[Numpy-discussion] Numpy and PEP 343
Travis Oliphant
oliphant at ee.byu.edu
Fri Mar 3 15:45:02 EST 2006
David M. Cooke wrote:
>cookedm at physics.mcmaster.ca (David M. Cooke) writes:
>
>
>>Yeah, I was going to stare at the assembly output from gcc before
>>deciding.
>>
>>Note that some functions that are supported directly by the CPU may
>>still be a function call because they aren't good enough.
>>
>>
>
>... and a quick test shows me that with gcc 4 under linux, only sqrt
>is inlined (and it does a test for NaN (I think) and calls the C
>library sqrt on that). On PowerPC under OS X, even sqrt is a function
>call.
>
>
There were some posts earlier showing that numarray had faster sin(x)
functions on an INTEL platform (with a particular compiler that I'm not
sure was specified). I'm pretty sure the difference was the function
call versus the inline sin computation that the compiler used.
I bet it depends on the compiler. But, you bring up a good point that
the on-chip provided functions might not be "math-lib" worthy according
to different compilers. It would be nice if there was still some way to
use them though letting the user decide if they were "good enough"
-Travis
More information about the NumPy-Discussion
mailing list