[gentoo-user] gcc-5.0 ?

[gentoo-user] gcc-5.0 ?

[james]

Hello,

I need access to the latest gcc compiler for some experimental work
with some GPU enabled coding. [1,2] 

So using this latest gcc compiler just for compiling some fancy new cluster
codes using RDMA, is liable (probably ?) to be a wee bit tricky.
Now, assuming I find an overlay somewhere, how do I go about ensuring
it is slotted and only used with specific syntax just for a few codes
and nothing on my general portage tree?

Also, I have not found a gcc-5.0 or gcc-5.1 in an overlay (yet), but
I did find this gcc-9999 by zorry[3]. It does not look to be gcc-5.x but I
am as yet uncertain on that.  

Any discussion or suggestions how to best get access to gcc-5.x
is most appreciated; as is methods to using it (control it) limited
to just a few codes.


James

[1] https://gcc.gnu.org/wiki/OpenACC

[2] https://gcc.gnu.org/gcc-5/changes.html#offload
under: OpenMP 4.0 specification

[3] http://gpo.zugaina.org/sys-devel/gcc

Re: [gentoo-user] gcc-5.0 ?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 335 bytes --]

On Tue, 21 Apr 2015 17:09:37 +0000 (UTC), james wrote:

> Also, I have not found a gcc-5.0 or gcc-5.1 in an overlay (yet),

eix -R -e gcc shows several options, the mpst recent of which appears to
be 5.0.0_alpha20150322 from the toolchain overlay.


-- 
Neil Bothwick

Hard work has a future payoff. Laziness pays off NOW!

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: gcc-5.0 ?

[james]

Neil Bothwick <neil <at> digimed.co.uk> writes:

> 
> On Tue, 21 Apr 2015 17:09:37 +0000 (UTC), james wrote:
> 
> > Also, I have not found a gcc-5.0 or gcc-5.1 in an overlay (yet),
> 
> eix -R -e gcc shows several options, the mpst recent of which appears to
> be 5.0.0_alpha20150322 from the toolchain overlay.


Wow, I never tried the -R option..... Very Very cool!
I was really hoping for 5.1; maybe the name of a dev on the edge?


How do you tell if a ~9999 is actually based on the nightlies,
5.1 or is just old ebuild with the .9999 extension somebody never
got around to renaming or deleting? Look at 

[4] "chromiumos" layman/chromiumos


and tell me (edumacate me?)....

James

[gentoo-user] Re: gcc-5.0 ?

[james]

Neil Bothwick <neil <at> digimed.co.uk> writes:

> eix -R -e gcc shows several options, the mpst recent of which appears to
> be 5.0.0_alpha20150322 from the toolchain overlay.

Interesting situation.  I have about half a dozen overlays set up
via layman. Zugaina does not cleanly sync for me:


<snip>
   Reading category 167|167 (100%) Finished             
[6] 'zugaina' /var/lib/layman/zugaina (cache:
parse|ebuild*#metadata-md5#metadata-assign#assign)
     Reading category  69|167 ( 41%): games-simulation .. * ERROR:
games-simulation/secondlife-1.22.1_rc::zugaina failed (depend phase):
 *   EAPI=0 is not supported
 * 
 * Call stack:
 *                     ebuild.sh, line 584:  Called source
'/var/lib/layman/zugaina/games-simulation/secondlife/secondlife-1.22.1_rc.ebuild'

<end_snip>

1. How do I skip the games portion of zugaina overlays ?

2.' eix -R -e  gcc'  yeilds:

[1] "AstroFloyd" layman/AstroFloyd
[2] "OSSDL" layman/OSSDL
[3] "ROKO__" layman/ROKO__
[4] "chromiumos" layman/chromiumos
[5] "dlang" layman/dlang
[6] "embedded-cross" layman/embedded-cross
[7] "funtoo-overlay" layman/funtoo-overlay
[8] "gentoo-arm" layman/gentoo-arm
[9] "heroxbd" layman/heroxbd
[10] "maggu2810-overlay" layman/maggu2810-overlay
[11] "sabayon-distro" layman/sabayon-distro
[12] "sekh" layman/sekh


Do our overlay lists matchup completely? Does the -R always check
the latest, or is their some updating syntax to ensure the "remotes"
are updated?


James

Re: [gentoo-user] Re: gcc-5.0 ?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 401 bytes --]

On Tue, 21 Apr 2015 19:09:19 +0000 (UTC), james wrote:

> Do our overlay lists matchup completely? Does the -R always check
> the latest, or is their some updating syntax to ensure the "remotes"
> are updated?

eix-remote update


-- 
Neil Bothwick

Physics is like sex: sure, it may give some practical results, but
  that's not why we do it.                            Richard Feynman

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-user] Re: gcc-5.0 ?

[Nikos Chantziaras]

On 21/04/15 21:14, james wrote:
> How do you tell if a ~9999 is actually based on the nightlies,
> 5.1 or is just old ebuild with the .9999 extension somebody never
> got around to renaming or deleting?

9999 are live ebuilds. Not based on any release or nightlies. They 
download the code from a version control repository (Git, Svn, etc.) in 
whatever state it currently is and build from that. The version before 
the 9999 usually specifies the branch. For example, 5.0.9999 would mean 
the latest state of the 5.0 branch (or whatever branch name would apply 
to that version, like "stable".)

[gentoo-user] Re: gcc-5.0 ?

[james]

Nikos Chantziaras <realnc <at> gmail.com> writes:

> 
> On 21/04/15 21:14, james wrote:
> > How do you tell if a ~9999 is actually based on the nightlies,
> > 5.1 or is just old ebuild with the .9999 extension somebody never
> > got around to renaming or deleting?
> 
> 9999 are live ebuilds. Not based on any release or nightlies. They 
> download the code from a version control repository (Git, Svn, etc.) in 
> whatever state it currently is and build from that. The version before 
> the 9999 usually specifies the branch. For example, 5.0.9999 would mean 
> the latest state of the 5.0 branch (or whatever branch name would apply 
> to that version, like "stable".)
> 
> 

Agreeded. Look at this gcc.9999.ebuild and you tell me what version
it is  (Overlay: chromiumos (layman):

http://gpo.zugaina.org/sys-devel/gcc


Sure there is a live ebuild for the latest gcc (5.1.x) ?
I just cannot find it .....


tia,
James

[gentoo-user] Re: gcc-5.0 ?

[Nikos Chantziaras]

On 22/04/15 17:58, james wrote:
> Nikos Chantziaras <realnc <at> gmail.com> writes:
>
>>
>> On 21/04/15 21:14, james wrote:
>>> How do you tell if a ~9999 is actually based on the nightlies,
>>> 5.1 or is just old ebuild with the .9999 extension somebody never
>>> got around to renaming or deleting?
>>
>> 9999 are live ebuilds. Not based on any release or nightlies. They
>> download the code from a version control repository (Git, Svn, etc.) in
>> whatever state it currently is and build from that. The version before
>> the 9999 usually specifies the branch. For example, 5.0.9999 would mean
>> the latest state of the 5.0 branch (or whatever branch name would apply
>> to that version, like "stable".)
>>
>
> Agreeded. Look at this gcc.9999.ebuild and you tell me what version
> it is  (Overlay: chromiumos (layman):
>
> http://gpo.zugaina.org/sys-devel/gcc

Lack of a version number always suggests latest "master" branch.

However, these are Chromium OS overlays. I don't think you're supposed 
to be using them on Gentoo. They're for Chromium OS. For all you know, 
that live ebuild can refer to the master branch of Google's GCC branch, 
and it might not even build or work correctly as a Gentoo compiler.

[gentoo-user] Re: gcc-5.0 ?

[james]

Nikos Chantziaras <realnc <at> gmail.com> writes:


> Lack of a version number always suggests latest "master" branch.

Good to know.

> However, these are Chromium OS overlays. I don't think you're supposed 
> to be using them on Gentoo. They're for Chromium OS. For all you know, 
> that live ebuild can refer to the master branch of Google's GCC branch, 
> and it might not even build or work correctly as a Gentoo compiler.


Bummer. So why does it show up, when I run "eix -R -3 gcc" if
it's not gentoo eligible?

<snip>

 (x86_64-pc-linux-gnu) 4.7.2-r51^ms[3] ~9999^ms[3]
[3] "chromiumos" layman/chromiumos


So is my best hope the "toolchain" repo ?

Do drop me a short message, if there is a  live or 5.1 gcc somewhere.
I'm itching (really bad) to test RDMA on Cephfs with some in-memory
codes, on my gentoo_GPU_linux_cluster_hack.......


thx,
James

Re: [gentoo-user] Re: gcc-5.0 ?

[Dutch Ingraham]

On 04/22/15 12:37, james wrote:
> Nikos Chantziaras <realnc <at> gmail.com> writes:
> 
> 
>> Lack of a version number always suggests latest "master" branch.
> 
> Good to know.
> 
>> However, these are Chromium OS overlays. I don't think you're supposed 
>> to be using them on Gentoo. They're for Chromium OS. For all you know, 
>> that live ebuild can refer to the master branch of Google's GCC branch, 
>> and it might not even build or work correctly as a Gentoo compiler.
> 
> 
> Bummer. So why does it show up, when I run "eix -R -3 gcc" if
> it's not gentoo eligible?
> 
> <snip>
> 
>  (x86_64-pc-linux-gnu) 4.7.2-r51^ms[3] ~9999^ms[3]
> [3] "chromiumos" layman/chromiumos
> 
> 
> So is my best hope the "toolchain" repo ?
> 
> Do drop me a short message, if there is a  live or 5.1 gcc somewhere.
> I'm itching (really bad) to test RDMA on Cephfs with some in-memory
> codes, on my gentoo_GPU_linux_cluster_hack.......
> 
> 
> thx,
> James
> 
> 
> 
> 
5.1 was released today:

https://gcc.gnu.org/gcc-5/

[gentoo-user] Re: gcc-5.0 ?

[Nikos Chantziaras]

On 22/04/15 19:37, james wrote:
> So is my best hope the "toolchain" repo ?
>
> Do drop me a short message, if there is a  live or 5.1 gcc somewhere.
> I'm itching (really bad) to test RDMA on Cephfs with some in-memory
> codes, on my gentoo_GPU_linux_cluster_hack.......

You can always compile and install locally in your $HOME directory. 5.1 
was just released, so you can try that.

[gentoo-user] Re: gcc-5.0 ?

[james]

Nikos Chantziaras <realnc <at> gmail.com> writes:


> > So is my best hope the "toolchain" repo ?

> You can always compile and install locally in your $HOME directory. 5.1 
> was just released, so you can try that.

I hear you. It'd take me a long time to figure out the settings, configs
and such. Beside I know that folks that do this sort of thing put
some time into learning the tricks.....

I'll wait at least until there is an ebuild of some kind. Those folks
(toolchain) that put out the 5.0.x builds should have one for 5.1
soon.......... I do agree with the subliminal suggestion that I should
find those gcc compile and install docs to read about the new options
and feature and where I needed them turned off or on, regardless of
how it is installed on my systems.


thx
James

Re: [gentoo-user] Re: gcc-5.0 ?

[Helmut Jarausch]

[-- Attachment #1: Type: text/plain, Size: 1065 bytes --]

On 04/23/2015 07:51:30 AM, james wrote:
> Nikos Chantziaras <realnc <at> gmail.com> writes:
> 
> 
> > > So is my best hope the "toolchain" repo ?
> 
> > You can always compile and install locally in your $HOME directory.
> 5.1
> > was just released, so you can try that.
> 
> I hear you. It'd take me a long time to figure out the settings,
> configs
> and such. Beside I know that folks that do this sort of thing put
> some time into learning the tricks.....
> 
> I'll wait at least until there is an ebuild of some kind. Those folks
> (toolchain) that put out the 5.0.x builds should have one for 5.1
> soon.......... I do agree with the subliminal suggestion that I should
> find those gcc compile and install docs to read about the new options
> and feature and where I needed them turned off or on, regardless of
> how it is installed on my systems.
> 

I've just renamed the  gcc-6.0.0_alpha20150412.ebuild (the 6 must be typo) from toolchain overlay
to gcc-5.1.0.ebuild which I have attached.
It worked just fine here.

Helmut


[-- Attachment #2: gcc-5.1.0.ebuild --]
[-- Type: text/plain, Size: 472 bytes --]

# Copyright 1999-2012 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: $

EAPI="5"
GCC_FILESDIR="${PORTDIR}/sys-devel/gcc/files"

inherit eutils toolchain

KEYWORDS=""
IUSE="debug"

RDEPEND=""
DEPEND="${RDEPEND}
	>=${CATEGORY}/binutils-2.20"

if [[ ${CATEGORY} != cross-* ]] ; then
	PDEPEND="${PDEPEND} elibc_glibc? ( >=sys-libs/glibc-2.12 )"
fi

src_prepare() {
	toolchain_src_prepare

	use debug && GCC_CHECKS_LIST="yes"
}

Re: [gentoo-user] Re: gcc-5.0 ?

[Stefan G. Weichinger]

On 23.04.2015 10:12, Helmut Jarausch wrote:

> I've just renamed the  gcc-6.0.0_alpha20150412.ebuild (the 6 must be typo) from toolchain overlay
> to gcc-5.1.0.ebuild which I have attached.
> It worked just fine here.

same here, just a bit later ;-)

I now try to build gcc-5.1.0 with itself ... and maybe later I will try
@system in a btrfs-subvolume.

just curious

Re: [gentoo-user] Re: gcc-5.0 ?

[Andreas K. Huettel]

Am Donnerstag, 23. April 2015, 07:51:30 schrieb james:
> I'll wait at least until there is an ebuild of some kind. Those folks
> (toolchain) that put out the 5.0.x builds should have one for 5.1
> soon.......... I do agree with the subliminal suggestion that I should
> find those gcc compile and install docs to read about the new options
> and feature and where I needed them turned off or on, regardless of
> how it is installed on my systems.
> 
> 
> thx
> James

You may want to lurk on bug 547470 :)
https://bugs.gentoo.org/show_bug.cgi?id=547470

-- 
Andreas K. Huettel
Gentoo Linux developer (council, perl, libreoffice)
dilfridge@gentoo.org
http://www.akhuettel.de/

[gentoo-user] Re: gcc-5.0 ?

[james]

Stefan G. Weichinger <lists <at> xunil.at> writes:

> 
> On 23.04.2015 10:12, Helmut Jarausch wrote:
> 
> > I've just renamed the  gcc-6.0.0_alpha20150412.ebuild (the 6 must be
typo) from toolchain overlay
> > to gcc-5.1.0.ebuild which I have attached.
> > It worked just fine here.

very cool! thx.




> I now try to build gcc-5.1.0 with itself ... and maybe later I will try
>  <at> system in a btrfs-subvolume.

Hello Stephan,

Very interesting. You do know that both cephfs-0.94 and gcc-5.1.x
have support for RDMA. It should really speed up some applications,
particularly if you are running Apache:(spark|storm) or other 
"in-memory" codes on top of Apache-mesos (ebuild in BGO).

The recently released (portage)t dev-java/sbt has gotten me much further
along toward a working apache-spark ebuild, also in BGO.

So things are "rocking" for low-latency, HPCC in gentoo. I only regret
that somebody smarter than me is doing all of this..... NONE of the 
old gentoo linux cluster devs are much interested in putting together
a gentoo cluster from 100% sources; and I find that most baffling,
particularly  Donnie Berkholz. Many are using clusters at their work,
based on other distros but little effort is being expended to bring
100% source solutions for clustering to gentoo. 

I do find lots of solutions for containers on remote (vendor) clouds and
binaries for hadoop and such. Nothing so that the rank and file gentoo
communities can build their High Performance Computer Clusters, (HPCC) from
100% sources. Strange, real strange, at least from where I sit....

THANKS for the help.
James




James

Re: [gentoo-user] Re: gcc-5.0 ?

[Stefan G. Weichinger]

On 23.04.2015 19:18, james wrote:
> Stefan G. Weichinger <lists <at> xunil.at> writes:
>> I now try to build gcc-5.1.0 with itself ... and maybe later I will try
>>  <at> system in a btrfs-subvolume.
> 
> Hello Stephan,
> 
> Very interesting. 

emerge -e @system

didn't get very far in my btrfs-subvolume (a snapshot of my current rootfs).

maybe a bit too early :-) (I didn't have the time to look into it closer)

But I run a recent kernel compiled with gcc-5.1.0 now ->

$ cat /proc/version
Linux version 4.0.0-gentoo (root@hiro) (gcc version 5.1.0 (Gentoo 5.1.0)
) #4 SMP Thu Apr 23 21:15:14 CEST 2015

I am sure it will enter portage soon ...

Re: [gentoo-user] Re: gcc-5.0 ?

[Stefan G. Weichinger]

On 23.04.2015 23:06, Stefan G. Weichinger wrote:

> I am sure it will enter portage soon ...

btw ... there you go:

https://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/sys-devel/gcc/gcc-5.1.0.ebuild?view=markup

[gentoo-user] Re: gcc-5.0 ?

[Nikos Chantziaras]

On 21/04/15 20:09, james wrote:
> [installing  gcc 5 system-wide through portage]

Now that 5.1 is in Portage (masked), you should keep in mind that 
emerging it will result in the 5.1 libraries being used, even if you 
keep 4.9 (or 4.8) as the default compiler.

This is not really guaranteed to work well. AFAIK, the only 
fully-supported configuration is having the latest emerged gcc version 
be the default compiler. It might still be best to install 5.1 locally 
outside of portage if you don't intent to make 5.1 the default system 
compiler.

[gentoo-user] Re: gcc-5.0 ?

[james]

Nikos Chantziaras <realnc <at> gmail.com> writes:


> On 21/04/15 20:09, james wrote:
> > [installing  gcc 5 system-wide through portage]

This is not my intention, system wide...

> Now that 5.1 is in Portage (masked), you should keep in mind that 
> emerging it will result in the 5.1 libraries being used, even if you 
> keep 4.9 (or 4.8) as the default compiler.

4.8 for now. I was going to google and noodle around this issue, but
since you brought it up....


> This is not really guaranteed to work well. AFAIK, the only 
> fully-supported configuration is having the latest emerged gcc version 
> be the default compiler. It might still be best to install 5.1 locally 
> outside of portage if you don't intent to make 5.1 the default system 
> compiler.

Naive question: Can't I use /usr/local/portage/..... I guess not,
nor my second 'dumb' idea to set it up as user 'portage'

Ok so what is the overview how to install it (hopefully using
standard gentoo methods) So that I can just use it for amd64 
codes to test on test machines? Also, what about standard libs and
custome libs -[1]?


Later on I want to compile with it on arm64  in a cross compile environment
for the arm64 target (Cortex A53). [2]


James

[1] http://joelinoff.com/blog/?p=1003

[2] http://www.arm.com/products/processors/armv8-architecture.php

[gentoo-user] Re: gcc-5.0 ?

[Martin Vaeth]

Nikos Chantziaras <realnc@gmail.com> wrote:
>
> Now that 5.1 is in Portage (masked), you should keep in mind that
> emerging it will result in the 5.1 libraries being used, even if you
> keep 4.9 (or 4.8) as the default compiler.

If you should really get problems with this, you can manually
remove the corresponding *5.1* line(s) from
/etc/ld.so.conf.d/05gcc-x86_64-pc-linux-gnu.conf
or change the order and then call env-update.
Note that calling gcc-config afterwards would recreate your
original file again, so you do not really have to safe it.

However, so far I had no problems. OTOH, I do not use a heavy
C++ desktop and had already recompiled practically everything with 4.9