataFrame.time_agg_small_num_groups ok
[ 56.88%] ··· ======== ============= ==========
-- ndim
-------- ------------------------
method 1 2
======== ============= ==========
sum 1.32±0.05ms 423±10ms
mean 1.03±0.02ms 455±6ms
======== ============= ==========
[ 56.98%] ··· groupby.GroupByDataFrame.time_init ok
[ 56.98%] ··· ====== =============
ndim
------ -------------
1 33.0±0.5μs
2 5.72±0.08ms
====== =============
[ 57.07%] ··· groupby.Resample.time_agg_large_num_groups ok
[ 57.07%] ··· ======== ========== ==========
-- ndim
-------- ---------------------
method 1 2
======== ========== ==========
sum 623±10ms 712±10ms
mean 685±4ms 910±9ms
======== ========== ==========
[ 57.17%] ··· groupby.Resample.time_agg_small_num_groups ok
[ 57.17%] ··· ======== ============ ============
-- ndim
-------- -------------------------
method 1 2
======== ============ ============
sum 6.05±0.2ms 7.22±0.3ms
mean 6.24±0.2ms 8.34±0.2ms
======== ============ ============
[ 57.27%] ··· groupby.Resample.time_init ok
[ 57.27%] ··· ====== =============
ndim
------ -------------
1 2.21±0.08ms
2 2.13±0.08ms
====== =============
[ 57.36%] ··· groupby.ResampleDask.time_agg_large_num_groups ok
[ 57.36%] ··· ======== ============ ============
-- ndim
-------- -------------------------
method 1 2
======== ============ ============
sum 4.98±0.02s 5.33±0.03s
mean 2.90±0.02s 3.18±0.01s
======== ============ ============
[ 57.46%] ··· groupby.ResampleDask.time_agg_small_num_groups ok
[ 57.46%] ··· ======== ============ ============
-- ndim
-------- -------------------------
method 1 2
======== ============ ============
sum 25.3±0.2ms 28.5±1ms
mean 18.3±0.5ms 20.3±0.7ms
======== ============ ============
[ 57.56%] ··· groupby.ResampleDask.time_init ok
[ 57.56%] ··· ====== ============
ndim
------ ------------
1 2.19±0.1ms
2 2.20±0.1ms
====== ============
```
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1040185743
https://github.com/pydata/xarray/pull/5922#issuecomment-955537002,https://api.github.com/repos/pydata/xarray/issues/5922,955537002,IC_kwDOAMm_X8449FZq,14371165,2021-10-30T17:33:11Z,2021-10-30T17:33:11Z,MEMBER,"These additions seems to double the total testing time. Won't we learn the same thing with a tenth of the values?
Did you run asv locally by the way? Is this fast on your own pc?","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1040185743