PERF: Regressions since v0.21 #18532
Labels
Performance
Memory or execution speed performance
Comments
|
@mroeschke if you want to re-run and update would be appreciated |
|
Sure thing. Will try to get to it by the end of the week. |
(edited by Tom to put the speedups in a details section) |
|
#18496 for the Series constructor. PR incoming. |
|
cc @jbrockmendel for thoughts on the timestamp ones |
|
#18164 made |
|
Yeah, in that case not a big deal I think. |
|
Looking briefly at the |
|
Having trouble reproducing the |
I see it clearly in the terminal (I suppose notebook repr takes a different path) with |
|
The slowing in If that's the case the slowdown should be expected? |
|
@mroeschke would you mind rerunning and changing the top of the PR here. |
|
Updated the regressions in my top comment |
|
Hmm looks like a lot of regressions. I can take a look at the GroupBy stuff over the next few days. |
|
Are these consistent across runs? IIRC correctly asv's use of the word "SIGNIFICANTLY" does not refer to statistical significance. |
|
I can run the suite one more time on my machine. |
|
Reminder: we also have http://pandas.pydata.org/speed/pandas/ |
|
Ahh, the benchmarks there look out of date :/ Looking into it now. |
|
@TomAugspurger I get the feeling the web portal may not be showing all of the benchmarks, at the very least those that are parametrized. For instance, it only shows two benchmarks from the groupby.GroupByMethods class, though I think the combinations of parameters there should generate 264 benchmarks
Any idea where to even start looking at that? |
|
I'm looking into it. When pandas added the HTML results to the .gitignore
we stopped updating. We at least have the data though.
…On Wed, Oct 24, 2018 at 11:40 AM William Ayd ***@***.***> wrote:
@TomAugspurger <https://github.com/TomAugspurger> I get the feeling the
web portal may not be showing all of the benchmarks, at the very least
those that are parametrized. For instance, it only shows two benchmarks
from the groupby.GroupByMethods class, though I think the combinations of
parameters there should generate 264 benchmarks
[image: image]
<https://user-images.githubusercontent.com/609873/47446765-c9fe5f00-d770-11e8-8437-5f1d4618541b.png>
Any idea where to even start looking at that?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#18532 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABQHIt0uMg2fTr8UeTSPLWFMHlOh7QBoks5uoJgEgaJpZM4QsZ91>
.
|
|
No luck yet. I'll need to do more debugging later.
On Wed, Oct 24, 2018 at 12:13 PM Tom Augspurger <tom.augspurger88@gmail.com>
wrote:
… I'm looking into it. When pandas added the HTML results to the .gitignore
we stopped updating. We at least have the data though.
On Wed, Oct 24, 2018 at 11:40 AM William Ayd ***@***.***>
wrote:
> @TomAugspurger <https://github.com/TomAugspurger> I get the feeling the
> web portal may not be showing all of the benchmarks, at the very least
> those that are parametrized. For instance, it only shows two benchmarks
> from the groupby.GroupByMethods class, though I think the combinations of
> parameters there should generate 264 benchmarks
>
> [image: image]
> <https://user-images.githubusercontent.com/609873/47446765-c9fe5f00-d770-11e8-8437-5f1d4618541b.png>
>
> Any idea where to even start looking at that?
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#18532 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/ABQHIt0uMg2fTr8UeTSPLWFMHlOh7QBoks5uoJgEgaJpZM4QsZ91>
> .
>
|
|
Here's the second run for the interested: At a high level there's consistency between runs. Regressions Speedups |
|
Short update, seems like we do *not* have data for this time period on the
benchmark machine. At some point, the environment became corrupted,
so that pandas failed to build. However, the Airflow process running these
didn't fail, so I didn't notice.
I've purged that env, and am re-running the benchmarks from the last few
commits. Those should finish in a few hours. I'll try to manually kick
of some runs for older commits, to fill in the gaps.
…On Wed, Oct 24, 2018 at 7:25 PM Matthew Roeschke ***@***.***> wrote:
Here's the second run for the interested:
At a high level there's consistency between runs.
Regressions
+ 55.9±10μs 1.52±0.04s 27224.82 indexing.IntervalIndexing.time_loc_list
+ 85.7±20μs 1.57±0.2s 18312.55 indexing.IntervalIndexing.time_getitem_list
+ 14.9±0.2μs 1.36±0.03ms 91.71 categoricals.CategoricalSlicing.time_getitem_bool_array('monotonic_decr')
+ 36.0±1ms 2.06±0.01s 57.32 offset.ApplyIndex.time_apply_index(<BusinessDay>)
+ 449±4ns 25.2±0.6μs 56.03 timestamp.TimestampProperties.time_weekday_name(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, None)
+ 455±6ns 25.4±2μs 55.83 timestamp.TimestampProperties.time_weekday_name(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 24.3±1ms 1.35±0.5s 55.59 period.DataFramePeriodColumn.time_setitem_period_column
+ 39.1±0.8ms 2.07±0.07s 52.89 offset.ApplyIndex.time_apply_index(<SemiMonthBegin: day_of_month=15>)
+ 39.5±1ms 2.05±0.04s 51.94 offset.ApplyIndex.time_apply_index(<SemiMonthEnd: day_of_month=15>)
+ 4.87±0.07ms 208±10ms 42.66 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<BusinessDay>)
+ 5.27±0.2ms 207±10ms 39.38 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<SemiMonthBegin: day_of_month=15>)
+ 5.43±0.1ms 209±10ms 38.40 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<SemiMonthEnd: day_of_month=15>)
+ 5.61±0.1ms 210±6ms 37.39 offset.OffsetSeriesArithmetic.time_add_offset(<BusinessDay>)
+ 15.4±0.4ms 549±20ms 35.71 timeseries.Iteration.time_iter_preexit(<function period_range at 0x11284df28>)
+ 6.35±0.2ms 217±10ms 34.12 offset.OffsetSeriesArithmetic.time_add_offset(<SemiMonthEnd: day_of_month=15>)
+ 6.33±0.2ms 213±10ms 33.64 offset.OffsetSeriesArithmetic.time_add_offset(<SemiMonthBegin: day_of_month=15>)
+ 389±7ns 12.9±0.8μs 33.27 indexing.MethodLookup.time_lookup_ix
+ 3.73±0.09ms 113±5ms 30.40 period.PeriodIndexConstructor.time_from_pydatetime('D')
+ 1.78±0.04ms 54.0±2ms 30.28 indexing.CategoricalIndexIndexing.time_get_indexer_list('monotonic_decr')
+ 453±8ns 9.84±0.7μs 21.72 timestamp.TimestampProperties.time_weekday_name(None, 'B')
+ 447±8ns 8.54±0.04μs 19.09 timestamp.TimestampProperties.time_weekday_name(None, None)
+ 5.25±0.1ms 99.7±0.9ms 18.99 timeseries.DatetimeIndex.time_timeseries_is_month_start('tz_aware')
+ 9.48±0.08ms 175±2ms 18.50 multiindex_object.Values.time_datetime_level_values_copy
+ 7.32±0.3μs 120±4μs 16.40 period.Indexing.time_get_loc
+ 6.64±0.07μs 69.0±1μs 10.39 period.Indexing.time_shallow_copy
+ 7.54±0.5ms 76.5±10ms 10.14 frame_methods.Repr.time_frame_repr_wide
+ 78.8±7ms 696±10ms 8.83 plotting.TimeseriesPlotting.time_plot_regular
+ 23.4±0.6ms 191±3ms 8.16 binary_ops.Ops2.time_frame_float_floor_by_zero
+ 7.71±0.1μs 60.1±5μs 7.80 index_object.Indexing.time_slice('Int')
+ 7.81±0.3μs 60.9±2μs 7.79 index_object.Indexing.time_slice_step('Int')
+ 83.5±20μs 587±20μs 7.03 groupby.GroupByMethods.time_dtype_as_field('float', 'cummin', 'transformation')
+ 82.7±4μs 580±50μs 7.01 groupby.GroupByMethods.time_dtype_as_field('datetime', 'cummin', 'transformation')
+ 83.2±4μs 580±40μs 6.97 groupby.GroupByMethods.time_dtype_as_field('datetime', 'cummin', 'direct')
+ 87.5±10μs 588±30μs 6.72 groupby.GroupByMethods.time_dtype_as_field('float', 'cummax', 'transformation')
+ 19.5±0.4μs 130±0.9μs 6.66 period.PeriodUnaryMethods.time_now('M')
+ 84.4±10μs 559±20μs 6.62 groupby.GroupByMethods.time_dtype_as_field('float', 'cummin', 'direct')
+ 85.7±9μs 567±30μs 6.62 groupby.GroupByMethods.time_dtype_as_field('float', 'cummax', 'direct')
+ 92.4±10μs 572±20μs 6.20 groupby.GroupByMethods.time_dtype_as_field('float', 'cumprod', 'direct')
+ 18.6±0.4ms 114±9ms 6.10 frame_methods.Dropna.time_dropna('any', 1)
+ 18.2±0.3ms 107±3ms 5.85 frame_methods.Dropna.time_dropna('any', 0)
+ 99.3±30μs 573±20μs 5.77 groupby.GroupByMethods.time_dtype_as_field('float', 'cumprod', 'transformation')
+ 97.4±6μs 559±20μs 5.74 groupby.GroupByMethods.time_dtype_as_field('float', 'cumsum', 'direct')
+ 96.6±5μs 546±70μs 5.65 groupby.GroupByMethods.time_dtype_as_field('float', 'cumsum', 'transformation')
+ 116±5μs 636±90μs 5.48 groupby.GroupByMethods.time_dtype_as_group('float', 'cummin', 'transformation')
+ 33.4±0.6μs 182±1μs 5.43 period.PeriodUnaryMethods.time_asfreq('M')
+ 122±5μs 660±50μs 5.39 groupby.GroupByMethods.time_dtype_as_group('int', 'cummax', 'direct')
+ 34.0±0.6μs 180±1μs 5.31 period.PeriodUnaryMethods.time_asfreq('min')
+ 125±10μs 660±30μs 5.27 groupby.GroupByMethods.time_dtype_as_group('int', 'cummin', 'direct')
+ 128±10μs 672±30μs 5.26 groupby.GroupByMethods.time_dtype_as_field('int', 'cummax', 'transformation')
+ 69.7±0.6μs 358±10μs 5.14 period.PeriodProperties.time_property('min', 'end_time')
+ 121±20μs 622±30μs 5.14 groupby.GroupByMethods.time_dtype_as_group('int', 'cummin', 'transformation')
+ 111±2μs 571±4μs 5.13 groupby.GroupByMethods.time_dtype_as_group('datetime', 'cummin', 'direct')
+ 125±7μs 635±20μs 5.10 groupby.GroupByMethods.time_dtype_as_group('int', 'cummax', 'transformation')
+ 127±5μs 646±30μs 5.09 groupby.GroupByMethods.time_dtype_as_field('int', 'cummax', 'direct')
+ 112±4μs 571±3μs 5.09 groupby.GroupByMethods.time_dtype_as_group('datetime', 'cummin', 'transformation')
+ 118±30μs 596±20μs 5.03 groupby.GroupByMethods.time_dtype_as_group('float', 'cummax', 'direct')
+ 123±5μs 614±50μs 5.01 groupby.GroupByMethods.time_dtype_as_field('int', 'cummin', 'transformation')
+ 117±10μs 581±10μs 4.96 groupby.GroupByMethods.time_dtype_as_group('float', 'cummin', 'direct')
+ 71.6±8μs 348±2μs 4.87 period.PeriodProperties.time_property('M', 'end_time')
+ 3.55±0.2μs 17.2±1μs 4.85 indexing.DataFrameStringIndexing.time_ix
+ 128±10μs 620±10μs 4.84 groupby.GroupByMethods.time_dtype_as_field('int', 'cummin', 'direct')
+ 142±10μs 687±60μs 4.83 groupby.GroupByMethods.time_dtype_as_group('float', 'cumsum', 'direct')
+ 137±8μs 655±60μs 4.79 groupby.GroupByMethods.time_dtype_as_field('int', 'cumsum', 'transformation')
+ 33.5±0.5ms 158±2ms 4.73 eval.Eval.time_and('python', 1)
+ 124±20μs 580±10μs 4.68 groupby.GroupByMethods.time_dtype_as_group('float', 'cummax', 'transformation')
+ 140±20μs 632±20μs 4.52 groupby.GroupByMethods.time_dtype_as_field('int', 'cumsum', 'direct')
+ 135±10μs 608±10μs 4.50 groupby.GroupByMethods.time_dtype_as_group('int', 'cumsum', 'direct')
+ 140±10μs 628±7μs 4.50 groupby.GroupByMethods.time_dtype_as_group('int', 'cumsum', 'transformation')
+ 141±10μs 627±20μs 4.44 groupby.GroupByMethods.time_dtype_as_group('float', 'cumsum', 'transformation')
+ 42.1±3ms 162±3ms 3.84 eval.Eval.time_and('python', 'all')
+ 37.6±1ms 144±9ms 3.84 frame_methods.Dropna.time_dropna('all', 0)
+ 64.7±1μs 237±1μs 3.66 period.PeriodUnaryMethods.time_to_timestamp('M')
+ 70.0±3μs 255±20μs 3.63 period.PeriodProperties.time_property('min', 'start_time')
+ 65.0±1μs 236±0.8μs 3.63 period.PeriodUnaryMethods.time_to_timestamp('min')
+ 41.8±2ms 151±7ms 3.61 frame_methods.Dropna.time_dropna('all', 1)
+ 65.4±1μs 235±1μs 3.59 period.PeriodProperties.time_property('M', 'start_time')
+ 54.1±1μs 188±20μs 3.47 period.Indexing.time_unique
+ 109±4ms 369±10ms 3.38 groupby.Groups.time_series_groups('int64_large')
+ 3.52±0.07μs 11.6±0.3μs 3.28 multiindex_object.GetLoc.time_med_get_loc
+ 252±20μs 823±50μs 3.27 groupby.GroupByMethods.time_dtype_as_field('int', 'cumprod', 'direct')
+ 266±30μs 850±100μs 3.20 groupby.GroupByMethods.time_dtype_as_group('float', 'cumprod', 'direct')
+ 3.57±0.3μs 11.2±0.3μs 3.14 multiindex_object.GetLoc.time_string_get_loc
+ 57.4±0.6μs 179±9μs 3.12 period.PeriodUnaryMethods.time_now('min')
+ 3.58±0.1ms 11.1±0.1ms 3.11 multiindex_object.GetLoc.time_med_get_loc_warm
+ 95.4±8μs 295±30μs 3.09 period.Algorithms.time_drop_duplicates('index')
+ 29.8±0.4ms 91.4±3ms 3.06 binary_ops.Ops.time_frame_multi_and(False, 'default')
+ 30.4±0.6ms 92.9±2ms 3.06 binary_ops.Ops.time_frame_multi_and(False, 1)
+ 5.82±0.1ms 17.6±0.4ms 3.02 frame_methods.Equals.time_frame_nonunique_unequal
+ 112±2μs 336±20μs 3.00 period.PeriodIndexConstructor.time_from_date_range('D')
+ 49.3±3μs 148±40μs 3.00 groupby.GroupByMethods.time_dtype_as_field('float', 'size', 'direct')
+ 260±20μs 776±40μs 2.98 groupby.GroupByMethods.time_dtype_as_group('int', 'cumprod', 'transformation')
+ 9.25±0.2μs 27.5±0.6μs 2.97 timestamp.TimestampProperties.time_is_month_start(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 5.75±0.08ms 17.0±0.1ms 2.96 frame_methods.Equals.time_frame_nonunique_equal
+ 51.1±2μs 148±30μs 2.90 groupby.GroupByMethods.time_dtype_as_field('float', 'size', 'transformation')
+ 259±30μs 751±50μs 2.89 groupby.GroupByMethods.time_dtype_as_field('int', 'cumprod', 'transformation')
+ 279±30μs 807±10μs 2.89 groupby.GroupByMethods.time_dtype_as_group('int', 'cumprod', 'direct')
+ 33.7±0.7ms 96.5±40ms 2.87 binary_ops.Ops.time_frame_multi_and(True, 1)
+ 154±5ms 440±30ms 2.85 groupby.Groups.time_series_groups('object_large')
+ 145±5μs 411±3μs 2.84 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x113681950>, False)
+ 22.9±1ms 65.0±2ms 2.84 groupby.ApplyDictReturn.time_groupby_apply_dict_return
+ 158±5μs 448±20μs 2.83 period.Indexing.time_intersection
+ 9.59±0.3μs 27.2±3μs 2.83 timestamp.TimestampProperties.time_is_month_end(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 147±5μs 413±6μs 2.82 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c0840>, False)
+ 57.6±6ms 161±20ms 2.79 groupby.GroupByMethods.time_dtype_as_field('float', 'unique', 'direct')
+ 157±5μs 439±5μs 2.79 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x113681950>, True)
+ 9.60±0.2μs 26.3±0.09μs 2.74 timestamp.TimestampProperties.time_is_year_start(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 9.73±0.2μs 26.7±0.4μs 2.74 timestamp.TimestampProperties.time_is_year_end(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 160±10μs 438±7μs 2.74 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c0840>, True)
+ 151±9μs 413±6μs 2.73 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c07b8>, False)
+ 151±9μs 412±20μs 2.73 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c08c8>, False)
+ 355±10μs 966±90μs 2.72 period.Algorithms.time_value_counts('index')
+ 164±7μs 445±8μs 2.71 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c07b8>, True)
+ 10.0±0.2μs 27.0±0.6μs 2.69 timestamp.TimestampProperties.time_is_quarter_start(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 163±20μs 439±20μs 2.69 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136c08c8>, True)
+ 9.74±0.3μs 26.1±0.1μs 2.68 timestamp.TimestampProperties.time_is_leap_year(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 57.5±2μs 154±5μs 2.68 period.Indexing.time_series_loc
+ 9.88±0.09μs 26.3±0.1μs 2.66 timestamp.TimestampProperties.time_is_quarter_end(<DstTzInfo 'Europe/Amsterdam' LMT+0:20:00 STD>, 'B')
+ 6.91±0.3μs 18.0±0.3μs 2.61 timestamp.TimestampAcrossDst.time_replace_across_dst
+ 58.7±4ms 152±10ms 2.59 groupby.GroupByMethods.time_dtype_as_field('float', 'unique', 'transformation')
+ 82.5±4ms 210±9ms 2.55 groupby.GroupByMethods.time_dtype_as_group('int', 'unique', 'transformation')
+ 39.2±0.4ms 99.5±2ms 2.54 binary_ops.Ops.time_frame_multi_and(True, 'default')
+ 288±30μs 730±9μs 2.54 groupby.GroupByMethods.time_dtype_as_group('float', 'cumprod', 'transformation')
+ 1.22±0.1ms 3.07±0.2ms 2.52 io.csv.ReadCSVDInferDatetimeFormat.time_read_csv(False, 'iso8601')
+ 803±50μs 2.00±0.03ms 2.49 io.csv.ReadCSVParseDates.time_multiple_date
+ 82.2±5ms 204±8ms 2.48 groupby.GroupByMethods.time_dtype_as_group('int', 'unique', 'direct')
+ 73.7±2ms 180±20ms 2.45 groupby.GroupByMethods.time_dtype_as_field('object', 'unique', 'direct')
+ 102±5ms 249±6ms 2.44 reshape.WideToLong.time_wide_to_long_big
+ 4.70±0.6ms 11.4±0.1ms 2.44 multiindex_object.GetLoc.time_small_get_loc_warm
+ 50.5±5ms 123±7ms 2.43 join_merge.MergeAsof.time_by_int
+ 38.9±0.5ms 94.4±2ms 2.43 frame_methods.Interpolate.time_interpolate(None)
+ 72.3±4ms 175±20ms 2.42 groupby.GroupByMethods.time_dtype_as_field('datetime', 'unique', 'direct')
+ 129±10ms 309±2ms 2.39 groupby.GroupByMethods.time_dtype_as_group('datetime', 'unique', 'direct')
+ 1.98±0.03ms 4.72±0.3ms 2.38 binary_ops.Timeseries.time_series_timestamp_compare(None)
+ 8.43±0.4μs 19.9±0.4μs 2.36 timestamp.TimestampOps.time_replace_tz(None)
+ 131±10ms 308±0.9ms 2.34 groupby.GroupByMethods.time_dtype_as_group('float', 'unique', 'direct')
+ 8.30±0.3μs 19.3±0.5μs 2.33 ctors.SeriesDtypesConstructors.time_dtindex_from_series
+ 1.97±0.02ms 4.57±0.01ms 2.32 binary_ops.Timeseries.time_timestamp_series_compare(None)
+ 58.6±6ms 134±3ms 2.29 groupby.GroupByMethods.time_dtype_as_field('int', 'unique', 'direct')
+ 58.1±4ms 133±2ms 2.29 groupby.GroupByMethods.time_dtype_as_field('int', 'unique', 'transformation')
+ 66.5±3μs 152±20μs 2.28 groupby.GroupByMethods.time_dtype_as_group('float', 'count', 'transformation')
+ 136±10ms 309±1ms 2.28 groupby.GroupByMethods.time_dtype_as_group('datetime', 'unique', 'transformation')
+ 845±20μs 1.91±0.2ms 2.26 indexing.NonNumericSeriesIndexing.time_getitem_list_like('datetime', 'unique_monotonic_inc')
+ 74.4±0.6ms 167±10ms 2.25 groupby.GroupByMethods.time_dtype_as_field('object', 'unique', 'transformation')
+ 70.5±2ms 157±10ms 2.23 groupby.GroupByMethods.time_dtype_as_field('datetime', 'unique', 'transformation')
+ 89.2±8ms 196±50ms 2.20 groupby.GroupByMethods.time_dtype_as_field('float', 'skew', 'direct')
+ 14.2±0.3μs 31.1±0.7μs 2.20 timestamp.TimestampOps.time_replace_tz('US/Eastern')
+ 2.29±0.2s 4.98±0.4s 2.17 replace.ReplaceDict.time_replace_series(False)
+ 178±2μs 386±5μs 2.17 multiindex_object.Values.time_datetime_level_values_sliced
+ 1.75±0.02ms 3.74±0.9ms 2.14 reshape.SimpleReshape.time_stack
+ 8.84±0.2ms 18.8±4ms 2.12 stat_ops.FrameOps.time_op('mad', 'float', 1, False)
+ 146±30ms 306±5ms 2.09 groupby.GroupByMethods.time_dtype_as_group('float', 'unique', 'transformation')
+ 95.3±7μs 197±80μs 2.07 groupby.GroupByMethods.time_dtype_as_field('float', 'sum', 'direct')
+ 83.1±4μs 172±50μs 2.06 series_methods.SeriesConstructor.time_constructor(None)
+ 524±10μs 1.06±0.01ms 2.02 groupby.GroupByMethods.time_dtype_as_group('object', 'unique', 'transformation')
+ 139±10μs 279±100μs 2.01 groupby.GroupByMethods.time_dtype_as_field('float', 'std', 'transformation')
+ 525±20μs 1.06±0.01ms 2.01 groupby.GroupByMethods.time_dtype_as_group('object', 'unique', 'direct')
+ 54.3±3μs 108±0.8μs 2.00 timeseries.AsOf.time_asof_single_early('DataFrame')
+ 51.1±3μs 101±10μs 1.99 groupby.GroupByMethods.time_dtype_as_field('datetime', 'size', 'direct')
+ 67.4±5μs 133±2μs 1.97 groupby.GroupByMethods.time_dtype_as_group('int', 'count', 'transformation')
+ 1.75±0.02ms 3.43±0.2ms 1.97 reshape.Melt.time_melt_dataframe
+ 51.7±1μs 102±5μs 1.96 groupby.GroupByMethods.time_dtype_as_field('object', 'size', 'transformation')
+ 1.06±0.1s 2.06±0.2s 1.95 stat_ops.FrameMultiIndexOps.time_op([0, 1], 'mad')
+ 84.4±2ms 163±6ms 1.94 join_merge.MergeAsof.time_by_object
+ 50.8±1μs 98.0±6μs 1.93 groupby.GroupByMethods.time_dtype_as_field('object', 'size', 'direct')
+ 2.67±0.4ms 5.12±0.7ms 1.91 reindex.DropDuplicates.time_frame_drop_dups_bool(True)
+ 867±70μs 1.66±0.01ms 1.91 io.csv.ReadCSVParseDates.time_baseline
+ 50.3±2μs 95.9±5μs 1.91 groupby.GroupByMethods.time_dtype_as_field('datetime', 'size', 'transformation')
+ 66.8±3μs 127±10μs 1.90 groupby.GroupByMethods.time_dtype_as_field('int', 'count', 'direct')
+ 67.0±6ms 127±3ms 1.90 frame_methods.Interpolate.time_interpolate('infer')
+ 51.9±7μs 98.0±4μs 1.89 groupby.GroupByMethods.time_dtype_as_group('int', 'size', 'transformation')
+ 1.29±0.2s 2.44±0.04s 1.89 timeseries.ToDatetimeNONISO8601.time_different_offset
+ 116±10ms 219±3ms 1.88 stat_ops.FrameMultiIndexOps.time_op(1, 'mad')
+ 50.0±2μs 94.0±0.6μs 1.88 groupby.GroupByMethods.time_dtype_as_group('datetime', 'size', 'direct')
+ 2.93±0.1s 5.50±0.3s 1.88 replace.ReplaceDict.time_replace_series(True)
+ 66.5±4μs 124±2μs 1.86 groupby.GroupByMethods.time_dtype_as_field('float', 'count', 'transformation')
+ 67.4±9μs 125±7μs 1.86 groupby.GroupByMethods.time_dtype_as_field('datetime', 'count', 'direct')
+ 65.6±3μs 122±4μs 1.86 groupby.GroupByMethods.time_dtype_as_field('float', 'count', 'direct')
+ 1.87±0.03s 3.48±0.07s 1.86 sparse.SparseDataFrameConstructor.time_constructor
+ 49.6±2μs 91.9±1μs 1.85 groupby.GroupByMethods.time_dtype_as_field('int', 'size', 'direct')
+ 927±30μs 1.71±0.03ms 1.85 frame_methods.Interpolate.time_interpolate_some_good(None)
+ 50.0±4μs 92.4±0.6μs 1.85 groupby.GroupByMethods.time_dtype_as_group('float', 'size', 'direct')
+ 8.77±0.09ms 16.2±0.7ms 1.84 stat_ops.FrameOps.time_op('mad', 'float', 1, True)
+ 66.1±3μs 122±8μs 1.84 groupby.GroupByMethods.time_dtype_as_field('datetime', 'count', 'transformation')
+ 50.4±3μs 92.7±1μs 1.84 groupby.GroupByMethods.time_dtype_as_field('int', 'size', 'transformation')
+ 50.8±2μs 93.3±0.7μs 1.84 groupby.GroupByMethods.time_dtype_as_group('float', 'size', 'transformation')
+ 50.8±2μs 92.8±0.3μs 1.83 groupby.GroupByMethods.time_dtype_as_group('datetime', 'size', 'transformation')
+ 49.9±0.6μs 91.2±0.3μs 1.83 groupby.GroupByMethods.time_dtype_as_group('object', 'size', 'direct')
+ 86.6±3μs 158±20μs 1.83 indexing.CategoricalIndexIndexing.time_getitem_bool_array('monotonic_decr')
+ 65.0±0.5μs 118±0.6μs 1.82 groupby.GroupByMethods.time_dtype_as_group('datetime', 'count', 'direct')
+ 52.4±4μs 95.3±1μs 1.82 groupby.GroupByMethods.time_dtype_as_group('int', 'size', 'direct')
+ 50.5±2μs 91.7±0.7μs 1.82 groupby.GroupByMethods.time_dtype_as_group('object', 'size', 'transformation')
+ 65.0±0.2μs 118±0.7μs 1.81 groupby.GroupByMethods.time_dtype_as_group('datetime', 'count', 'transformation')
+ 87.8±5ms 159±10ms 1.81 groupby.GroupByMethods.time_dtype_as_field('float', 'skew', 'transformation')
+ 64.3±0.6μs 115±0.4μs 1.79 groupby.GroupByMethods.time_dtype_as_group('object', 'count', 'direct')
+ 14.3±0.5μs 25.5±0.7μs 1.78 ctors.SeriesDtypesConstructors.time_index_from_array_floats
+ 66.5±4μs 118±4μs 1.78 groupby.GroupByMethods.time_dtype_as_field('int', 'count', 'transformation')
+ 2.68±0.06ms 4.76±1ms 1.78 rolling.VariableWindowMethods.time_rolling('DataFrame', '50s', 'int', 'min')
+ 30.3±2ms 53.4±2ms 1.76 binary_ops.Ops.time_frame_comparison(False, 'default')
+ 18.9±0.3μs 33.4±0.6μs 1.76 ctors.SeriesDtypesConstructors.time_dtindex_from_index_with_series
+ 343±10μs 597±80μs 1.74 groupby.GroupByMethods.time_dtype_as_field('float', 'sem', 'transformation')
+ 29.3±0.4μs 50.8±1μs 1.73 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136818c8>, True)
+ 102±2μs 176±3μs 1.72 frame_methods.GetDtypeCounts.time_frame_get_dtype_counts
+ 1.09±0.02ms 1.87±0.06ms 1.71 io.csv.ReadCSVDInferDatetimeFormat.time_read_csv(False, 'ymd')
+ 31.7±2ms 53.4±0.4ms 1.69 binary_ops.Ops.time_frame_comparison(False, 1)
+ 3.80±0.1μs 6.41±0.7μs 1.69 inference.ToNumericDowncast.time_downcast('int32', None)
+ 2.90±0.3ms 4.88±0.3ms 1.69 rolling.VariableWindowMethods.time_rolling('Series', '50s', 'int', 'max')
+ 191±2ms 318±20ms 1.66 sparse.SparseDataFrameConstructor.time_from_scipy
+ 186±10μs 308±50μs 1.65 groupby.GroupByMethods.time_dtype_as_field('object', 'last', 'direct')
+ 138±10ms 226±4ms 1.64 groupby.GroupByMethods.time_dtype_as_group('int', 'skew', 'direct')
+ 3.03±0.2ms 4.94±1ms 1.63 rolling.Methods.time_rolling('DataFrame', 1000, 'float', 'std')
+ 123±5ms 199±4ms 1.62 frame_methods.Iteration.time_iterrows
+ 3.66±0.09μs 5.91±0.4μs 1.62 offset.OnOffset.time_on_offset(<MonthBegin>)
+ 71.9±10μs 116±0.3μs 1.62 groupby.GroupByMethods.time_dtype_as_group('object', 'count', 'transformation')
+ 117±0.9μs 187±20μs 1.61 indexing.DataFrameNumericIndexing.time_iloc_dups
+ 4.59±0.05ms 7.39±0.6ms 1.61 categoricals.Rank.time_rank_string_cat_ordered
+ 4.80±0.06ms 7.68±0.6ms 1.60 categoricals.Rank.time_rank_int_cat
+ 2.96±0.1ms 4.73±0.8ms 1.59 rolling.Methods.time_rolling('DataFrame', 10, 'int', 'std')
+ 98.5±5μs 156±8μs 1.59 groupby.GroupByMethods.time_dtype_as_field('float', 'var', 'direct')
+ 1.07±0.02ms 1.69±0.03ms 1.59 timeseries.ResampleDataFrame.time_method('min')
+ 6.15±0.1μs 9.74±0.08μs 1.58 timestamp.TimestampOps.time_replace_None('US/Eastern')
+ 164±5μs 259±3μs 1.58 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x113681730>, True)
+ 4.54±0.2ms 7.15±0.7ms 1.58 categoricals.Rank.time_rank_int_cat_ordered
+ 1.07±0.04ms 1.69±0.01ms 1.58 timeseries.ResampleDataFrame.time_method('max')
+ 21.4±1μs 33.7±1μs 1.57 ctors.SeriesDtypesConstructors.time_index_from_array_string
+ 1.70±0.03ms 2.67±0.2ms 1.57 io.csv.ReadCSVDInferDatetimeFormat.time_read_csv(True, 'iso8601')
+ 46.3±4μs 72.6±3μs 1.57 timeseries.SortIndex.time_get_slice(False)
+ 92.8±6μs 145±10μs 1.56 groupby.GroupByMethods.time_dtype_as_field('float', 'prod', 'transformation')
+ 112±2μs 175±1μs 1.56 timeseries.DatetimeIndex.time_unique('dst')
+ 89.0±5μs 138±30μs 1.55 groupby.GroupByMethods.time_dtype_as_field('datetime', 'max', 'direct')
+ 95.6±10μs 148±40μs 1.55 groupby.GroupByMethods.time_dtype_as_field('datetime', 'min', 'direct')
+ 150±5μs 232±2μs 1.55 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x113681730>, False)
+ 477±40μs 737±40μs 1.55 groupby.GroupByMethods.time_dtype_as_group('int', 'sem', 'transformation')
+ 534±6μs 824±2μs 1.54 indexing.MultiIndexing.time_frame_ix
+ 3.56±0.2ms 5.50±0.9ms 1.54 rolling.Methods.time_rolling('Series', 10, 'float', 'std')
+ 2.87±0.1ms 4.43±0.2ms 1.54 rolling.Methods.time_rolling('DataFrame', 10, 'float', 'std')
+ 1.70±0.03ms 2.60±0.2ms 1.53 io.csv.ReadCSVDInferDatetimeFormat.time_read_csv(True, 'ymd')
+ 282±6μs 432±10μs 1.53 indexing.NonNumericSeriesIndexing.time_getitem_label_slice('datetime', 'nonunique_monotonic_inc')
+ 91.8±3ms 140±7ms 1.53 groupby.GroupByMethods.time_dtype_as_field('int', 'skew', 'transformation')
+ 2.64±0.08ms 4.03±0.4ms 1.53 categoricals.Concat.time_union
+ 4.44±0.03ms 6.78±0.05ms 1.53 categoricals.Rank.time_rank_int
+ 8.50±0.1ms 13.0±5ms 1.53 stat_ops.FrameOps.time_op('mad', 'float', 0, True)
+ 1.69±0.09ms 2.58±0.5ms 1.52 io.csv.ReadCSVFloatPrecision.time_read_csv(';', '.', 'high')
+ 128±6μs 195±4μs 1.52 groupby.GroupByMethods.time_dtype_as_group('float', 'first', 'direct')
+ 1.77±0.08ms 2.68±0.5ms 1.52 io.csv.ReadCSVFloatPrecision.time_read_csv(';', '.', None)
+ 10.0±0.3ms 15.2±0.3ms 1.52 eval.Query.time_query_datetime_column
+ 96.1±10μs 145±20μs 1.51 groupby.GroupByMethods.time_dtype_as_field('float', 'sum', 'transformation')
+ 124±9μs 188±8μs 1.51 groupby.GroupByMethods.time_dtype_as_group('float', 'last', 'direct')
+ 3.39±0.04ms 5.14±0.2ms 1.51 frame_methods.Apply.time_apply_pass_thru
+ 128±9μs 194±8μs 1.51 groupby.GroupByMethods.time_dtype_as_group('float', 'first', 'transformation')
+ 2.96±0.02μs 4.48±0.1μs 1.51 categoricals.CategoricalSlicing.time_getitem_scalar('non_monotonic')
+ 727±60ms 1.10±0.06s 1.51 groupby.GroupByMethods.time_dtype_as_group('float', 'mad', 'direct')
+ 6.22±0.04ms 9.28±0.5ms 1.49 frame_methods.Apply.time_apply_lambda_mean
+ 80.2±4μs 119±20μs 1.49 groupby.GroupByMethods.time_dtype_as_field('float', 'shift', 'direct')
+ 861±30μs 1.27±0.06ms 1.47 period.Indexing.time_align
+ 3.59±0.3ms 5.29±0.4ms 1.47 reindex.DropDuplicates.time_frame_drop_dups_bool(False)
+ 130±3μs 191±20μs 1.47 groupby.GroupByMethods.time_dtype_as_field('object', 'count', 'transformation')
+ 13.4±0.3ms 19.7±5ms 1.47 reshape.PivotTable.time_pivot_table
+ 154±10μs 226±8μs 1.47 groupby.GroupByMethods.time_dtype_as_field('int', 'var', 'transformation')
+ 250±5μs 366±10μs 1.46 frame_ctor.FromRecords.time_frame_from_records_generator(None)
+ 50.0±2ms 73.0±4ms 1.46 index_object.IndexAppend.time_append_range_list
+ 3.08±0.01μs 4.50±0.1μs 1.46 categoricals.CategoricalSlicing.time_getitem_scalar('monotonic_incr')
+ 2.68±0.1ms 3.91±0.2ms 1.46 rolling.VariableWindowMethods.time_rolling('DataFrame', '50s', 'int', 'max')
+ 23.8±0.4ms 34.7±0.2ms 1.46 frame_methods.Equals.time_frame_object_unequal
+ 117±3ms 171±3ms 1.46 sparse.SparseSeriesToFrame.time_series_to_frame
+ 468±30ms 679±40ms 1.45 groupby.GroupByMethods.time_dtype_as_group('int', 'mad', 'transformation')
+ 3.05±0.09ms 4.41±0.1ms 1.45 gil.ParallelRolling.time_rolling('var')
+ 299±10ms 432±10ms 1.45 groupby.GroupByMethods.time_dtype_as_field('float', 'mad', 'transformation')
+ 304±20ms 439±20ms 1.44 groupby.GroupByMethods.time_dtype_as_field('int', 'mad', 'direct')
+ 11.7±0.6μs 16.9±0.4μs 1.44 offset.OffestDatetimeArithmetic.time_apply(<DateOffset: days=2, months=2>)
+ 1.65±0.1ms 2.38±0.1ms 1.44 io.csv.ReadCSVFloatPrecision.time_read_csv(',', '.', 'high')
+ 286±7μs 411±20μs 1.44 groupby.GroupByMethods.time_dtype_as_group('float', 'median', 'direct')
+ 734±80ms 1.05±0.06s 1.43 groupby.GroupByMethods.time_dtype_as_group('float', 'mad', 'transformation')
+ 2.22±0.08ms 3.16±0.04ms 1.43 frame_methods.Interpolate.time_interpolate_some_good('infer')
+ 169±20ms 241±2ms 1.43 indexing.NumericSeriesIndexing.time_getitem_list_like(<class 'pandas.core.indexes.numeric.Float64Index'>, 'unique_monotonic_inc')
+ 86.8±0.7ms 124±3ms 1.43 frame_methods.Apply.time_apply_axis_1
+ 218±10ms 310±6ms 1.42 groupby.GroupByMethods.time_dtype_as_group('float', 'skew', 'direct')
+ 2.95±0.2ms 4.19±0.2ms 1.42 rolling.VariableWindowMethods.time_rolling('Series', '50s', 'int', 'min')
+ 215±30ms 305±2ms 1.42 groupby.GroupByMethods.time_dtype_as_group('float', 'skew', 'transformation')
+ 132±10μs 187±9μs 1.42 groupby.GroupByMethods.time_dtype_as_group('float', 'last', 'transformation')
+ 3.77±0.08ms 5.33±1ms 1.41 rolling.Methods.time_rolling('Series', 1000, 'float', 'std')
+ 95.2±6μs 135±4μs 1.41 groupby.GroupByMethods.time_dtype_as_field('float', 'min', 'transformation')
+ 9.73±0.1μs 13.8±0.2μs 1.41 timestamp.TimestampConstruction.time_parse_iso8601_tz
+ 240±7μs 338±7μs 1.41 groupby.GroupByMethods.time_dtype_as_group('float', 'head', 'transformation')
+ 222±4ms 311±6ms 1.40 frame_methods.Duplicated.time_frame_duplicated_wide
+ 687±20μs 963±80μs 1.40 groupby.GroupByMethods.time_dtype_as_field('datetime', 'value_counts', 'direct')
+ 127±5μs 177±8μs 1.40 groupby.GroupByMethods.time_dtype_as_field('object', 'count', 'direct')
+ 94.5±5μs 132±10μs 1.40 groupby.GroupByMethods.time_dtype_as_field('int', 'shift', 'transformation')
+ 242±8μs 339±5μs 1.40 groupby.GroupByMethods.time_dtype_as_group('float', 'head', 'direct')
+ 92.7±10μs 129±1μs 1.40 groupby.GroupByMethods.time_dtype_as_field('float', 'first', 'transformation')
+ 96.8±7μs 135±5μs 1.39 groupby.GroupByMethods.time_dtype_as_field('float', 'max', 'transformation')
+ 840±40μs 1.17±0.2ms 1.39 groupby.GroupByMethods.time_dtype_as_field('float', 'value_counts', 'direct')
+ 2.88±0.2ms 4.01±0.1ms 1.39 rolling.VariableWindowMethods.time_rolling('Series', '50s', 'float', 'max')
+ 755±20μs 1.05±0.01ms 1.39 timeseries.ResampleDataFrame.time_method('mean')
+ 391±20μs 542±2μs 1.39 groupby.GroupByMethods.time_dtype_as_group('float', 'sem', 'direct')
+ 7.02±0.1μs 9.73±0.7μs 1.39 index_object.Indexing.time_get_loc_sorted('Int')
+ 139±8μs 192±10μs 1.38 groupby.GroupByMethods.time_dtype_as_group('int', 'first', 'direct')
+ 98.0±4μs 135±4μs 1.38 groupby.GroupByMethods.time_dtype_as_field('float', 'min', 'direct')
+ 1.74±0.1ms 2.41±0.1ms 1.38 io.csv.ReadCSVFloatPrecision.time_read_csv(';', '_', 'round_trip')
+ 14.1±0.7μs 19.4±0.09μs 1.38 offset.OffestDatetimeArithmetic.time_apply_np_dt64(<DateOffset: days=2, months=2>)
+ 94.6±7μs 130±1μs 1.38 groupby.GroupByMethods.time_dtype_as_field('float', 'max', 'direct')
+ 392±10μs 540±3μs 1.38 groupby.GroupByMethods.time_dtype_as_group('float', 'sem', 'transformation')
+ 1.96±0.09ms 2.69±0.6ms 1.38 rolling.Methods.time_rolling('DataFrame', 1000, 'float', 'count')
+ 118±3μs 161±7μs 1.37 groupby.GroupByMethods.time_dtype_as_field('int', 'last', 'transformation')
+ 125±8μs 171±10μs 1.37 groupby.GroupByMethods.time_dtype_as_group('int', 'last', 'transformation')
+ 433±10μs 593±200μs 1.37 reindex.Reindex.time_reindex_columns
+ 473±30ms 646±20ms 1.37 groupby.GroupByMethods.time_dtype_as_group('int', 'mad', 'direct')
+ 88.1±7μs 120±6μs 1.36 groupby.GroupByMethods.time_dtype_as_field('datetime', 'shift', 'direct')
+ 2.82±0.1ms 3.83±0.07ms 1.36 rolling.VariableWindowMethods.time_rolling('Series', '50s', 'float', 'min')
+ 1.47±0.06ms 1.99±0.1ms 1.35 groupby.Datelike.time_sum('date_range')
+ 27.0±0.3ms 36.5±2ms 1.35 strings.Methods.time_get
+ 31.7±0.7ms 42.8±2ms 1.35 indexing.InsertColumns.time_insert
+ 100±8μs 135±2μs 1.35 groupby.GroupByMethods.time_dtype_as_field('float', 'prod', 'direct')
+ 849±30μs 1.14±0.1ms 1.35 groupby.GroupByMethods.time_dtype_as_field('float', 'value_counts', 'transformation')
+ 6.92±0.1ms 9.28±0.3ms 1.34 frame_methods.Apply.time_apply_np_mean
+ 434±7μs 583±8μs 1.34 categoricals.CategoricalSlicing.time_getitem_list('non_monotonic')
+ 69.8±1μs 93.6±7μs 1.34 indexing.NonNumericSeriesIndexing.time_getitem_label_slice('datetime', 'unique_monotonic_inc')
+ 706±30μs 945±60μs 1.34 groupby.GroupByMethods.time_dtype_as_field('datetime', 'value_counts', 'transformation')
+ 277±10μs 370±8μs 1.34 groupby.GroupByMethods.time_dtype_as_group('float', 'prod', 'transformation')
+ 122±1μs 163±0.4μs 1.33 groupby.GroupByMethods.time_dtype_as_group('datetime', 'last', 'direct')
+ 275±10μs 366±20μs 1.33 groupby.GroupByMethods.time_dtype_as_group('int', 'prod', 'direct')
+ 50.0±1μs 66.5±2μs 1.33 frame_ctor.FromNDArray.time_frame_from_ndarray
+ 118±3μs 156±10μs 1.33 groupby.GroupByMethods.time_dtype_as_group('object', 'first', 'transformation')
+ 129±5μs 171±4μs 1.33 groupby.GroupByMethods.time_dtype_as_group('datetime', 'max', 'transformation')
+ 19.2±0.3μs 25.4±0.8μs 1.33 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136818c8>, False)
+ 131±6μs 174±20μs 1.33 groupby.GroupByMethods.time_dtype_as_field('int', 'first', 'direct')
+ 131±6μs 174±6μs 1.32 groupby.GroupByMethods.time_dtype_as_group('datetime', 'max', 'direct')
+ 122±2μs 161±3μs 1.32 groupby.GroupByMethods.time_dtype_as_field('int', 'last', 'direct')
+ 5.67±0.05ms 7.48±0.1ms 1.32 reindex.DropDuplicates.time_frame_drop_dups_na(True)
+ 6.26±0.5ms 8.23±0.2ms 1.32 strings.Cat.time_cat(0, None, None, 0.001)
+ 1.72±0.09ms 2.26±0.02ms 1.32 io.csv.ReadCSVFloatPrecision.time_read_csv(';', '_', None)
+ 596±20ns 784±60ns 1.31 index_object.Indexing.time_get('String')
+ 128±2μs 169±3μs 1.31 groupby.GroupByMethods.time_dtype_as_group('datetime', 'min', 'direct')
+ 127±8μs 166±9μs 1.31 groupby.GroupByMethods.time_dtype_as_group('int', 'last', 'direct')
+ 2.78±0.1ms 3.65±0.07ms 1.31 rolling.VariableWindowMethods.time_rolling('DataFrame', '50s', 'float', 'max')
+ 135±6μs 177±3μs 1.31 groupby.GroupByMethods.time_dtype_as_field('int', 'first', 'transformation')
+ 3.73±0.2ms 4.87±0.03ms 1.31 rolling.Methods.time_rolling('Series', 10, 'int', 'std')
+ 130±5μs 170±0.7μs 1.31 groupby.GroupByMethods.time_dtype_as_group('datetime', 'min', 'transformation')
+ 145±5μs 189±10μs 1.30 groupby.GroupByMethods.time_dtype_as_group('int', 'first', 'transformation')
+ 135±6μs 175±1μs 1.30 groupby.GroupByMethods.time_dtype_as_field('int', 'max', 'transformation')
+ 149±10μs 194±10μs 1.30 groupby.GroupByMethods.time_dtype_as_group('float', 'var', 'transformation')
+ 1.72±0.1ms 2.24±0.2ms 1.30 io.csv.ReadCSVFloatPrecision.time_read_csv(',', '_', 'round_trip')
+ 1.71±0.1ms 2.22±0.03ms 1.30 io.csv.ReadCSVFloatPrecision.time_read_csv(',', '_', None)
+ 6.28±0.2ms 8.12±0.2ms 1.29 categoricals.Rank.time_rank_string_cat
+ 1.24±0.09ms 1.61±0.04ms 1.29 sparse.FromCoo.time_sparse_series_from_coo
+ 125±7μs 162±4μs 1.29 groupby.GroupByMethods.time_dtype_as_group('datetime', 'last', 'transformation')
+ 61.6±3μs 79.5±3μs 1.29 frame_ctor.FromSeries.time_mi_series
+ 7.89±0.4μs 10.2±0.03μs 1.29 offset.OnOffset.time_on_offset(<YearEnd: month=12>)
+ 1.23±0.05μs 1.59±0.07μs 1.29 index_object.Indexing.time_get('Float')
+ 233±9μs 299±10μs 1.29 groupby.GroupByMethods.time_dtype_as_group('datetime', 'head', 'direct')
+ 343±5μs 442±10μs 1.29 timeseries.ResetIndex.time_reest_datetimeindex(None)
+ 58.7±1ms 75.5±0.5ms 1.29 stat_ops.Correlation.time_corr('spearman')
+ 31.7±0.5ms 40.7±3ms 1.29 stat_ops.FrameMultiIndexOps.time_op(0, 'mad')
+ 9.40±0.1μs 12.1±1μs 1.28 timestamp.TimestampProperties.time_is_leap_year(None, 'B')
+ 450±8μs 577±3μs 1.28 categoricals.CategoricalSlicing.time_getitem_list('monotonic_incr')
+ 4.38±0.2ms 5.60±1ms 1.28 rolling.Pairwise.time_pairwise(1000, 'corr', False)
+ 3.00±0.08ms 3.83±0.1ms 1.28 timeseries.ToDatetimeISO8601.time_iso8601_format
+ 3.81±0.09s 4.87±0.08s 1.28 period.DataFramePeriodColumn.time_set_index
+ 103±3μs 132±2μs 1.28 join_merge.Concat.time_concat_empty_right(0)
+ 145±10μs 185±7μs 1.28 groupby.GroupByMethods.time_dtype_as_group('int', 'min', 'transformation')
+ 311±8μs 397±80μs 1.28 groupby.GroupByMethods.time_dtype_as_field('object', 'tail', 'direct')
+ 9.29±0.3μs 11.8±0.9μs 1.27 offset.OffestDatetimeArithmetic.time_apply(<YearBegin: month=1>)
+ 1.75±0.1ms 2.23±0.09ms 1.27 io.csv.ReadCSVFloatPrecision.time_read_csv(',', '.', None)
+ 296±8ms 375±5ms 1.27 frame_methods.Nunique.time_frame_nunique
+ 99.4±7μs 126±4μs 1.27 groupby.GroupByMethods.time_dtype_as_group('int', 'shift', 'transformation')
+ 228±3μs 288±3μs 1.27 groupby.GroupByMethods.time_dtype_as_group('object', 'head', 'direct')
+ 153±1ms 194±2ms 1.26 replace.Convert.time_replace('DataFrame', 'Timedelta')
+ 2.39±0.1ms 3.02±0.2ms 1.26 io.csv.ReadCSVFloatPrecision.time_read_csv(';', '.', 'round_trip')
+ 11.3±0.3ms 14.2±0.3ms 1.26 categoricals.Constructor.time_regular
+ 8.15±0.1ms 10.3±0.2ms 1.26 stat_ops.FrameOps.time_op('mad', 'float', 0, False)
+ 19.1±0.3ms 24.1±0.3ms 1.26 stat_ops.FrameMultiIndexOps.time_op(0, 'kurt')
+ 13.9±0.5ms 17.5±0.2ms 1.26 join_merge.Concat.time_concat_series(0)
+ 229±10μs 288±4μs 1.26 timeseries.DatetimeIndex.time_normalize('dst')
+ 108±8μs 135±3μs 1.26 groupby.GroupByMethods.time_dtype_as_field('float', 'var', 'transformation')
+ 2.84±0.3ms 3.58±0.04ms 1.26 rolling.VariableWindowMethods.time_rolling('DataFrame', '50s', 'float', 'min')
+ 2.81±0.1ms 3.53±0.2ms 1.26 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(',', '_', 'round_trip')
+ 301±7μs 379±30μs 1.26 groupby.GroupByMethods.time_dtype_as_field('object', 'head', 'transformation')
+ 7.75±0.2ms 9.74±0.5ms 1.26 indexing.InsertColumns.time_assign_with_setitem
+ 234±3μs 294±4μs 1.26 groupby.GroupByMethods.time_dtype_as_group('datetime', 'head', 'transformation')
+ 236±8μs 296±4μs 1.26 groupby.GroupByMethods.time_dtype_as_group('object', 'tail', 'transformation')
+ 140±8μs 175±2μs 1.25 groupby.GroupByMethods.time_dtype_as_field('int', 'max', 'direct')
+ 149±2ms 186±20ms 1.25 binary_ops.Ops2.time_frame_float_div_by_zero
+ 1.98±0.1ms 2.47±0.2ms 1.25 rolling.VariableWindowMethods.time_rolling('DataFrame', '1d', 'float', 'sum')
+ 54.3±0.5μs 67.8±2μs 1.25 timeseries.SortIndex.time_get_slice(True)
+ 269±20μs 336±10μs 1.25 groupby.GroupByMethods.time_dtype_as_group('int', 'head', 'direct')
+ 91.0±2μs 113±30μs 1.25 groupby.GroupByMethods.time_dtype_as_field('datetime', 'first', 'direct')
+ 92.7±4μs 115±2μs 1.25 groupby.GroupByMethods.time_dtype_as_field('int', 'shift', 'direct')
+ 189±10μs 235±5μs 1.24 groupby.GroupByMethods.time_dtype_as_field('int', 'std', 'transformation')
+ 3.43±0.2ms 4.26±0.09ms 1.24 rolling.Methods.time_rolling('DataFrame', 1000, 'int', 'std')
+ 10.1±0.2ms 12.6±0.3ms 1.24 categoricals.CategoricalSlicing.time_getitem_bool_array('non_monotonic')
+ 75.1±2ms 93.1±4ms 1.24 frame_methods.ToHTML.time_to_html_mixed
+ 129±2μs 159±10μs 1.24 inference.NumericInferOps.time_subtract(<class 'numpy.int8'>)
+ 28.1±0.5ms 34.8±0.1ms 1.24 binary_ops.Timeseries.time_timestamp_ops_diff_with_shift('US/Eastern')
+ 124±6μs 153±0.9μs 1.23 groupby.GroupByMethods.time_dtype_as_field('float', 'median', 'transformation')
+ 118±4μs 146±5μs 1.23 groupby.GroupByMethods.time_dtype_as_field('object', 'shift', 'transformation')
+ 427±10μs 527±6μs 1.23 timeseries.ResetIndex.time_reest_datetimeindex('US/Eastern')
+ 2.22±0.03ms 2.74±0.07ms 1.23 groupby.Transform.time_transform_multi_key4
+ 259±8μs 318±4μs 1.23 groupby.GroupByMethods.time_dtype_as_field('int', 'head', 'transformation')
+ 84.9±2μs 104±0.9μs 1.23 groupby.GroupByMethods.time_dtype_as_group('object', 'shift', 'direct')
+ 8.56±0.06ms 10.5±0.1ms 1.23 stat_ops.Rank.time_rank('Series', False)
+ 44.1±1ms 54.1±2ms 1.23 frame_methods.Equals.time_frame_object_equal
+ 263±10μs 322±10μs 1.23 groupby.GroupByMethods.time_dtype_as_field('datetime', 'head', 'direct')
+ 2.39±0.1ms 2.93±0.2ms 1.23 io.csv.ReadCSVFloatPrecision.time_read_csv(',', '.', 'round_trip')
+ 587±4μs 720±2μs 1.23 groupby.GroupByMethods.time_dtype_as_group('object', 'value_counts', 'transformation')
+ 9.10±0.2μs 11.1±0.4μs 1.23 offset.OffestDatetimeArithmetic.time_apply(<YearEnd: month=12>)
+ 65.0±2μs 79.6±2μs 1.23 timeseries.SortIndex.time_sort_index(True)
+ 3.29±0.04μs 4.02±0.2μs 1.22 indexing.CategoricalIndexIndexing.time_getitem_scalar('monotonic_incr')
+ 10.2±0.1ms 12.5±0.09ms 1.22 gil.ParallelRolling.time_rolling('std')
+ 11.5±0.3ms 14.1±0.4ms 1.22 timedelta.TimedeltaOps.time_add_td_ts
+ 728±40μs 890±30μs 1.22 groupby.GroupByMethods.time_dtype_as_group('int', 'value_counts', 'direct')
+ 146±10μs 179±7μs 1.22 groupby.GroupByMethods.time_dtype_as_group('int', 'max', 'transformation')
+ 261±7μs 319±4μs 1.22 groupby.GroupByMethods.time_dtype_as_field('int', 'head', 'direct')
+ 102±6μs 124±4μs 1.22 groupby.GroupByMethods.time_dtype_as_field('float', 'mean', 'transformation')
+ 9.42±0.1ms 11.5±0.2ms 1.22 stat_ops.Rank.time_average_old('Series', True)
+ 6.53±0.1ms 7.96±0.1ms 1.22 groupby.Transform.time_transform_multi_key1
+ 68.1±0.6μs 83.0±7μs 1.22 indexing.DataFrameNumericIndexing.time_loc
+ 9.29±0.1ms 11.3±0.1ms 1.22 stat_ops.Rank.time_average_old('Series', False)
+ 138±2μs 168±3μs 1.22 join_merge.Concat.time_concat_empty_right(1)
+ 152±0.6ms 184±6ms 1.22 binary_ops.Ops2.time_frame_int_div_by_zero
+ 8.43±0.1ms 10.2±2ms 1.21 algorithms.Hashing.time_series_string
+ 43.1±0.7μs 52.3±3μs 1.21 indexing.NonNumericSeriesIndexing.time_getitem_pos_slice('string', 'nonunique_monotonic_inc')
+ 691±30μs 838±20μs 1.21 groupby.GroupByMethods.time_dtype_as_field('object', 'value_counts', 'transformation')
+ 9.48±0.1ms 11.5±0.04ms 1.21 frame_methods.MaskBool.time_frame_mask_floats
+ 614±10μs 744±90μs 1.21 frame_methods.Quantile.time_frame_quantile(1)
+ 86.9±0.7μs 105±0.3μs 1.21 groupby.GroupByMethods.time_dtype_as_group('datetime', 'shift', 'transformation')
+ 1.54±0.02ms 1.86±0.06ms 1.21 stat_ops.SeriesMultiIndexOps.time_op(1, 'prod')
+ 8.70±0.09ms 10.5±0.2ms 1.21 stat_ops.Rank.time_rank('Series', True)
+ 3.47±0.2ms 4.19±0.06ms 1.21 io.sas.SAS.time_read_msgpack('xport')
+ 88.0±3μs 106±0.7μs 1.21 groupby.GroupByMethods.time_dtype_as_group('float', 'shift', 'direct')
+ 601±20μs 726±4μs 1.21 groupby.GroupByMethods.time_dtype_as_group('object', 'value_counts', 'direct')
+ 284±20μs 343±10μs 1.21 groupby.GroupByMethods.time_dtype_as_group('int', 'median', 'transformation')
+ 86.6±2μs 105±0.9μs 1.21 groupby.GroupByMethods.time_dtype_as_group('object', 'shift', 'transformation')
+ 122±0.7ms 147±10ms 1.21 replace.Convert.time_replace('Series', 'Timestamp')
+ 255±20μs 307±2μs 1.21 groupby.GroupByMethods.time_dtype_as_group('float', 'tail', 'direct')
+ 35.6±0.7ms 42.8±1ms 1.20 io.csv.ReadCSVCategorical.time_convert_direct
+ 712±10μs 857±5μs 1.20 groupby.GroupByMethods.time_dtype_as_group('datetime', 'value_counts', 'direct')
+ 5.75±0.2ms 6.92±0.2ms 1.20 rolling.VariableWindowMethods.time_rolling('DataFrame', '1d', 'int', 'std')
+ 8.11±0.09ms 9.73±0.3ms 1.20 groupby.MultiColumn.time_col_select_numpy_sum
+ 305±10μs 366±20μs 1.20 groupby.GroupByMethods.time_dtype_as_field('object', 'head', 'direct')
+ 281±20μs 336±20μs 1.20 groupby.GroupByMethods.time_dtype_as_field('int', 'median', 'direct')
+ 140±1μs 168±4μs 1.20 join_merge.Concat.time_concat_empty_left(1)
+ 62.7±1ms 75.1±1ms 1.20 io.sas.SAS.time_read_msgpack('sas7bdat')
+ 54.1±1ms 64.8±2ms 1.20 stat_ops.SeriesMultiIndexOps.time_op(1, 'mad')
+ 71.5±3μs 85.6±6μs 1.20 inference.ToNumeric.time_from_str('ignore')
+ 715±9μs 853±8μs 1.19 groupby.GroupByMethods.time_dtype_as_group('datetime', 'value_counts', 'transformation')
+ 274±3ms 327±3ms 1.19 groupby.Apply.time_copy_overhead_single_col
+ 705±20μs 841±8μs 1.19 groupby.GroupByMethods.time_dtype_as_field('object', 'value_counts', 'direct')
+ 53.8±0.9μs 63.7±0.3μs 1.18 frame_methods.XS.time_frame_xs(0)
+ 35.3±1μs 41.7±1μs 1.18 offset.OffestDatetimeArithmetic.time_subtract(<DateOffset: days=2, months=2>)
+ 91.4±2μs 108±2μs 1.18 groupby.GroupByMethods.time_dtype_as_field('datetime', 'first', 'transformation')
+ 463±10μs 544±50μs 1.18 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<DateOffset: days=2, months=2>)
+ 266±10μs 313±1μs 1.18 groupby.GroupByMethods.time_dtype_as_field('float', 'head', 'direct')
+ 9.38±0.07ms 11.0±0.06ms 1.17 groupby.MultiColumn.time_cython_sum
+ 9.36±0.2μs 11.0±0.1μs 1.17 timestamp.TimestampProperties.time_is_year_start(None, 'B')
+ 20.7±0.3ms 24.2±2ms 1.17 stat_ops.SeriesMultiIndexOps.time_op(1, 'skew')
+ 1.00±0.01μs 1.17±0.04μs 1.17 timestamp.TimestampConstruction.time_parse_iso8601_no_tz
+ 275±10μs 320±7μs 1.16 groupby.GroupByMethods.time_dtype_as_field('int', 'sum', 'transformation')
+ 2.90±0.1ms 3.37±0.2ms 1.16 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(',', '_', 'high')
+ 1.74±0.02ms 2.02±0.08ms 1.16 timeseries.ResampleSeries.time_resample('period', '1D', 'ohlc')
+ 9.39±0.2μs 10.9±0.07μs 1.16 timestamp.TimestampProperties.time_is_quarter_end(None, 'B')
+ 60.2±1ms 69.7±5ms 1.16 frame_ctor.FromDicts.time_nested_dict_int64
+ 205±6ms 238±3ms 1.16 strings.Split.time_split(True)
+ 120±2ms 138±6ms 1.16 replace.Convert.time_replace('Series', 'Timedelta')
+ 144±2μs 166±3μs 1.15 groupby.GroupByMethods.time_dtype_as_group('datetime', 'cumcount', 'transformation')
+ 6.11±0.1ms 7.04±0.05ms 1.15 strings.Cat.time_cat(0, ',', '-', 0.001)
+ 2.14±0.02ms 2.46±0.04ms 1.15 binary_ops.Ops.time_frame_comparison(True, 1)
+ 227±20μs 261±4μs 1.15 groupby.GroupByMethods.time_dtype_as_group('int', 'var', 'transformation')
+ 1.56±0.01ms 1.79±0.01ms 1.15 stat_ops.SeriesMultiIndexOps.time_op(0, 'mean')
+ 2.80±0.1ms 3.21±0.02ms 1.15 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(';', '_', 'high')
+ 2.20±0.03ms 2.52±0.01ms 1.14 series_methods.IsIn.time_isin('object')
+ 2.94±0.05ms 3.36±0.2ms 1.14 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(',', '_', None)
+ 40.2±0.3ms 46.0±0.3ms 1.14 algorithms.Factorize.time_factorize_float(True)
+ 2.89±0.09ms 3.29±0.05ms 1.14 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(';', '_', None)
+ 9.74±0.4μs 11.1±0.2μs 1.14 timestamp.TimestampProperties.time_is_month_start(None, 'B')
+ 2.03±0.03ms 2.31±0.09ms 1.14 binary_ops.Timeseries.time_series_timestamp_compare('US/Eastern')
+ 552±10μs 628±10μs 1.14 ctors.SeriesConstructors.time_series_constructor(<class 'list'>, False)
+ 18.9±0.3ms 21.5±0.5ms 1.14 reindex.DropDuplicates.time_frame_drop_dups_na(False)
+ 208±8μs 237±5μs 1.14 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<MonthBegin>)
+ 217±5μs 246±5μs 1.14 offset.OffsetDatetimeIndexArithmetic.time_add_offset(<MonthEnd>)
+ 5.24±0.3ms 5.94±0.3ms 1.13 io.csv.ReadCSVDInferDatetimeFormat.time_read_csv(True, 'custom')
+ 519±10μs 588±50μs 1.13 indexing.DataFrameNumericIndexing.time_bool_indexer
+ 257±20μs 291±30μs 1.13 groupby.GroupByMethods.time_dtype_as_field('int', 'prod', 'direct')
+ 3.65±0.1ms 4.13±0.9ms 1.13 binary_ops.Ops.time_frame_comparison(True, 'default')
+ 3.57±0.1ms 4.03±0.3ms 1.13 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(',', '.', 'round_trip')
+ 582±8μs 655±10μs 1.13 ctors.SeriesConstructors.time_series_constructor(<class 'list'>, True)
+ 1.57±0.01ms 1.77±0.07ms 1.13 reshape.SimpleReshape.time_unstack
+ 2.05±0.04ms 2.31±0.02ms 1.13 stat_ops.FrameMultiIndexOps.time_op(1, 'mean')
+ 2.44±0.07ms 2.74±0.03ms 1.13 groupby.CountMultiInt.time_multi_int_count
+ 409±7μs 460±5μs 1.12 timeseries.DatetimeIndex.time_unique('repeated')
+ 9.70±0.3μs 10.9±0.1μs 1.12 timestamp.TimestampProperties.time_is_year_end(None, 'B')
+ 1.41±0.08μs 1.57±0.02μs 1.12 timestamp.TimestampConstruction.time_parse_today
+ 434±10μs 486±30μs 1.12 frame_methods.Quantile.time_frame_quantile(0)
+ 9.86±0.2μs 11.0±0.2μs 1.12 timestamp.TimestampProperties.time_is_quarter_start(None, 'B')
+ 3.47±0.2ms 3.88±0.04ms 1.12 io.csv.ReadCSVFloatPrecision.time_read_csv_python_engine(';', '.', 'round_trip')
+ 9.06±0.1ms 10.1±0.1ms 1.12 io.hdf.HDFStoreDataFrame.time_query_store_table
+ 1.42±0.06μs 1.58±0.01μs 1.12 timestamp.TimestampConstruction.time_parse_now
+ 208±7μs 232±2μs 1.11 groupby.GroupByMethods.time_dtype_as_group('object', 'nunique', 'direct')
+ 11.8±0.07ms 13.0±0.5ms 1.11 index_object.Ops.time_modulo('float')
+ 1.62±0.02ms 1.79±0.03ms 1.11 timeseries.ResampleDatetetime64.time_resample
+ 320±10μs 355±5μs 1.11 groupby.GroupByMethods.time_dtype_as_group('int', 'nunique', 'transformation')
+ 1.03±0.01ms 1.14±0.03ms 1.11 replace.FillNa.time_replace(True)
+ 2.07±0.01ms 2.28±0.02ms 1.10 stat_ops.SeriesMultiIndexOps.time_op(0, 'std')
+ 2.87±0.06ms 3.17±0.2ms 1.10 stat_ops.SeriesMultiIndexOps.time_op(1, 'sem')
Speedups
- 4.52±0.02ms 4.10±0.04ms 0.91 frame_methods.NSort.time_nlargest_two_columns('last')
- 3.47±0.2ms 3.15±0.02ms 0.91 sparse.ArithmeticBlock.time_make_union(nan)
- 3.55±0.1ms 3.21±0.02ms 0.90 sparse.ArithmeticBlock.time_division(0)
- 28.7±0.9ms 25.9±0.1ms 0.90 groupby.Nth.time_series_nth_any('float32')
- 33.6±2μs 30.2±0.3μs 0.90 offset.OffestDatetimeArithmetic.time_subtract(<Day>)
- 4.84±0.4ms 4.34±0.04ms 0.90 timeseries.DatetimeAccessor.time_dt_accessor_normalize
- 4.86±0.07μs 4.34±0.03μs 0.89 timedelta.TimedeltaConstructor.time_from_np_timedelta
- 6.63±0.9μs 5.92±0.1μs 0.89 timedelta.TimedeltaConstructor.time_from_datetime_timedelta
- 4.29±0.2ms 3.83±0.02ms 0.89 timeseries.DatetimeIndex.time_normalize('repeated')
- 114±4ms 101±2ms 0.89 gil.ParallelGroupbyMethods.time_loop(4, 'sum')
- 62.6±2ms 55.6±0.5ms 0.89 io.sql.WriteSQLDtypes.time_to_sql_dataframe_column('sqlalchemy', 'datetime')
- 2.76±0.02ms 2.43±0.02ms 0.88 frame_methods.NSort.time_nlargest_one_column('last')
- 3.13±0.07ms 2.76±0.08ms 0.88 frame_methods.NSort.time_nsmallest_one_column('last')
- 801±50μs 705±10μs 0.88 offset.OffsetSeriesArithmetic.time_add_offset(<Day>)
- 1.43±0.06ms 1.25±0.01ms 0.87 stat_ops.SeriesOps.time_op('median', 'float', True)
- 42.2±2ms 36.8±0.8ms 0.87 io.sql.WriteSQLDtypes.time_to_sql_dataframe_column('sqlalchemy', 'float')
- 1.93±0.2μs 1.68±0.03μs 0.87 timedelta.TimedeltaConstructor.time_from_missing
- 17.4±1μs 15.1±0.6μs 0.87 offset.OffestDatetimeArithmetic.time_apply_np_dt64(<BusinessDay>)
- 45.0±0.9ms 39.0±0.4ms 0.87 io.sql.WriteSQLDtypes.time_to_sql_dataframe_column('sqlalchemy', 'bool')
- 6.10±0.05μs 5.28±0.04μs 0.86 timedelta.TimedeltaConstructor.time_from_unit
- 44.4±2ms 38.2±1ms 0.86 io.sql.WriteSQLDtypes.time_to_sql_dataframe_column('sqlalchemy', 'float_with_nan')
- 2.37±0.1ms 2.04±0.02ms 0.86 timeseries.DatetimeIndex.time_timeseries_is_month_start('repeated')
- 8.17±0.1ms 7.00±0.03ms 0.86 stat_ops.FrameOps.time_op('mean', 'float', 1, False)
- 41.5±0.5ms 35.3±0.2ms 0.85 io.sql.WriteSQLDtypes.time_to_sql_dataframe_column('sqlalchemy', 'int')
- 60.4±3ms 51.3±2ms 0.85 gil.ParallelGroupbyMethods.time_loop(2, 'mean')
- 21.3±0.2ms 17.9±0.06ms 0.84 timeseries.DatetimeIndex.time_normalize('tz_aware')
- 7.86±0.3μs 6.60±0.04μs 0.84 offset.OnOffset.time_on_offset(<BusinessMonthBegin>)
- 199±5ms 166±1ms 0.83 io.stata.Stata.time_read_stata('td')
- 8.60±0.7μs 7.08±0.06μs 0.82 offset.OnOffset.time_on_offset(<SemiMonthEnd: day_of_month=15>)
- 9.77±0.2ms 8.03±0.07ms 0.82 strings.Cat.time_cat(0, ',', None, 0.001)
- 5.11±0.1ms 4.18±0.07ms 0.82 frame_methods.NSort.time_nlargest_two_columns('first')
- 35.3±7ms 28.8±0.4ms 0.82 groupby.Nth.time_groupby_nth_all('object')
- 62.4±3ms 50.7±0.6ms 0.81 gil.ParallelGroupbyMethods.time_loop(2, 'prod')
- 10.3±0.09ms 8.34±0.1ms 0.81 io.hdf.HDFStoreDataFrame.time_store_info
- 1.61±0.5ms 1.30±0.03ms 0.80 stat_ops.SeriesOps.time_op('median', 'float', False)
- 15.7±1μs 12.6±0.1μs 0.80 timedelta.TimedeltaConstructor.time_from_components
- 63.1±5ms 50.3±1ms 0.80 gil.ParallelGroupbyMethods.time_loop(2, 'sum')
- 24.9±0.7μs 19.8±0.1μs 0.80 offset.OffestDatetimeArithmetic.time_subtract_10(<BusinessDay>)
- 9.84±1ms 7.80±0.1ms 0.79 io.sql.WriteSQLDtypes.time_read_sql_query_select_column('sqlalchemy', 'float_with_nan')
- 133±2μs 105±1μs 0.79 offset.OffestDatetimeArithmetic.time_apply(<CustomBusinessMonthBegin>)
- 129±6μs 100±2μs 0.78 offset.OffestDatetimeArithmetic.time_subtract_10(<CustomBusinessMonthBegin>)
- 141±5ms 109±0.8ms 0.78 offset.OffsetSeriesArithmetic.time_add_offset(<CustomBusinessMonthBegin>)
- 17.6±1μs 13.4±0.7μs 0.76 offset.OffestDatetimeArithmetic.time_add(<BusinessDay>)
- 21.5±2μs 16.2±0.4μs 0.75 timeseries.AsOf.time_asof_single('Series')
- 10.2±2μs 7.70±0.1μs 0.75 timeseries.AsOf.time_asof_single_early('Series')
- 97.1±1ms 71.4±4ms 0.74 frame_methods.Describe.time_series_describe
- 191±5ms 139±3ms 0.73 timeseries.DatetimeIndex.time_to_pydatetime('tz_aware')
- 20.1±0.2ms 14.6±0.6ms 0.73 algorithms.Factorize.time_factorize_int(True)
- 5.60±0.2μs 4.04±0.2μs 0.72 timeseries.DatetimeIndex.time_get('tz_naive')
- 149±10μs 107±0.8μs 0.72 offset.OffestDatetimeArithmetic.time_apply_np_dt64(<CustomBusinessMonthBegin>)
- 16.0±4ms 11.4±0.09ms 0.71 groupby.Nth.time_series_nth('datetime')
- 238±20ms 168±3ms 0.71 io.stata.Stata.time_read_stata('tc')
- 9.71±0.8ms 6.86±0.4ms 0.71 groupby.Categories.time_groupby_ordered_nosort
- 44.0±5ms 30.9±0.2ms 0.70 plotting.TimeseriesPlotting.time_plot_irregular
- 134±10μs 94.2±0.9μs 0.70 offset.OffestDatetimeArithmetic.time_subtract(<CustomBusinessMonthBegin>)
- 321±5ms 225±10ms 0.70 frame_methods.Describe.time_dataframe_describe
- 15.9±4ms 11.1±0.1ms 0.70 groupby.Nth.time_series_nth('float64')
- 11.3±0.3ms 7.93±0.02ms 0.70 inference.DateInferOps.time_subtract_datetimes
- 1.38±0.02ms 961±10μs 0.69 stat_ops.SeriesOps.time_op('median', 'int', True)
- 11.4±2ms 7.79±0.1ms 0.68 timeseries.AsOf.time_asof('DataFrame')
- 290±5ns 197±4ns 0.68 timedelta.TimedeltaProperties.time_timedelta_days
- 430±60ns 291±1ns 0.68 indexing.MethodLookup.time_lookup_loc
- 22.7±0.4μs 15.3±0.2μs 0.68 offset.OffestDatetimeArithmetic.time_add_10(<YearBegin: month=1>)
- 1.45±0.06ms 975±10μs 0.67 stat_ops.SeriesOps.time_op('median', 'int', False)
- 5.88±0.4μs 3.92±0.06μs 0.67 timeseries.DatetimeIndex.time_get('dst')
- 2.46±0.03ms 1.64±0.01ms 0.66 groupby.RankWithTies.time_rank_ties('float32', 'first')
- 2.54±0.02ms 1.67±0.03ms 0.66 groupby.RankWithTies.time_rank_ties('float64', 'dense')
- 2.51±0.3ms 1.65±0.03ms 0.65 groupby.RankWithTies.time_rank_ties('int64', 'max')
- 2.50±0.08ms 1.64±0.01ms 0.65 groupby.RankWithTies.time_rank_ties('float64', 'first')
- 51.0±2μs 33.3±4μs 0.65 categoricals.IsMonotonic.time_categorical_series_is_monotonic_decreasing
- 2.53±0.01ms 1.64±0.01ms 0.65 groupby.RankWithTies.time_rank_ties('float64', 'min')
- 13.8±2ms 8.96±0.1ms 0.65 strings.Cat.time_cat(0, ',', None, 0.15)
- 2.58±0.07ms 1.67±0.01ms 0.65 groupby.RankWithTies.time_rank_ties('float32', 'dense')
- 107±2ms 69.3±1ms 0.65 index_object.IndexAppend.time_append_int_list
- 11.9±1ms 7.63±0.2ms 0.64 frame_methods.ToString.time_to_string_floats
- 2.56±0.1ms 1.63±0.01ms 0.64 groupby.RankWithTies.time_rank_ties('float32', 'average')
- 2.77±0.3ms 1.75±0.06ms 0.63 rolling.Quantile.time_quantile('Series', 10, 'int', 1, 'higher')
- 2.56±0.7ms 1.62±0ms 0.63 groupby.RankWithTies.time_rank_ties('datetime64', 'max')
- 6.92±0.3ms 4.37±0.03ms 0.63 stat_ops.FrameOps.time_op('median', 'int', 0, True)
- 312±3ns 196±1ns 0.63 timedelta.TimedeltaProperties.time_timedelta_microseconds
- 6.93±0.3ms 4.33±0.05ms 0.62 stat_ops.FrameOps.time_op('median', 'int', 0, False)
- 2.64±0.1ms 1.65±0.01ms 0.62 groupby.RankWithTies.time_rank_ties('float64', 'max')
- 2.69±0.2ms 1.66±0.01ms 0.62 groupby.RankWithTies.time_rank_ties('int64', 'first')
- 2.65±0.08ms 1.64±0.01ms 0.62 groupby.RankWithTies.time_rank_ties('float32', 'min')
- 31.0±1μs 19.2±0.9μs 0.62 offset.OffestDatetimeArithmetic.time_subtract_10(<QuarterEnd: startingMonth=3>)
- 2.67±0.1ms 1.65±0.01ms 0.62 groupby.RankWithTies.time_rank_ties('float64', 'average')
- 3.45±0.09ms 2.13±0.05ms 0.62 ctors.SeriesConstructors.time_series_constructor(<function SeriesConstructors.<lambda> at 0x1136817b8>, True)
- 474±100ms 290±5ms 0.61 indexing.NonNumericSeriesIndexing.time_getitem_list_like('datetime', 'nonunique_monotonic_inc')
- 26.6±1μs 16.2±0.2μs 0.61 offset.OffestDatetimeArithmetic.time_subtract(<YearBegin: month=1>)
- 2.64±0.1ms 1.61±0.01ms 0.61 groupby.
|
|
There seem to be some jumps in some of the benchmarks since the last run, eg http://pandas.pydata.org/speed/pandas/#algorithms.Factorize.time_factorize_float? (and it also does not show a big difference in the output above of @mroeschke ) |
|
I think we'll close this in favor of #30790 (0.25 -> 1.0). |
4 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
xref
Speedups
The text was updated successfully, but these errors were encountered: