acos-inputs
Paul Zimmermann
Paul.Zimmermann@inria.fr
Thu Aug 12 08:35:33 GMT 2021
Dear Siddhesh,
> If one had to think in an algorithm-agnostic manner, the one factor that
> typically determines performance of such functions is the precision
> required to compute an accurate result for a given input. A set of
> inputs such that the precision required to compute f(x) is a uniform
> distribution in the minimum and maximum precision required to compute
> f(x) may tell us more about the performance of the function.
assume the maximum precision is 10^6 bits, and is needed for only one
input (out of the 2^64 possible for double precision). If you consider
this input, the result of "make bench" will be highly biased, although
in real application this case will never happen.
> Do you think something like this makes sense? I don't know if
> performance analysis of libm functions based on precision requirements
> has been done before, so maybe there is literature out there that may
> have dealt with this already.
I'm not aware of such literature. Anyway, I will use my own set of inputs
for my application, since the glibc ones give weird results.
Best regards,
Paul
More information about the Libc-alpha
mailing list