This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Set fast unaligned load flag for new Intel microarchitecture.
- From: Dmitrieva Liubov <liubov dot dmitrieva at gmail dot com>
- To: Andreas Jaeger <aj at suse dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Fri, 14 Jun 2013 18:53:14 +0400
- Subject: Re: [PATCH] Set fast unaligned load flag for new Intel microarchitecture.
- References: <CAHjhQ92vybopdpv02o+QtW3Gq++nzbb3VJ=+pCzZG3cDMmLokg at mail dot gmail dot com> <51BAFAD7 dot 90603 at suse dot com>
I fixed:
2013-06-14 Liubov Dmitrieva <liubov.dmitrieva@intel.com>
* sysdeps/x86_64/multiarch/init-arch.c (__init_cpu_features):
Set bit_Fast_Unaligned_Load for Intel Silvermont architecture
diff --git a/sysdeps/x86_64/multiarch/init-arch.c b/sysdeps/x86_64/multiarch/ini
index 7daaf46..e10f0f6 100644
--- a/sysdeps/x86_64/multiarch/init-arch.c
+++ b/sysdeps/x86_64/multiarch/init-arch.c
@@ -77,7 +77,12 @@ __init_cpu_features (void)
/* BSF is slow on Atom. */
__cpu_features.feature[index_Slow_BSF] |= bit_Slow_BSF;
break;
-
+ case 0x37:
+ /* Unaligned load versions are faster than SSSE3
+ on Silvermont. */
+ __cpu_features.feature[index_Fast_Unaligned_Load]
+ |= bit_Fast_Unaligned_Load;
+ break;
default:
/* Unknown family 0x06 processors. Assuming this is one
of Core i3/i5/i7 processors if AVX is available. */
Should I commit this myself? I have never done the commit myself.
--
Liubov Dmitrieva
Intel Corporation
On Fri, Jun 14, 2013 at 3:13 PM, Andreas Jaeger <aj@suse.com> wrote:
> On 06/14/2013 12:35 PM, Dmitrieva Liubov wrote:
>>
>> I have small patch for new Intel Silvermont machines.
>>
>>
>> http://newsroom.intel.com/community/intel_newsroom/blog/2013/05/06/intel-launches-low-power-high-performance-silvermont-microarchitecture
>>
>> Ok for 2.18?
>>
>> I checked this on my machine and see that strcpy, ... unaligned
>> versions are faster than ssse3 versions.
>>
>> ChangeLog:
>>
>> 2013-06-14 Liubov Dmitrieva <liubov.dmitrieva@intel.com>
>>
>> * .sysdeps/x86_64/multiarch/init-arch.c: Set
>> bit_Fast_Unaligned_Load for new
>> Intel Silvermont architecture.
>
>
> This needs proper formatting, please check before committing - and name the
> function you change like:
> * sysdeps/x86_64/multiarch/init-arch.c (__init_cpu_features):
>
>
>>
>> diff --git a/sysdeps/x86_64/multiarch/init-arch.c
>> b/sysdeps/x86_64/multiarch/ini
>> index 7daaf46..23b1b98 100644
>> --- a/sysdeps/x86_64/multiarch/init-arch.c
>> +++ b/sysdeps/x86_64/multiarch/init-arch.c
>> @@ -77,7 +77,11 @@ __init_cpu_features (void)
>> /* BSF is slow on Atom. */
>> __cpu_features.feature[index_Slow_BSF] |= bit_Slow_BSF;
>> break;
>> -
>> + case 0x37:
>> + /* Silvermont. */
>
>
> Say in the comment that unaligned versions are faster than SSSE3.
>
> Otherwise fine,
> Andreas
>
>
>> + __cpu_features.feature[index_Fast_Unaligned_Load]
>> + |= bit_Fast_Unaligned_Load;
>> + break;
>> default:
>> /* Unknown family 0x06 processors. Assuming this is one
>> of Core i3/i5/i7 processors if AVX is available. */
>>
>>
>>
>> --
>> Liubov Dmitrieva
>> Intel Corporation
>>
>
>
> --
> Andreas Jaeger aj@{suse.com,opensuse.org} Twitter/Identica: jaegerandi
> SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
> GF: Jeff Hawn,Jennifer Guild,Felix Imendörffer,HRB16746 (AG Nürnberg)
> GPG fingerprint = 93A3 365E CE47 B889 DF7F FED1 389A 563C C272 A126