On 7/27/2014 9:44 AM, Ruben Van Boxem wrote:
> Hi guys,
>
> I stumbled upon a strange question on Stackoverflow.com:
> http://stackoverflow.com/q/24962205/256138
>
> Is he right to think the generated code is suboptimal? Or is this an
> ABI requirement? Can GCC be improved in this regard?
Unless
On 7/27/2014 23:44, Ruben Van Boxem wrote:
> Hi guys,
>
> I stumbled upon a strange question on Stackoverflow.com:
> http://stackoverflow.com/q/24962205/256138
>
> Is he right to think the generated code is suboptimal? Or is this an ABI
> requirement? Can GCC be improved in this regard?
>
> I ca
On 7/27/2014 23:50, André Hentschel wrote:
> Am 26.07.2014 um 01:54 schrieb JonY:
>> On 7/26/2014 06:20, André Hentschel wrote:
>>> +double log2(double x) +{ +return softmath_log(x) / softmath_log(2); +}
>
>> Any reason not to expand softmath_log(2) into a constant? Otherwise, it
>> looks OK.
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Am 26.07.2014 um 01:54 schrieb JonY:
> On 7/26/2014 06:20, André Hentschel wrote:
>> +double log2(double x) +{ +return softmath_log(x) / softmath_log(2); +}
>
> Any reason not to expand softmath_log(2) into a constant? Otherwise, it looks
> OK.
>
Hi guys,
I stumbled upon a strange question on Stackoverflow.com:
http://stackoverflow.com/q/24962205/256138
Is he right to think the generated code is suboptimal? Or is this an ABI
requirement? Can GCC be improved in this regard?
I can point him to this mailing list and ask him to supply reprod
And while I think it would be a good idea to add this function to
intrin-impl.h so that this "intrinsic" function would be, well,
intrinsic (instead of imported), that seems like a discussion for
another day.
Hey, it's another day...
I originally thought it would make sense to do this incre
