View previous topic :: View next topic |
Author |
Message |
mv Watchman
Joined: 20 Apr 2005 Posts: 6749
|
Posted: Fri Apr 30, 2021 6:57 am Post subject: |
|
|
freke wrote: | I don't think graphite should be dangerous - I've used it since ~9.1.0 |
I hardly had any compile problems with it, but a lot of runtime problems: missing or crashing features of some packages which worked without graphite; I forgot the details. OTOH, my attempts with graphite are much older. |
|
Back to top |
|
|
papu l33t
Joined: 25 Jan 2008 Posts: 709 Location: Sota algun pi o alzina...
|
Posted: Fri Apr 30, 2021 3:34 pm Post subject: |
|
|
Hu wrote: | papu wrote: | Does it make any sense to recompile again gcc 11 with gcc 11 ? | No. gcc's bootstrap process should have already done this for you. |
aaah very nice _________________ "~amd64" --cpu 7700 non-x --DDR5 2x16GB 6000MHz --gpu RX 470 |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9691 Location: almost Mile High in the USA
|
Posted: Fri Apr 30, 2021 3:50 pm Post subject: |
|
|
Though this has been something gcc has been doing since forever, sometimes I wish it were possible to not do this automatically to save on build time, and then we could just emerge gcc again at a more opportune time to get the new optimizations on gcc itself.
Use model dependent I guess. Then it also could be said, if a host notices all of the remote distcc helpers already have the same version of gcc that's currently being built, there's technically no need to recompile itself if it solely uses all the remote distcc gcc's? _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
papu l33t
Joined: 25 Jan 2008 Posts: 709 Location: Sota algun pi o alzina...
|
Posted: Fri Apr 30, 2021 4:07 pm Post subject: |
|
|
freke wrote: | papu wrote: | hi all , i just have updated to gcc 11 with lto+gpo , gpo takes double time compilation
are mesurable the benefits of using pgo ?
Does it make any sense to recompile again gcc 11 with gcc 11 ?
i takes 2021-04-28T16:09:10 >>> sys-devel/gcc-11.1.0: 2:45:39
https://i.imgur.com/TD8BuzF.jpg
and also a question about the use
- - graphite : Add support for the framework for loop optimizations based on a polyhedral intermediate representation
What benefit would it have to activate it and if it is a dangerous optimization, which breaks packages ... ?
thanks |
I don't think graphite should be dangerous - I've used it since ~9.1.0 and I don't have any packages which I've turned it off for - lto I had to turn of for a few select packages.
(running console-only servers)
I've been running with these graphite/lto-flags
Code: | CFLAGS="-fgraphite-identity -floop-interchange -ftree-loop-distribution -floop-strip-mine -floop-block"
CFLAGS="${CFLAGS} -flto=4 -fuse-linker-plugin -fno-fat-lto-objects" |
|
i not have idea where i have to put this FLAGS and what flags will be better for my cpu, i have to add globally on my make.conf or just an env for gcc?
I don't know the benefits of graphite , may be less binary sizes?
Code: |
CFLAGS="-march=native -O2 -pipe"
CXXFLAGS="${CFLAGS}"
CHOST="x86_64-pc-linux-gnu"
MAKEOPTS="-j4"
ABI_X86="32 64"
ACCEPT_LICENSE="*"
ACCEPT_KEYWORDS="~amd64"
...
|
_________________ "~amd64" --cpu 7700 non-x --DDR5 2x16GB 6000MHz --gpu RX 470 |
|
Back to top |
|
|
|