logs archiveBotHelp.net / Freenode / #2f30 / 2015 / September / 25 / 2
Evil_Bob
the linux-specific ACPI code is to blame also though
(as far as i understand it)
stateless
yeah acpi is up there with all the other scary sh*t
Evil_Bob
when debugging it the code seemed fragile, it had various windows check in it too
stateless
what u mean windows check?
Evil_Bob
i really should learn about hardware and kernel programming, it seems really interesting to me, but also hard to dive into
FRIGN
Evil_Bob: twss
Evil_Bob
the ACPI code had vendor checks, openbsd "emulates" windows 2003 ACPI iirc
stateless
ah
everyone does that yes, because vendors do not implement the acpi spec but some random subset with incorrect semantics
windows always gets it right because microsoft works with the vendors
so emulating windows is the only way
the openbsd acpi code is also pretty good
but this sh*t has lots of hw specific quirks
easy to break
ok time for reboot
hope X starts :P
Evil_Bob
yeah and when you specify a custom vendor like linux does who knows which parts you might break :P
its like flipping a random switch
heh good luck!
stateless
back
it worked :>
Evil_Bob
btw i also use git now to sync the OpenBSD repo
just git add .; git commit -m "sync" :P
stateless
lol yeah
easier to see changes
http://sta.li/upgrade
hahahaha
love the "done"
Evil_Bob
yeah sh*ts insanely simple
all distros will jizz on our package management tbh
stateless
lol
Evil_Bob
cd /
git pull
JIZZ IN MY PANTS
https://www.youtube.com/watch?v=VLnWf1sQkjY
stateless
Evil_Bob, i want video offloading however
i dont think it works/does it now on openbsd
too much cpu usage for 1080p stuff
and i saw a post somewhere that someone is working on it
Evil_Bob
hmm i think it works, did you try -vo opengl-hq or so?
stateless
i had -vo opengl
that doesn't look accelerated
Evil_Bob
i havent tried watching movies on it yet (small screen) could try later
stateless
i mean it works but it burns a lot of cpu
Evil_Bob
yeah
FRIGN
Evil_Bob: did you find anything in xargs(1)?
Evil_Bob
FRIGN: im le tired
FRIGN
ok
gn8 guys!
atmc
question to the linux sysadmins here
imagine _one_ computing box
N users get a login, but they should be restricted to only use full system capacity (computing) if no other "job" is running currently. how can i realize such a scheduling system?
like if a user executes a command regularly he can just use 10% cpu and 10% ram
but if he "submits" a job he is able to use full capacity (but the job is only started when other jobs finish)
dsockwell
you'd use a resource manager like SLURM or hadoop
atmc
even on just one machine?
dsockwell
are you blind?
atmc
".. of all sizes"
?
dsockwell
"sh*t flows downhill" ?
are we going to start talking in quatrains next?
atmc
i'm talking hindu next
dsockwell
your people killed Jesus
atmc
i think i'm going to use slurm then
never liked apache
reminds me of tomcat
so many hacks
dsockwell
if you use SLURM and then, like, ulimit or nice or something
so only the SLURM user can go full-bore
that'll implement what you want
atmc
do i even need the user accounts with slurm?
probably
dsockwell
i don't remember i haven't used slurm in 4 years
but when i use the cloud
i use the WHOLE CLOUD
atmc
so you work in that area?
dsockwell
no.
atmc
hm
still debating what distribution the workstation will run on
(not the slurm machine)
stateless
atmc, sounds like working around broken schedulers
:P
or like task pinning
"pin it cuz the scheduler sux!"
:P
atmc
:D
stateless
the nice value?
wtf
:P
atmc
over 90000
stateless
:P
atmc
thats gonna be fun when that machine is actually running code
i hope we find a place where i dont get the noise
:D
stateless
http://www.networkworld.com/article/2986176/opensource-subnet/the-world-we-leave-our-children-how-i-became-a-free-software-extremist.html
morning hatred
that guy is fu*king retarded
and so are his children
atmc
lol
« prev 1 2 3 4 5 6 7 next »