>compile linux & drivers >with -Ofast -march=native

Attached: pepedevil_by_rpglord-dbebq3y.png (400x400, 9K)

Other urls found in this thread:

lemire.me/blog/2018/07/25/it-is-more-complicated-than-i-thought-mtune-march-in-gcc/
twitter.com/SFWRedditVideos

why is that /devillish/ ? I'm pretty sure that's the right way to do it if you don't wish to wait ages

lmao retard
vectorization cannot be used in the kernel

Is this what you guys mean by devs?

Attached: 1519932692304.jpg (400x299, 14K)

>he doesn't know it's already -may

>php artisan config:cache

Watch even senior devs not understanding why the database is STILL pointing to homestead

>-g -O0 -fno-inline

have some patience instead of risking yourself for a massive fuckup

>-ffast-math

Attached: 1554252988071.gif (640x480, 1.26M)

I always hear this but have never had bugs appear. What could possibly go wrong?

anything. it's easily the most important thing you need to compile so you really shouldn't take any risk

>aka -Ointel

gcc -march=k12 -Intoddlers_btfo -Deeznuts

Oh mama

Attached: scrot.png (916x725, 134K)

wait is that true?
Just changing compile flags can fuck up the kernel?

>The absolute state of C & UB

It can't, GCC heuristics have gotten incredible since 3.x and 4.x when everything really would go to shit

Hey it actually works pretty well

-O3 -march=native the world

Sometimes -march=native has worse options than -march=, output gcc's verbose flags and see the diff

native takes only those that are supported by the system

Use -march=native -mtune=native then.

Why don't I feel any cconsequences?

nope, march=native implies mtune=native, you can google it ;-)

Yes, no, and sort of.
lemire.me/blog/2018/07/25/it-is-more-complicated-than-i-thought-mtune-march-in-gcc/

read the comments below the post, it was a bug of an old gcc af version, which got fixed

>still compiling in 2019
lol?