Skip to content

Conversation

@jreback
Copy link
Contributor

@jreback jreback commented Jul 7, 2014

closes #7633

close to what it was in 0.14.0
key was to not keep recomputing whether an index hasnans every time we need it (it is now cached).
further min/max are optimized if the index is monotonic

------------------------------------------------------------------------------- Test name | head[ms] | base[ms] | ratio | ------------------------------------------------------------------------------- timeseries_timestamp_downsample_mean | 4.5697 | 7.8034 | 0.5856 | dataframe_resample_min_string | 1.8380 | 2.5294 | 0.7266 | dataframe_resample_min_numpy | 1.8580 | 2.5463 | 0.7297 | dataframe_resample_max_numpy | 1.8887 | 2.5803 | 0.7320 | dataframe_resample_max_string | 1.9130 | 2.5553 | 0.7486 | dataframe_resample_mean_numpy | 2.6687 | 3.3340 | 0.8004 | dataframe_resample_mean_string | 2.7773 | 3.3080 | 0.8396 | timeseries_period_downsample_mean | 12.2183 | 11.6010 | 1.0532 | Ratio < 1.0 means the target commit is faster then the baseline. Seed used: 1234 Target [d2d30c7] : PERF: better perf on min/max on indices not containing NaT for DatetimeIndex/PeriodIndex Base [e060616] : DOC: minor corrections in v0.14.1 
@jreback jreback added this to the 0.14.1 milestone Jul 7, 2014
jreback added a commit that referenced this pull request Jul 7, 2014
PERF: better perf on min/max on indices not containing NaT for DatetimeIndex/PeriodsIndex
@jreback jreback merged commit d6eace7 into pandas-dev:master Jul 7, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Performance Memory or execution speed performance

1 participant