Re: [PATCH 00/10] Optimized math routines

[Szabolcs Nagy <szabolcs.nagy@arm.com>]

On 06/07/18 17:27, Carlos O'Donell wrote:
> On 07/06/2018 11:46 AM, Szabolcs Nagy wrote:
>> On 06/07/18 13:43, Carlos O'Donell wrote:
>>> On 07/06/2018 04:47 AM, Szabolcs Nagy wrote:
>>>> Optimized exp, exp2, log, log2, pow, sinf, cosf and sincosf
>>>> implementations.
>>>
>>> Is it your intent to have these included in 2.28?
>>>
>>
>> (resending as my previous mail seems to be lost)
>>
>> yes, i'd like to add it to the 'desirable in 2.28' list
>> if Joseph is ok with the code, but i see he is not available
>> right now for review.
>>
>> i don't know how other maintainers feel about such change,
>> there needs to be an ulp update (i'm willing to do that for
>> targets i can access hw for testing).
> 
> Where there any unanswered questions in your v4 review?
> 
> Do you think v4 is basically as good as it will get?
> 
> Who were the people who signed off on the review?
> 

i made two minor modifications (codegen is not affected):
- removed an unused configuration option from sincosf.
- fixed a comment in pow.

i ran glibc tests on powerpc64le-linux-gnu and i686-linux-gnu
too now.

i will also update the improvement numbers in the commit messages
to reflect dynamic linked master vs new instead of static linked
master vs nowrapper improvements (i didn't repost the patchset
just to change those numbers though).

i propose this change to be included in glibc 2.28 (added it to
the desired features), i'm willing to do further tweaks if
necessary.

i believe the code provides the documented math quality
guarantees and does not introduce regressions on any target
(other than ulp bound update is needed after the exp patch).

the wrapper removal is deferred to glibc 2.29 (it requires
some changes to the libm alias machinery, it changes abi
and target maintainers may need to tweak the ifunc mechanism
where appropriate for optimal performance)