This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] A new meta intrinsic header file for current and future x86 instrinsics.
- From: "Richard Guenther" <richard dot guenther at gmail dot com>
- To: "rajagopal, dwarak" <dwarak dot rajagopal at amd dot com>
- Cc: "Uros Bizjak" <ubizjak at gmail dot com>, "Guo, Xuepeng" <xuepeng dot guo at intel dot com>, gcc-patches at gcc dot gnu dot org, jakub at redhat dot com, "Lu, Hongjiu" <hongjiu dot lu at intel dot com>, "Ye, Joey" <joey dot ye at intel dot com>, "Lin, Weiliang" <weiliang dot lin at intel dot com>, "Harle, Christophe" <christophe dot harle at amd dot com>, "Sebastian Pop" <sebpop at gmail dot com>
- Date: Wed, 19 Nov 2008 21:44:34 +0100
- Subject: Re: [PATCH] A new meta intrinsic header file for current and future x86 instrinsics.
- References: <820531547ADB6847AF89CC220F6128F1737F@pdsmsx001.ccr.corp.intel.com> <4923023C.3090106@gmail.com> <9E1304B144EBEB4C97F4162BFAC4788602D3A6CD@SAUSEXMB2.amd.com>
On Wed, Nov 19, 2008 at 9:30 PM, rajagopal, dwarak
<dwarak.rajagopal@amd.com> wrote:
>>
>> If there are no further comments or objections from AMD people in the
>> next 48 hours, and since this header file is intended for _Intel_
>> intrinsics, it is OK for mainline.
>>
>> BTW: I guess there will be a header file (ammintrin.h ?) introduced to
>> include AMD defined intrinsics, so gcc can still add x86intrin.h that
>> will include both immintrin.h and eventual ammintrin.h. Hopefully,
> these
>> two intrinsics won't step onto each other toes.
>>
>> I hope this eventual solution is acceptable to both contributors.
>>
>> Thanks,
>> Uros.
>>
>
> Hi,
>
> This proposal is not acceptable to AMD for the following reasons:
>
> - GCC is an open source compiler that is vendor agnostic. We do not want
> to see vendor specific intrinsic files, as the ISAs are architecture
> specific primarily. This is how it has been till now (eg: emmintrin.h,
> xmmintrin.h, etc). Also there are many examples when AMD adopted SSE
> instructions after Intel created them, and Intel adopted AMD
> instructions after AMD came up with them (latest example: popcnt).
>
> - We recommend that there is only 1 intrinsic file used for all x86
> intrinsics. This seemed to have been agreed upon
> (http://gcc.gnu.org/ml/gcc/2008-04/msg00081.html). We prefer that this
> file be neutral and vendor agnostic, which immintrin.h is not.
I agree. Providing this headerfile is of no use for GCC users - it constrains
them to Intel CPU features. Dwarak, can you based on the ideas in immintrin.h
provide an implementation of x86intrin.h we can add to GCC?
As a note to Intel folks, as the issue has been raised some time ago sticking
to the immintrin.h plan shows quite some disrespect to the GCC community.
Thanks,
Richard.