aserio changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/
pat[w] has quit [Quit: Leaving]
eschnett has joined #ste||ar
daissgr has quit [Ping timeout: 276 seconds]
Smasher has quit [Remote host closed the connection]
<hkaiser>
bibek: yt?
K-ballo has quit [Quit: K-ballo]
hkaiser has quit [Quit: bye]
mcopik has quit [Ping timeout: 256 seconds]
<Zwei>
Hi again. I'm using ubuntu 17.10, I've installed all the prerequisites except for HDF5. I'm gett this config error: http://codepad.org/ZpfBOY4T
<Zwei>
How do I tell it to use jemalloc?
<Zwei>
note: I also installed tcmalloc via "sudo apt-get install libtcmalloc-minimal4", doesn't seem to help.
<Zwei>
I'm not setting something correctly...
nanashi55 has quit [Ping timeout: 265 seconds]
nanashi55 has joined #ste||ar
parsa has joined #ste||ar
parsa has quit [Quit: Zzzzzzzzzzzz]
<zao>
Zwei: -DHPX_WITH_MALLOC=jemalloc for example.
<zao>
Some of those _ROOT directories should probably point to the directory holding {include,lib}, depends a bit based on what package it's about.
<zao>
If I remember correctly, I think that the Linux branch of the malloc section of our CMakeLists strongly prefers tcmalloc for some reason.
<zao>
In general, all the knobs that HPX exposes to the builder are on the form HPX_WITH_<X>
<zao>
HPX_WITH_CXX14=ON, HPX_WITH_MALLOC=system, etc.
<Zwei>
zao: I see, thanks. I'll give it go now :)
<Zwei>
-- Configuring done
<Zwei>
Yaaaay
<zao>
Woo!
<Zwei>
building now.
<Zwei>
Hopefully soon I can contribute :)
<Zwei>
So nice to find a HPC community using the latest C++ standard
<zao>
A build on my Ryzen takes around 28 minutes including the building of tests.
<zao>
One of the things that's a bit annoying with CircleCI and Appveyor, as those build on way fewer cores.
<zao>
Which is why there's also our buildbot and the recent work on JB's build thingie.
* zao
jiggles off to work
david_pfander has joined #ste||ar
simbergm has quit [Ping timeout: 272 seconds]
simbergm has joined #ste||ar
simbergm has quit [Ping timeout: 248 seconds]
simbergm has joined #ste||ar
K-ballo has joined #ste||ar
parsa has joined #ste||ar
hkaiser has joined #ste||ar
<bibek>
hkaiser: I am here
K-ballo has quit [Ping timeout: 248 seconds]
<hkaiser>
bibek: all should work now, please retry
<bibek>
I just finished compiling phylanx and all tests pass with your latest changes
<bibek>
I will move on to doing performance comparisions vs the old code now
<hkaiser>
bibek: thanks
<hkaiser>
bibek: so you can confirm that the slicing problems are fixed?
<bibek>
hkaiser: yes sir, on the tests, yes. LRA example I will find out in 10 mins.
<hkaiser>
bibek: great
<bibek>
hkaiser: wow, lra csv with 10k rows and columns finished 5 iterations in 1.05 seconds
<hkaiser>
how does this compare to the numpy version?
<bibek>
very similar time
<hkaiser>
perfect
<hkaiser>
bibek: pls comment on the PR, I'll merge it later
<bibek>
will do , thanks
eschnett has quit [Quit: eschnett]
bibek has quit [Quit: Konversation terminated!]
diehlpk has joined #ste||ar
<diehlpk>
jbjnr_, Calling you
RostamLog has joined #ste||ar
rtohid has joined #ste||ar
eschnett has joined #ste||ar
RostamLog_ has joined #ste||ar
twwright_ has joined #ste||ar
parsa has joined #ste||ar
RostamLog has quit [Ping timeout: 265 seconds]
parsa| has quit [Ping timeout: 256 seconds]
twwright has quit [Ping timeout: 264 seconds]
twwright_ is now known as twwright
aserio has joined #ste||ar
rtohid has quit [Ping timeout: 265 seconds]
parsa[w] has quit [Ping timeout: 265 seconds]
zbyerly_ has quit [Ping timeout: 265 seconds]
diehlpk has quit [Ping timeout: 248 seconds]
<heller_>
hkaiser: you should look at the message I wrote a few minutes before yours ;)
parsa has quit [Ping timeout: 252 seconds]
<heller_>
hkaiser: I think it's not too little parallelism, as the number of tasks is quite high. Large idle rates can come from too much parallelism as well
parsa has joined #ste||ar
K-ballo has joined #ste||ar
<aserio>
twwright: Is the CCT site down?
aserio has quit [Quit: aserio]
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
<hkaiser>
heller_: didnt I say that as well?
<heller_>
sure
<heller_>
hkaiser: hmm, i read it as if you'd assume too little parallelism
<hkaiser>
can't tell without seeing the code, it's all just conjecture
aserio has joined #ste||ar
<hkaiser>
, sorry
<hkaiser>
yah, they have too little parallelism
<aserio>
hkaiser: can you access cct.lsu.edu?
<hkaiser>
even if they have alrge amount of tasks those have too many cross dependencies preventing from running in parallel
<hkaiser>
aserio: I cannot
<heller_>
giving the average task duration, i'd go for scheduling overheads though
<hkaiser>
that I said, definitely
<aserio>
thks
<hkaiser>
heller_: but please respond yourself, I might see things the wrong way
<heller_>
I did
<hkaiser>
yah, cct is down :/
<heller_>
"will be used throughout the remainder of this" this is flagged as plagiarism
<hkaiser>
lol
<K-ballo>
hah
<aserio>
hkaiser: Congrats on the sucess of your Phylanx speedups!
<hkaiser>
thanks
bibek has joined #ste||ar
<heller_>
hkaiser: do my messages not come through the list?
<hkaiser>
heller_: cct is down
<aserio>
heller_: yours just posted
<heller_>
ok
<heller_>
sent it 20 minutes ago ;)
<aserio>
I can't log into my own instional resources so there's that
<aserio>
*institutional
<heller_>
poor boy
* heller_
likes grammarly
<heller_>
yeah, it found a plagiarism from one of our papers
<hkaiser>
if it's our paper it's not plagiarism, is it?
<heller_>
not in the german academic system
<hkaiser>
lol
<hkaiser>
you have to copy the whole thing, like all the ministers did
<hkaiser>
;)
<heller_>
:P
<diehlpk_work>
mailinglist seems to work again.
<diehlpk_work>
Got a bunch of notifications for stellar group
<twwright>
aserio, yes, ITS is working on fixing the connectivity issues
<aserio>
twwright: thanks for the update
hkaiser has quit [Quit: bye]
samuel has joined #ste||ar
pree has joined #ste||ar
pree has quit [Read error: Connection reset by peer]
david_pfander has quit [Ping timeout: 246 seconds]
hkaiser has joined #ste||ar
vamatya has joined #ste||ar
vamatya has quit [Client Quit]
aserio has quit [Ping timeout: 265 seconds]
hkaiser has quit [Quit: bye]
vamatya has joined #ste||ar
<K-ballo>
is hello_world_component build by CI ?
aserio has joined #ste||ar
mcopik has joined #ste||ar
Vir has quit [Ping timeout: 265 seconds]
Smasher has joined #ste||ar
rtohid has joined #ste||ar
<jbjnr_>
K-ballo: not unless examples are built, currently I only build tests on daint/etc
<jbjnr_>
not sure about circel-ci etc
<heller_>
circle-ci should build them
<heller_>
jbjnr_: welcome back!
eschnett has quit [Quit: eschnett]
<K-ballo>
circle ci builds examples, but hello_world_component is not part of the cmake build
<heller_>
ok
<heller_>
i think hello_world_component is not built
<heller_>
with the examples
eschnett has joined #ste||ar
jaafar has quit [Ping timeout: 248 seconds]
Vir has joined #ste||ar
mcopik has quit [Ping timeout: 276 seconds]
hkaiser has joined #ste||ar
jaafar has joined #ste||ar
<aserio>
I have a question about clients: can you create an empty client and then attach a server to them later?
<aserio>
hkaiser: ^^
<aserio>
Would this be the operator= overload....
<jbjnr_>
heller_: Hi. Not "welcome back" - I've been here all over the holiday period, but trying to do more work and less chatting is my new year's resolution.
<hkaiser>
aserio: yes, should work
<hkaiser>
jbjnr_: we'll miss your snarky remaks, though
<hkaiser>
remarks, even
jaafar_ has joined #ste||ar
jaafar has quit [Ping timeout: 264 seconds]
<jbjnr_>
hkaiser: my resolution's don't usually last more than a few days, so fear not.
<jbjnr_>
Happy new year btw.
<hkaiser>
ok, what a relieve
<hkaiser>
Happy New Year to you as as well
<aserio>
hkaiser: what do I need to set the client equal to?
<hkaiser>
another client
<aserio>
hkaiser: I am trying to set the client equal to server.get_id()
<aserio>
ah
<aserio>
so I can't take an existing client and assign it to a server...
<hkaiser>
that should work as well
<aserio>
I am getting a has credits assertion
<hkaiser>
where do you get that server instance from?
Smasher has quit [Remote host closed the connection]
<aserio>
I am restoring it
<hkaiser>
normally you don't see the server, just the id
<aserio>
So I was hoping to have the user pass me an empty client...
<aserio>
and I fill it
<hkaiser>
you should restore it using new_<>
<aserio>
ah, and then assign the new client to the old one
<hkaiser>
otherwise it will not be properly registered in agas
<aserio>
that would make sense
<hkaiser>
nod
<hkaiser>
aserio : look at how migration is implemented, this should be similar
<aserio>
nod, I was looking earlier... but for the wrong thing