[PATCH] Simple sqrt implementation for sphereflake

Daniel Dunbar daniel at zuster.org
Thu May 8 10:27:10 PDT 2014

This seems reasonable, given the primary intent of these tests. The only
problem I see with this is that there is actually some value in testing the
compiler's handling of the real sqrt(), but that is something probably more
properly done by specific functional tests written by someone with a very
good idea of exactly what is allowed.

 - Daniel

On Thu, May 8, 2014 at 4:34 AM, Renato Golin <renato.golin at linaro.org>wrote:

> Hi ddunbar,
> Reports about different results depending on which GCC was used to run the
> test-suite have prompted me to implement sqrt() as a simple function to
> avoid library issues. The results on x86_64, ARM and AArch64 are identical,
> even across Clang and GCC 4.8.
> The new implementation should be safe for values between 10^-10 and 10^10.
> For this particular case, though, it's more than enough.
> The images remained pretty. ;)
> http://reviews.llvm.org/D3667
> Files:
>   SingleSource/Benchmarks/Misc-C++/Large/sphereflake.cpp
>   SingleSource/Benchmarks/Misc-C++/Large/sphereflake.reference_output
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140508/4dc806c8/attachment.html>

More information about the llvm-commits mailing list