vbench (?) .ix slowness

[wesm]

i got this email and didn't know if this was a perf regression we should care about

Wes,

I'm stuck on something and I was hoping for your help.

I wrote a function a year back that iterates element by element through a
DataFrame of about 1200 columns by 2000 rows.  It runs some basic operations,
but nothing crazy.  It used to take something like 5 seconds to complete, and
now maybe 100x that.  The only thing I can think of that's changed is my pandas
version (and it's dependancies such as numpy).

You can see my stackoverflow question about the same here:
http://stackoverflow.com/questions/17538902/is-there-a-way-to-speed-up-this-pandas-function

At the bottom of my question is a simple example function I just created (to
try and isolate the problem) that seems to take very long to complete (73
seconds) on my computer.  Does this sound right to you? Wondering if you have
any thoughts, as I'm not sure what to do now.  I know vectorized functions are
preferred, I have certain cases that require elementwise functions and would
like to figure out a way to get decent speed.

Thanks for any help!

[jreback]

pretty sure this was covered in #4198. .ix perf didn't regress from 0.11-0.12 at all (nor did we really touch it much). That said, if the OP would provide a complete example w/data I'll take a look (aside from the fact that this entire operation can be vectorized)

[jreback]

closing as no regression