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/
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
parsa has quit [Quit: Zzzzzzzzzzzz]
diehlpk_mobile has quit [Ping timeout: 256 seconds]
daissgr has quit [Ping timeout: 260 seconds]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Ping timeout: 268 seconds]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile2 has joined #ste||ar
diehlpk_mobile has quit [Ping timeout: 240 seconds]
diehlpk_mobile2 has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
parsa has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile2 has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile2 has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile2 has joined #ste||ar
diehlpk_mobile has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
Anushi1998 has joined #ste||ar
nanashi55 has quit [Ping timeout: 255 seconds]
anushi has quit [Remote host closed the connection]
EverYoung has quit [Read error: Connection reset by peer]
nanashi55 has joined #ste||ar
eschnett has joined #ste||ar
EverYoung has joined #ste||ar
gedaj__ has quit [Read error: Connection reset by peer]
gedaj_ has joined #ste||ar
hkaiser has joined #ste||ar
eschnett has quit [Quit: eschnett]
eschnett has joined #ste||ar
hkaiser has quit [Quit: bye]
K-ballo has quit [Quit: K-ballo]
EverYoung has quit [Remote host closed the connection]
anushi has joined #ste||ar
diehlpk_mobile has joined #ste||ar
diehlpk_mobile2 has quit [Ping timeout: 248 seconds]
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile2 has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile2 has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Read error: Connection reset by peer]
diehlpk_mobile2 has joined #ste||ar
diehlpk_mobile2 has quit [Read error: Connection reset by peer]
diehlpk_mobile has joined #ste||ar
Anushi1998 has quit [Quit: Leaving]
hkaiser has joined #ste||ar
<parsa>
hkaiser: #290 is still broken… removing that if block wasn't sufficient
sharonhsl has joined #ste||ar
gedaj_ has quit [Quit: Leaving]
nanashi55 has quit [Ping timeout: 256 seconds]
nanashi55 has joined #ste||ar
<hkaiser>
parsa: ok
<hkaiser>
what's th eissue?
<hkaiser>
(and good that you added a test ;-)
<parsa>
it says "abandoning not ready shared state: HPX(broken_promise)" on the return statement
<hkaiser>
parsa: ahh, the promise never gets set now
<parsa>
i have no idea what causes that … trying to find out
<hkaiser>
in this case you could simply return make_ready_future(nil{});
<parsa>
why does it happen on return
<parsa>
okay, that's easy
<hkaiser>
because the prose gets destructed without ever set_value being called on it
<hkaiser>
promise*
<hkaiser>
we might teach the compiler about nil, that would enable things like foo(nil, 1, 2, 3)
<parsa>
also can we add BOOST_CONFIG_SUPPRESS_OUTDATED_MESSAGE to our CMake files?
<parsa>
or at least to our AppVeyor config
<hkaiser>
parsa: sure, pls add it to appveyor
<parsa>
will do
<anushi>
hkaiser, Have you reviewed that pull request ? What's that I missed?
Anushi1998 has joined #ste||ar
<hkaiser>
anushi: I have not looked yet, sorry
<hkaiser>
Antrix[m]: is taeguk happy with it now?
<hkaiser>
anushi: ^^
<Anushi1998>
hkaiser: No problem, have a view whenever you are free :)
<hkaiser>
Anushi1998: is taeguk happy now?
<Anushi1998>
I have not asked taeguk, I can't find him on IRC but i will ask him on PR rightaway
<hkaiser>
Anushi1998: he was the one raising the queston related to the seeds
sharonhsl has quit [Quit: sharonhsl]
EverYoung has joined #ste||ar
<hkaiser>
parsa: how many attempts will we need for the circleci stuff?
parsa has quit [Quit: Zzzzzzzzzzzz]
<hkaiser>
parsa[w]: how many attempts will we need for the circleci stuff?
<Anushi1998>
hkaiser: Sorry for that, I have now asked him to review again :)
<hkaiser>
thanks
EverYoung has quit [Ping timeout: 240 seconds]
<hkaiser>
Anushi1998: I think we should stick with random_device{} to generate the seed
<Anushi1998>
hkaiser: Okay,as you say :)
<hkaiser>
Anushi1998: this does not have to be portable
<hkaiser>
if we need a specific seed we can explicitly supply that through the command line
<Anushi1998>
Yes, actually my point was that should we use seeder sequence instead of single number?
<hkaiser>
that makes it more difficult to reproduce results
<hkaiser>
doesn't it?
<hkaiser>
if all relies on a single seed results can be made reproducible by providing a specific seed
<Anushi1998>
Yes it will but it make our random generator better , there has to be a tradeoff :/
<hkaiser>
the main focus for testing should be reproducibility
<Anushi1998>
Okay then I think we are good to go once taeguk is also satisfied :)
<hkaiser>
the quality of the random number sequences is secondary
<hkaiser>
Anushi1998: ok
<Anushi1998>
Also please give me some time for fixing broken documentation because me and K-ballo have figured out that this problem is occurring because our Stellar-bot is not working properly.I will need some time to go through it :(
<hkaiser>
k
<hkaiser>
no worries
<hkaiser>
we're not rushing anywhere
<Anushi1998>
Great, thanks:)
anushi has quit [Remote host closed the connection]
anushi has joined #ste||ar
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
mcopik has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
Anushi1998 has quit [Quit: Leaving]
<jbjnr>
something interesting happened yesterday
simbergm has joined #ste||ar
CaptainRubik has joined #ste||ar
mcopik has joined #ste||ar
<simbergm>
jbjnr: what happened yesterday?
<simbergm>
I will try to fix the pycicle github status, there may be some extra entries on the dashboard but I'll try to keep it to a minimum
<jbjnr>
For the first time, I got results with my cholesky, where the new numa guided executor got better results than the non guided one
<jbjnr>
simbergm: ^
nikunj has joined #ste||ar
<jbjnr>
simbergm: regarding the dashboard - I really don't care if you mess it up from time to time - it's there to help us fix problems and if you have to do 20 build over and over again to find something and fix it, then that's what it's there for
<simbergm>
jbjnr: awesome, well done!
<simbergm>
yeah, sure, I'll try to not do it like last time though :)
<simbergm>
and I realized I can turn off the dashboard submissions so there might be no extra submissions
<simbergm>
jbjnr: so you have it rebased on latest(-ish) master now?
<jbjnr>
yes. I finally got it working
<jbjnr>
took longer than expected and my PR will be <huge>
<jbjnr>
when it happens
<jbjnr>
not really huge, but touches a lot of files
EverYoung has joined #ste||ar
<simbergm>
are you only talking about the added numa hint? or does it contain other changes?
nikunj has quit [Quit: Page closed]
EverYoung has quit [Ping timeout: 245 seconds]
<jbjnr>
the numa hint, the new scheduler and the new executors
<jbjnr>
they form a set, the hints are useless without the executor and scheduler
jakub_golinowski has joined #ste||ar
mcopik has quit [Ping timeout: 240 seconds]
<heller_>
jbjnr: well done!
<heller_>
so we are there?
<jbjnr>
heller_: not yet. The small blocks are still a problem, but in all my previous tests, the new executor actualy made things slower, but this time I see a speed increase with some combinations of settings. I want to run another test in a minute to discover more
<heller_>
great
<heller_>
so profiling results by the end of the day?
mcopik has joined #ste||ar
<jbjnr>
we'll see
<jbjnr>
hope so
<jbjnr>
still want to try your branch though. where is it?
<heller_>
I don't think I have it all together at the moment
<jbjnr>
want me to finish it off?
<heller_>
#3146 was the last batch of changes
<heller_>
after that would have been through, I wanted to remove wait_or_add_new finally
<heller_>
gotta run to entertain students...
<jbjnr>
k
nikunj has joined #ste||ar
Anushi1998 has joined #ste||ar
marco has joined #ste||ar
marco is now known as Guest47721
<nikunj>
Is there any resource related to hpx runtime system where I get to know about how hpx initializes various components and functions, perhaps a video or doc?
Anushi1998 has quit [Remote host closed the connection]
Anushi1998 has joined #ste||ar
<jbjnr>
nikunj: not really. You have to pick the brains of one of us mostly hkaiser and heller have worked on that code.
<nikunj>
@jbjnr: ok, I'll ask hkaiser and heller.
<heller_>
nikunj: as a first entry point, use the debugger and step through the initialization procedure
<heller_>
I'm pretty busy till the end of April and not of much use there right now
<nikunj>
@heller_: oh. In that case I'll try my hand at hpx debugger to better understand the initialization procedure
<heller_>
nikunj: could you maybe squash your commits?
hkaiser has quit [Quit: bye]
<heller_>
nikunj: and you shouldn't remove the copyright
<heller_>
especially since all you did was copy &paste
diehlpk_mobile has quit [Ping timeout: 260 seconds]
anushi_ has quit [Remote host closed the connection]
anushi_ has joined #ste||ar
anushi_ is now known as Anushi1998
thomasd has joined #ste||ar
jakub_golinowski has quit [Ping timeout: 276 seconds]
<thomasd>
Hi, I'm having some issues using channels. channel.get() is blocking at some point in my code, and I'd like to understand why. Can I share a somewhat minimal example (~100 lines)?
<thomasd>
hmm, seems I'm here at the wrong time :)
<simbergm>
thomasd: might be worth posting a link in any case, someone might be able to look a bit later (I would also have a look but I don't know much about HPX's channels)
<thomasd>
ok, here's a link https://codeshare.io/GkbJwA (is there any preferred pasting site? I just googled...)
mcopik has quit [Ping timeout: 256 seconds]
<simbergm>
doesn't really matter as long as it works :) I think people here tend to use github gists
<thomasd>
probably the code doesn't make a lot of sense (and I removed a lot of context), but my problem (at this point) is that the code blocks on the second iteration of "old.recv.get(t)", whereas I thought it would pass through all iterations of my "step()" function, only to block when waiting for the result at the end.
<nikunj>
@heller_: Ohh, I'll replace the copyrights with previous ones. Are you satisfied with everything else?
<simbergm>
thomasd: uhm, where do you set the value for the channel?
<thomasd>
I don't (yet) :)
<thomasd>
but as far as I understand, get() should always immediately return a future, which becomes ready when the corresponding value is set
<simbergm>
yeah, you're right... heller_ tends to know about these things, maybe he can help you better
<heller_>
it should not block
<heller_>
how do you know it blocks?
<thomasd>
I assume it blocks, because the "std::cout" print statements after that don't appear
<thomasd>
it doesn't block for the first iteration of the "for(...j...)" loop, but it does on the second one
<thomasd>
it doesn't (seem to) block if I replace the channel.get() with a make_ready_future(...)
<heller_>
as a first guess, this looks like a problem with dangling reference or something alike
parsa has joined #ste||ar
<heller_>
I have to run now...
parsa has quit [Client Quit]
<thomasd>
heller_: there is a lot of copying and assignment going on with all those boost::multi_array's, I suppose. any tips on how I can find out more?
<K-ballo>
nikunj: you've missattributed copyright, look carefully at the files changed tab
<K-ballo>
unless that was intentional? I don't fully understand why files are moving from tests to examples
<K-ballo>
this has become a much more complex changeset than my ticket initially intended, that was for just one new command in ci build instructions, I'll abstain
wash has joined #ste||ar
<nikunj>
@K-ballo: @heller_ asked me to build it this way instead of building the example it was previously building.
<nikunj>
^^told
<K-ballo>
so I'm reading in the PR discussion, hence while I'll be abstaining
CaptainRubik has quit [Ping timeout: 260 seconds]
viraj has joined #ste||ar
viraj has left #ste||ar [#ste||ar]
aserio has joined #ste||ar
<Zwei>
hey K-ballo, sorry to bother you for advice again. I'm up to chapter 7 of C++ concurrency in action, and it's about lock-free data structures. Just wondering how much is this used in HPX?
<Zwei>
or is it mainly just mutexes and such?
<K-ballo>
we use some lock-free data structures, but we know nothing about them
<K-ballo>
we just use boost.lockfree
<zao>
I'd expect us to have stolen most from Boost.
<Zwei>
Ah, okay :)
<zao>
Which hopefully have done a decent job of it :)
<K-ballo>
we wouldn't dare implementing those ourselves :P
<Zwei>
yeah, it seems complicated
<Zwei>
I'll slowly work through this chapter tho :)
<Zwei>
Well, not "complicated" - but very tricky to get ight
<Zwei>
right*
<K-ballo>
it takes many millennia until an implementation is proven to be right
<Zwei>
yeah ^^
sharonhsl has joined #ste||ar
<Zwei>
K-ballo: when you say "proven" right, do you mean mathematically, using logic, or just loads of testing?
<Zwei>
or both...
<K-ballo>
all of the above, plus use in a production environment, and any other factor you might think of
<Zwei>
heh
<Zwei>
cool :)
sharonhsl has left #ste||ar [#ste||ar]
<simbergm>
jbjnr: pycicle #19, mind having a look? and merging, restarting pycicle if you're happy?
thomasd has quit [Ping timeout: 268 seconds]
parsa[[w]] has joined #ste||ar
parsa[w] has quit [Ping timeout: 252 seconds]
anushi has joined #ste||ar
aserio has quit [Ping timeout: 252 seconds]
anushi is now known as Anushi1998
anushi_ is now known as anushi
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 260 seconds]
aserio has joined #ste||ar
<parsa[[w]]>
error: no member named 'scoped_unlock' in namespace 'hpx::util' :'''(
<StefanLSU>
libc++abi.dylib: terminating with uncaught exception of type std::runtime_error: partitioner::setup_pools: Default pool default has no threads assigned. Please rerun with --hpx:threads=X and check the pool thread assignment
<StefanLSU>
Abort trap: 6
<hkaiser>
yah, somebody showed me the same just yesterday
<hkaiser>
looks like some problem with hwloc on mac - we have not investigated this yet