This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 28/28] powerpc: refactor logb{f,l}
- From: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- To: "Gabriel F. T. Gomes" <gabriel at inconstante dot eti dot br>
- Cc: libc-alpha at sourceware dot org
- Date: Mon, 8 Jul 2019 15:31:23 -0300
- Subject: Re: [PATCH 28/28] powerpc: refactor logb{f,l}
- References: <20190329133529.22523-1-adhemerval.zanella@linaro.org> <20190329133529.22523-29-adhemerval.zanella@linaro.org> <20190705012404.GD7290@tereshkova>
On 04/07/2019 22:24, Gabriel F. T. Gomes wrote:
> On Fri, Mar 29 2019, Adhemerval Zanella wrote:
>> The power7 logb implementation does not show a performance gain on
>> ISA 2.07+ chips with faster floating-point to GRP instructions
>> (currently POWER8 and POWER9).
>
> Looks good to me. Thanks.
>
> Reviewed-by: Gabriel F. T. Gomes <gabrielftg@linux.ibm.com>
>
>> - asm (
>> - "xxland %x0,%x1,%x2\n"
>> - "fcfid %0,%0"
>> - : "=f" (ret)
>> - : "f" (x), "f" (mask.d));
>> + asm ("xxland %x0,%x1,%x2\n"
>> + "fcfid %0,%0"
>> + : "=d" (ret)
>> + : "d" (x), "d" (0x7ff0000000000000ULL));
>
> OK. I don't actually understand when the 'f' constraint is supposed to
> be used, but 'd' looks good to me.
>
>> -static const double two7m1 = -127.0; /* -2**7 + 1 */
>> [...]
>> - if (__builtin_expect (ret > -two7m1, 0))
>> + if (ret > 127.0)
>
> OK. Double negative.
>
>> --- a/sysdeps/powerpc/powerpc64/fpu/multiarch/Makefile
>> +++ /dev/null
>
> Hurray. :)
>
> Thanks, Adhemerval.
>
> Sorry for taking too long to review this patch set (I tried to review
> them all before the freeze, but it wasn't so easy for me).
>
If no one opposes it, I will commit this shortly.