Hi,
2011/10/18 Satoru KURASHIKI <[email protected]>:
> hi,
>
> On Tue, Oct 18, 2011 at 10:21 AM, Nobuhiro Iwamatsu
> <[email protected]> wrote:
>> Because these architecture does not support "-march=native".
>> I made patch which revise this problem. Could you check and apply this patch?
>
> yes, and I plan to contact upstream to fix this (because configure script
> seems
> to be broken) suggesting a patch as follows:
>
> --- configure.orig 2011-10-18 00:56:16.033903126 +0900
> +++ configure 2011-10-18 01:26:46.366979251 +0900
> @@ -298,9 +298,10 @@
> }
>
> {
> + $_ = `uname -m 2>&1`;
> my $common = '';
> if (GccIsSupport($global{'gcc_version'}, 4, 2, 0)) {
> - $common .= '-march=native';
> + $common .= '-march=native' if /86/i;
> } else {
> $common .= '-march=k8' if /x86_64/i;
> }
>
> Your patch removes this if-clause entirely, which drops optimize
> options even in x86 architecture. It seems to make a (little bit)
> drastic cut, doesn't it?
Well, '-march=native' option varies in an option of the optimization
by building environment.
For example, the following optimization is made when we use this
option in C2Q and C2D.
C2D:
-march=core2 -mcx16 -msahf -mno-movbe -mno-aes -mno-pclmul
-mno-popcnt -mno-abm -mno-lwp -mno-fma -mno-fma4 -mno-xop -mno-bmi
-mno-tbm -mno-avx -mno-sse4.2 -mno-sse4.1 --param l1-cache-size=32
--param l1 -cache-line-size=64 --param l2-cache-size=4096
-mtune=core2
C2Q:
-march=core2 -mcx16 -msahf -mno-movbe -mno-aes -mno-pclmul
-mno-popcnt -mno-abm -mno-lwp -mno-fma -mno-fma4 -mno-xop -mno-bmi
-mno-tbm -mno-avx -mno-sse4.2 -msse4.1 --param l1-cache-size=32
--param l1-ca che-line-size=64 --param l2-cache-size=3072
-mtune=core2
When a binary optimized by the latest CPU is built, user may not work
with the old CPU.
Best regards,
Nobuhiro
--
Nobuhiro Iwamatsu
iwamatsu at {nigauri.org / debian.org}
GPG ID: 40AD1FA6
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
