Update asv config + fix some broken benchmarks #12563
Conversation
pv
commented
Mar 8, 2016
pv
commented
- Enable platform-dependent config in asv (needs asv git version for it to do something)
- Enable wheel cache in asv (in asv git version)
- Fix a few easily fixed broken benchmarks
| self.df = DataFrame(np.random.randn(20000, 100)) | ||
| self.df2 = DataFrame(np.random.randn(20000, 100)) | ||
| self.df3 = DataFrame(np.random.randn(20000, 100)) | ||
| self.df4 = DataFrame(np.random.randn(20000, 100)) | ||
|
|
||
| def time_eval_frame_chained_cmp_python(self): | ||
| pd.eval('df < df2 < df3 < df4', engine='python') | ||
| if threads == 1: |
There was a problem hiding this comment.
hmm maybe should set this explicity each time something is run to 1 (then you can set in this particular one to higher)
There was a problem hiding this comment.
I'm not sure what would be a good value for threads=='all', could you explain how you'd like it. Note each benchmark is run in its own process, so those with threads!=1 get the default value on pandas startup.
There was a problem hiding this comment.
oh, each is in own thread, then ok. I just didn't want to set a 'global', but if spinning off proc then its fine.
|
can you rebase this. we can prob throw this in 0.18.0; it doesn't affect user code. Can you run vs 0.17.1 and post anything interesting? |
|
Rebased. |
|
benchmarks: https://gist.github.com/pv/8ff97c3f6faa62fd6a5e --- not sure if all of those are real, or fluctuations (laptop under load) |
|
I think we need to update the perf docs in contributing? |
|
Not sure --- they already instruct to install the asv git version. |
| // are installed. The 'python' key is required. The exclude rules | ||
| // do not apply to includes. | ||
| // | ||
| // In addition to package names, the following keys are available: |
There was a problem hiding this comment.
so how does one choose the installation method / use this, ideally would ha e examples for conda (default) / pip
Add requirements rules for conda and virtualenv (requires asv git version). Enable asv wheel cache.
- eval.eval_frame_*: fix up local variables - groupby: range() -> list(range()) and using np.random.choice - plotting: ensure matplotlib Agg backend is used - packers: fix data file paths
|
Added explanation on how to choose between conda / virtualenv / existing environment to contributer docs. |
|
thanks @pv pls check the docs when built (prob a few hours) and if necessary issue a follow-up |
|
should these be ignored as well? |
|
sorry....these look like mine. nvm! |
