[gentoo-user] using icc with portage

[gentoo-user] using icc with portage

[Stefan G. Weichinger]

Anyone else using Intel's compiler, icc?

I do for quite a while now.
Followed http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage

I still prefer gcc over icc so I use

http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage#.2Fetc.2Fportage.2Fbashrc

to only use icc for stuff I list in /etc/portage/package.icc

-

Lately I get compilation-errors for packages that aren't listed in that
file, for example dev-python/numpy-1.6.1

It fails with log-lines telling me that icc was used to (try to) compile it:

[..]
icc: command line warning #10156: ignoring option '-fp'; no argument
required
icc: error #10236: File not found:  'l1-cache-line-size=64'
icc: command line warning #10156: ignoring option '-fp'; no argument
required
icc: error #10236: File not found:  'l2-cache-size=4096'
error: Command "icc -pthread -shared -Wl,-O1 -Wl,--as-needed -shared -O2
-D_FORTIFY_SOURCE=2 -march=core2 -mcx16 -msahf --param l1-cache-size=32
--param l1-cache-line-size=64 --param l2-cache-size=4096 -mtune=core2
-fno-strict-aliasing
build-2.7/temp.linux-x86_64-2.7/build-2.7/src.linux-x86_64-2.7/numpy/core/src/_sortmodule.o
-L/usr/lib64 -Lbuild-2.7/temp.linux-x86_64-2.7 -lnpymath -lm -lpython2.7
-o build-2.7/lib.linux-x86_64-2.7/numpy/core/_sort.so" failed with exit
status 1


# grep numpy /etc/portage/*
#

The shell is bash:

# echo $SHELL
/bin/bash

hmm

Yeah, I know, I could simply get rid of icc again.

But maybe someone in here has an idea why this fails.
Maybe it somehow needs some small fix somewhere.

Stefan

Re: [gentoo-user] using icc with portage

[Tamer Higazi]

ICC would be only interisting if I would have a fallback sollution, that
doesn't exist right now.


For example, if ICC wouldn't compile that it falls back automaticly to gcc.


As said, too sad :(


Tamer

Am 05.09.2011 23:43, schrieb Stefan G. Weichinger:
> 
> Anyone else using Intel's compiler, icc?
> 
> I do for quite a while now.
> Followed http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage
> 
> I still prefer gcc over icc so I use
> 
> http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage#.2Fetc.2Fportage.2Fbashrc
> 
> to only use icc for stuff I list in /etc/portage/package.icc
> 
> -
> 
> Lately I get compilation-errors for packages that aren't listed in that
> file, for example dev-python/numpy-1.6.1
> 
> It fails with log-lines telling me that icc was used to (try to) compile it:
> 
> [..]
> icc: command line warning #10156: ignoring option '-fp'; no argument
> required
> icc: error #10236: File not found:  'l1-cache-line-size=64'
> icc: command line warning #10156: ignoring option '-fp'; no argument
> required
> icc: error #10236: File not found:  'l2-cache-size=4096'
> error: Command "icc -pthread -shared -Wl,-O1 -Wl,--as-needed -shared -O2
> -D_FORTIFY_SOURCE=2 -march=core2 -mcx16 -msahf --param l1-cache-size=32
> --param l1-cache-line-size=64 --param l2-cache-size=4096 -mtune=core2
> -fno-strict-aliasing
> build-2.7/temp.linux-x86_64-2.7/build-2.7/src.linux-x86_64-2.7/numpy/core/src/_sortmodule.o
> -L/usr/lib64 -Lbuild-2.7/temp.linux-x86_64-2.7 -lnpymath -lm -lpython2.7
> -o build-2.7/lib.linux-x86_64-2.7/numpy/core/_sort.so" failed with exit
> status 1
> 
> 
> # grep numpy /etc/portage/*
> #
> 
> The shell is bash:
> 
> # echo $SHELL
> /bin/bash
> 
> hmm
> 
> Yeah, I know, I could simply get rid of icc again.
> 
> But maybe someone in here has an idea why this fails.
> Maybe it somehow needs some small fix somewhere.
> 
> Stefan
> 

Re: [gentoo-user] using icc with portage

[justin]

[-- Attachment #1: Type: text/plain, Size: 2387 bytes --]

On 9/5/11 11:43 PM, Stefan G. Weichinger wrote:
> 
> Anyone else using Intel's compiler, icc?

Hi Stefan,

try to stick to gcc as most pacakges will compile with it.

I personally use icc/ifort for some sience packages and see speedups of
calculation between 2-25x depending on the *FLAGS. But this needs much
optimization of the flags. Nevertheless interesting for performance
critical apps.

Another compiler which was recently released after a long time as closed
source app into the open source world is the ekopath compiler suite
(ekopath(-bin) and path64) which proofed to have the best optimization
of all compilers in benchmarks. But same as icc, it might not work with
some packages.

And never use it with the kernel.

justin


> 
> I do for quite a while now.
> Followed http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage
> 
> I still prefer gcc over icc so I use
> 
> http://www.gentoo-wiki.info/HOWTO_ICC_and_Portage#.2Fetc.2Fportage.2Fbashrc
> 
> to only use icc for stuff I list in /etc/portage/package.icc
> 
> -
> 
> Lately I get compilation-errors for packages that aren't listed in that
> file, for example dev-python/numpy-1.6.1
> 
> It fails with log-lines telling me that icc was used to (try to) compile it:
> 
> [..]
> icc: command line warning #10156: ignoring option '-fp'; no argument
> required
> icc: error #10236: File not found:  'l1-cache-line-size=64'
> icc: command line warning #10156: ignoring option '-fp'; no argument
> required
> icc: error #10236: File not found:  'l2-cache-size=4096'
> error: Command "icc -pthread -shared -Wl,-O1 -Wl,--as-needed -shared -O2
> -D_FORTIFY_SOURCE=2 -march=core2 -mcx16 -msahf --param l1-cache-size=32
> --param l1-cache-line-size=64 --param l2-cache-size=4096 -mtune=core2
> -fno-strict-aliasing
> build-2.7/temp.linux-x86_64-2.7/build-2.7/src.linux-x86_64-2.7/numpy/core/src/_sortmodule.o
> -L/usr/lib64 -Lbuild-2.7/temp.linux-x86_64-2.7 -lnpymath -lm -lpython2.7
> -o build-2.7/lib.linux-x86_64-2.7/numpy/core/_sort.so" failed with exit
> status 1
> 
> 
> # grep numpy /etc/portage/*
> #
> 
> The shell is bash:
> 
> # echo $SHELL
> /bin/bash
> 
> hmm
> 
> Yeah, I know, I could simply get rid of icc again.
> 
> But maybe someone in here has an idea why this fails.
> Maybe it somehow needs some small fix somewhere.
> 
> Stefan
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 267 bytes --]

Re: [gentoo-user] using icc with portage

[Stefan G. Weichinger]

Am 2011-09-07 07:19, schrieb justin:
> On 9/5/11 11:43 PM, Stefan G. Weichinger wrote:
>> 
>> Anyone else using Intel's compiler, icc?
> 
> Hi Stefan,
> 
> try to stick to gcc as most pacakges will compile with it.
> 
> I personally use icc/ifort for some sience packages and see
> speedups of calculation between 2-25x depending on the *FLAGS. But
> this needs much optimization of the flags. Nevertheless interesting
> for performance critical apps.
> 
> Another compiler which was recently released after a long time as 
> closed source app into the open source world is the ekopath
> compiler suite (ekopath(-bin) and path64) which proofed to have the
> best optimization of all compilers in benchmarks. But same as icc,
> it might not work with some packages.
> 
> And never use it with the kernel.

Thanks to both of you (Justin, Tamer).

All this doesn't answer my question, but OK, I will find my way ...

Greets, Stefan

Re: [gentoo-user] using icc with portage

[Tamer Higazi]

As I say, I did once. There is no fallback to gcc if icc wouldn't
compile a package.

I know, that the ICC compiler promise to give more performance....
However, collect your experience and speak with the gentoo maintainer
for the icc compiler packages to have a fallback routine. Would be
really great. I am thinking to get in the next month a core I7. To
compile gentoo on it, would be SUPER!

Am 07.09.2011 10:05, schrieb Stefan G. Weichinger:
> Am 2011-09-07 07:19, schrieb justin:
>> On 9/5/11 11:43 PM, Stefan G. Weichinger wrote:
>>>
>>> Anyone else using Intel's compiler, icc?
>>
>> Hi Stefan,
>>
>> try to stick to gcc as most pacakges will compile with it.
>>
>> I personally use icc/ifort for some sience packages and see
>> speedups of calculation between 2-25x depending on the *FLAGS. But
>> this needs much optimization of the flags. Nevertheless interesting
>> for performance critical apps.
>>
>> Another compiler which was recently released after a long time as 
>> closed source app into the open source world is the ekopath
>> compiler suite (ekopath(-bin) and path64) which proofed to have the
>> best optimization of all compilers in benchmarks. But same as icc,
>> it might not work with some packages.
>>
>> And never use it with the kernel.
> 
> Thanks to both of you (Justin, Tamer).
> 
> All this doesn't answer my question, but OK, I will find my way ...
> 
> Greets, Stefan
> 

Re: [gentoo-user] using icc with portage

[justin]

[-- Attachment #1: Type: text/plain, Size: 1950 bytes --]

On 9/9/11 9:39 PM, Tamer Higazi wrote:
> As I say, I did once. There is no fallback to gcc if icc wouldn't
> compile a package.
> 
> I know, that the ICC compiler promise to give more performance....
> However, collect your experience and speak with the gentoo maintainer
> for the icc compiler packages to have a fallback routine. Would be
> really great. I am thinking to get in the next month a core I7. To
> compile gentoo on it, would be SUPER!

How would one distinguish a failed build because of some library
incompatibility and a real compilation problem with compiler. I dodn't
see a general way to distiguish compiler problems from other problems.

Nevertheless, please try the icc package from sci overlay. Ot works
quite smooth.

If you like speed, try the gold linker for c++ packages like chromium,
qt-libs, libreoffice and similar.

> 
> Am 07.09.2011 10:05, schrieb Stefan G. Weichinger:
>> Am 2011-09-07 07:19, schrieb justin:
>>> On 9/5/11 11:43 PM, Stefan G. Weichinger wrote:
>>>>
>>>> Anyone else using Intel's compiler, icc?
>>>
>>> Hi Stefan,
>>>
>>> try to stick to gcc as most pacakges will compile with it.
>>>
>>> I personally use icc/ifort for some sience packages and see
>>> speedups of calculation between 2-25x depending on the *FLAGS. But
>>> this needs much optimization of the flags. Nevertheless interesting
>>> for performance critical apps.
>>>
>>> Another compiler which was recently released after a long time as 
>>> closed source app into the open source world is the ekopath
>>> compiler suite (ekopath(-bin) and path64) which proofed to have the
>>> best optimization of all compilers in benchmarks. But same as icc,
>>> it might not work with some packages.
>>>
>>> And never use it with the kernel.
>>
>> Thanks to both of you (Justin, Tamer).
>>
>> All this doesn't answer my question, but OK, I will find my way ...
>>
>> Greets, Stefan
>>
> 
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 267 bytes --]