Re: [gentoo-user] Specific MAKEOPTS for one package

2018-10-01 Thread Walter Dnes
On Mon, Oct 01, 2018 at 12:34:25PM +0100, Klaus Ethgen wrote

> First, cpufreq-set seems to be not available on gentoo anymore. Doesn't
> matter, just giving it a try.

  I wrote up a bash script (Yes!) to do cpu speed control.  As a script,
it has no external dependancies.  The only requirements are that the CPU
supports frequency scaling, and that the appropriate CPU governors are
compiled into the kernel.

-- 
Walter Dnes 
I don't run "desktop environments"; I run useful applications



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-10-01 Thread Michael Orlitzky
On 10/01/2018 07:34 AM, Klaus Ethgen wrote:
> 
> First, cpufreq-set seems to be not available on gentoo anymore. Doesn't
> matter, just giving it a try.
> 

You want sys-power/cpupower these days.



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-10-01 Thread Klaus Ethgen
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Fr den 28. Sep 2018 um  5:19 schrieb Walter Dnes:
> On Tue, Sep 25, 2018 at 02:25:50PM +0100, Klaus Ethgen wrote
> > I have set MAKEOPTS="-j3" in my make.conf.
>   I assume you have 2 cores on your cpu and you're adding 1 as per the
> ancient rule.

True.

> That may not be the best option anyways.  See
> https://blogs.gentoo.org/ago/2013/01/14/makeopts-jcore-1-is-not-the-best-optimization/
> The author of the blog post ran multiple tests, and concluded that
> setting MAKEOPTS -j to the actual number of cores is slightly faster.
> Years ago, the gcc compiler was less efficient, and adding 1 to -j did
> improve compilation speed.  That is apparently not the case today.

Interesting thoughts.

First, cpufreq-set seems to be not available on gentoo anymore. Doesn't
matter, just giving it a try.

It came out, that all build times have nearly the same sys times but the
user and real times vary. Funny, this is a two CPU machine but the build
with -j1 consumes the least user time. Just the real value gives the
same pattern than on the site.

Regards
   Klaus
- -- 
Klaus Ethgen   http://www.ethgen.ch/
pub  4096R/4E20AF1C 2011-05-16Klaus Ethgen 
Fingerprint: 85D4 CA42 952C 949B 1753  62B3 79D0 B06F 4E20 AF1C
-BEGIN PGP SIGNATURE-
Comment: Charset: ISO-8859-1
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=a+QA
-END PGP SIGNATURE-



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-27 Thread Walter Dnes
On Tue, Sep 25, 2018 at 02:25:50PM +0100, Klaus Ethgen wrote
> Hi Folks,
> 
> I have set MAKEOPTS="-j3" in my make.conf.

  I assume you have 2 cores on your cpu and you're adding 1 as per the
ancient rule.  That may not be the best option anyways.  See
https://blogs.gentoo.org/ago/2013/01/14/makeopts-jcore-1-is-not-the-best-optimization/
The author of the blog post ran multiple tests, and concluded that
setting MAKEOPTS -j to the actual number of cores is slightly faster.
Years ago, the gcc compiler was less efficient, and adding 1 to -j did
improve compilation speed.  That is apparently not the case today.

-- 
Walter Dnes 
I don't run "desktop environments"; I run useful applications



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-26 Thread james
On 9/25/18 10:05 AM, Klaus Ethgen wrote:
> Thanks Dale and Michael, that was exactly I was searching for. (I think.
> ;-)
> 
> Am Di den 25. Sep 2018 um 15:00 schrieb Peter Humphrey:
>> On Tuesday, 25 September 2018 14:25:50 BST Klaus Ethgen wrote:
> 
>>> Is there an option to set -j1 just for firefox package? (and rust, that
>>> has the same problem)?
> 
>> As Dale and Michael said, but you may find -j2 acceptable; It's what I use 
>> for 
>> everything here.
> 
> No. Firefox and Rust is such a beast. It kills my machine even with -j2.
> 
> Regards
>Klaus
> 
> 

Memory may be your issue. A wide variety of places to look at::
ebuild
kernel
userland


I'd suggest you start with reading up on 'jemalloc' as a place to start.
Hopefully other will 'chime in' on debugging/controlling memory
constrained systems and large memory requirements on older/limited
resources.

hth,
James



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-25 Thread Klaus Ethgen
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Thanks Dale and Michael, that was exactly I was searching for. (I think.
;-)

Am Di den 25. Sep 2018 um 15:00 schrieb Peter Humphrey:
> On Tuesday, 25 September 2018 14:25:50 BST Klaus Ethgen wrote:
> 
> > Is there an option to set -j1 just for firefox package? (and rust, that
> > has the same problem)?
> 
> As Dale and Michael said, but you may find -j2 acceptable; It's what I use 
> for 
> everything here.

No. Firefox and Rust is such a beast. It kills my machine even with -j2.

Regards
   Klaus
- -- 
Klaus Ethgen   http://www.ethgen.ch/
pub  4096R/4E20AF1C 2011-05-16Klaus Ethgen 
Fingerprint: 85D4 CA42 952C 949B 1753  62B3 79D0 B06F 4E20 AF1C
-BEGIN PGP SIGNATURE-
Comment: Charset: ISO-8859-1

iQGzBAEBCgAdFiEEMWF28vh4/UMJJLQEpnwKsYAZ9qwFAluqQKkACgkQpnwKsYAZ
9qz/fAwAi+ocD3k/0MlpsXn67vhM6vSYvZM9cqhWcDwhxZhRbjgmjYgfpeOe8JhE
Rey7lrqm/K2HtNZQXSOWiR/pa4Qkgoa8fgV+eWROfYQsQeRM95AzMU+aBMr353BI
mW1BWLzHBDt4N7zk9VIRDDji/2sqohpqyRo9AzzAmJMVtFEccJQjQO/ppcz1eMfg
4DdxBVB7UNoa35jKttCBMl4aEjHtYmw48WRNqjnZA1rnEgozLNn/R74uxfoILQRY
58qBZZjBucO7GSg6X9S8OAOosJmYxpXcih4uejiiWyKiUBN+JZo8XHrhyw2kUHZ0
dy8rSqx6cTE5ga0t5YClwlWxYz/l6iaQPqPG7IRZZUAz1Mfh4Dyc7SIcxA6eu0Z+
lYGxq6ZhXARxds/bjDGYdKK1lXS+AUvQCT3t8V8VwdmIV2yuPmC2Gtt2yiXiU3y8
ns0mel1hm8o16T1eprrQGahhc3ooBJrcE3/hCyIAYyz/RMyHfXzM4M9RVx1k7V7i
JEHMQsoZ
=qjIU
-END PGP SIGNATURE-



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-25 Thread Peter Humphrey
On Tuesday, 25 September 2018 14:25:50 BST Klaus Ethgen wrote:

> Is there an option to set -j1 just for firefox package? (and rust, that
> has the same problem)?

As Dale and Michael said, but you may find -j2 acceptable; It's what I use for 
everything here.

-- 
Regards,
Peter.






Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-25 Thread Dale
Klaus Ethgen wrote:
> Hi Folks,
>
> I have set MAKEOPTS="-j3" in my make.conf. That is great for most
> packages but when I compile firefox that makes my box exploding as it
> consumes simply to much RAM.
>
> Is there an option to set -j1 just for firefox package? (and rust, that
> has the same problem)?
>
> Regards
>    Klaus


I think you are looking for this:

https://wiki.gentoo.org/wiki//etc/portage/package.env

It's been a while since I've had to do this but I'm fairly sure that
will do what you want. 

Hope that helps.

Dale

:-)  :-) 



Re: [gentoo-user] Specific MAKEOPTS for one package

2018-09-25 Thread Michael Orlitzky
On 09/25/2018 09:25 AM, Klaus Ethgen wrote>
> Is there an option to set -j1 just for firefox package? (and rust, that
> has the same problem)?
> 

https://wiki.gentoo.org/wiki//etc/portage/package.env




[gentoo-user] Specific MAKEOPTS for one package

2018-09-25 Thread Klaus Ethgen
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Hi Folks,

I have set MAKEOPTS="-j3" in my make.conf. That is great for most
packages but when I compile firefox that makes my box exploding as it
consumes simply to much RAM.

Is there an option to set -j1 just for firefox package? (and rust, that
has the same problem)?

Regards
   Klaus
- -- 
Klaus Ethgen   http://www.ethgen.ch/
pub  4096R/4E20AF1C 2011-05-16Klaus Ethgen 
Fingerprint: 85D4 CA42 952C 949B 1753  62B3 79D0 B06F 4E20 AF1C
-BEGIN PGP SIGNATURE-
Comment: Charset: ISO-8859-1
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=H7z8
-END PGP SIGNATURE-