Re: [PATCH 1/N] x86_64 vectorization support: vectorized math functions addition to Glibc

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Fri, Sep 12, 2014 at 12:13 PM, Carlos O'Donell <carlos@redhat.com> wrote:
> On 09/12/2014 03:05 PM, H.J. Lu wrote:
>> On Fri, Sep 12, 2014 at 8:10 AM, Carlos O'Donell <carlos@redhat.com> wrote:
>>> On 09/12/2014 11:01 AM, H.J. Lu wrote:
>>>> On Thu, Sep 11, 2014 at 5:10 PM, Carlos O'Donell <carlos@redhat.com> wrote:
>>>>> On 09/11/2014 04:57 PM, H.J. Lu wrote:
>>>>>>> That doesn't answer my question.  Maybe glibc 2.21 provides such versions
>>>>>>> for all x86 ISAs there are at present, up to AVX512 - and then a new
>>>>>>> extension AVX1024 appears.  When GCC 7 is used with glibc 2.21 headers and
>>>>>>> -mavx1024, it must not try to generate calls to the AVX1024 functions,
>>>>>>> because glibc 2.21 doesn't have such functions.  But maybe glibc 2.26 adds
>>>>>>> the AVX1024 functions.  So something needs to be different in the headers
>>>>>>> of 2.26 to inform GCC 7 that AVX1024 versions of the functions are
>>>>>>> available.  And I think that means the directive that communicates
>>>>>>> function availability to the compiler needs to identify the set of ISAs
>>>>>>> for which versions of the function in question are available.
>>>>>>>
>>>>>>
>>>>>> Wouldn't it be better to put libmvec in GCC instead?
>>>>>
>>>>> That's certainly a discussion we can have.
>>>>>
>>>>> What do you see as the pros and cons?
>>>>>
>>>>
>>>> It depends on who are the main target users of this library.
>>>> If it is mainly for programmers to use them directly in their
>>>> applications, mostly independent of compilers, it should be
>>>> in glibc.  But if it is mainly used by GCC, it should be in
>>>> GCC, just like other run-time libraries.
>>>
>>> The former. I want users to be able to call these functions
>>> directly regardless of the compiler. The same goes for the
>>> ppc-related API that has been in use for a long time by
>>> developers there.
>>>
>>> The compiler can certainly make use of these functions, and
>>> any more standard cross-machine GNU API we design, but it
>>> should always be possible to call them directly.
>>>
>>> Does this mean libmvec should be in glibc?
>>>
>>
>> If the target users are programmers,  we should make it easier
>> to use for programmers.  We can provide a generic API with a
>> generic implementation.  Each target can provide an optimized
>> version which is transparent to users.  We can use IFUNC to
>> select the best version at run-time.
>
> I think such an implementation is orthogonal to exposing the
> already documented vector functions supported by Intel?
>
> Similarly for IBM.
>
> First a foremost we should support users who are expecting
> to be able to call the functions Intel and IBM have already
> documented.
>
> Second to that we can create a new API?
>
> Note that the generic API might by very difficult to design,
> which is why I don't suggest we tackle it first.
>

I think it is the best solution for our users.


-- 
H.J.