hkaiser 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/ | GSoD: https://developers.google.com/season-of-docs/
K-ballo has quit [Quit: K-ballo]
weilewei has joined #ste||ar
nikunj has quit [Ping timeout: 265 seconds]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Ping timeout: 276 seconds]
weilewei has quit [Remote host closed the connection]
hkaiser has quit [Ping timeout: 264 seconds]
nikunj has joined #ste||ar
<heller> jbjnr: hey
<zao> Ooh, a force push to the s11n branch. Still broken, tho :)
<zao> Odd, master can't link libhpxd.so
<zao> Does this failure ring any bells? `multiple definition of 'hpx::force_linking()`
<zao> Let's pave the build dir and see.
<zao> Clean build did the trick. Got to love CMake.
<zao> (cleaned dir, that is, just a clean did nada)
<zao> I'm looking briefly at the problem the person mailed hpx-users about with PAPI. If I haven't built "components", it will give the kind of error they showed.
<zao> `runtime_support::load_components: command line processing: unrecognised option '--hpx:papi-event-info=all'`
<zao> If I build "components" first, I get a nice fat segfault out of `bin/hello_world_2 --hpx:papi-event-info=all`
<zao> I'm at work so can't really spend any more time on it, but it seems hecked up :D
<heller> simbergm: jbjnr: ?
<simbergm> heller: here
<heller> simbergm: any updates regarding the workshop?
<simbergm> john says no
<simbergm> heller: ^
<heller> ok
<heller> so we'll assume that it'll happen?
K-ballo has joined #ste||ar
rori has joined #ste||ar
K-ballo has quit [Quit: K-ballo]
K-ballo has joined #ste||ar
hkaiser has joined #ste||ar
<nikunj> hkaiser: yt?
<hkaiser> nikunj: here
<nikunj> hkaiser: I checked the deadlines for IPDPS
<hkaiser> yah, those are over
<nikunj> the abstract deadline is 7th and paper deadline is 14th Oct
<nikunj> are the over really?
<nikunj> i checked here: http://www.ipdps.org
<hkaiser> but the workshop deadlines are still some time out
<hkaiser> this is for the main program, I believe
<nikunj> ohh yeah, the workshop paper deadlines are over
<nikunj> I couldn't find anything for CCGrid
<hkaiser> no
<hkaiser> the call for workshops is over, not the actual workshop deadlines
<hkaiser> those have not even been announced yet
<nikunj> ohh yes, you are right
<nikunj> what would you suggest then?
<hkaiser> ccgrid will come up later
<hkaiser> there is also ISC in germany next summer with many workshops
<nikunj> Let me check ISC
<hkaiser> let me try to find the list of conferences coming up next
aserio has joined #ste||ar
<hkaiser> nikunj: https://www.isc-hpc.com/
<nikunj> thanks!
<nikunj> should you need any help from my side, please let me know
<hkaiser> well, I think Rod will be in contact wrt FLeCSI
<nikunj> yes I'm doing FlecSi work these days, just learnt a thing or two about dockers and now trying building and FlecSi
<nikunj> also, I've booked flight tickets. 24th Nov - 11th Jan
<nikunj> hkaiser: could you send a letter of invitation for the above mentioned dates just like you did for summers? I will need it to apply for a NOC
<hkaiser> nikunj: do you still hav ethat letter we sent last time?
<nikunj> yes I do
<nikunj> should I email that back to you?
<hkaiser> yes, pls
<nikunj> email sent
<hkaiser> nikunj: ok, we'll send another one
weilewei has joined #ste||ar
hkaiser has quit [Ping timeout: 250 seconds]
weilewei has quit [Remote host closed the connection]
<diehlpk_work> nikunj, CCgrid deadline is December 12th
<diehlpk_work> tianyi93, Prepares a paper for this event
akheir has quit [Remote host closed the connection]
rori has quit [Quit: WeeChat 1.9.1]
maxwellr96 has joined #ste||ar
<heller> jbjnr: simbergm: any news?
<simbergm> heller I think we'll know only on Monday...
jaafar has quit [Ping timeout: 240 seconds]
<heller> simbergm: alright
jbjnr_ has joined #ste||ar
<jbjnr_> heller:we will hear a final decision on monday about the course, but today nobody registered after the final reminder about the course, so it's very probable the course is cancelled.
<jbjnr_> ^will be cancelled
<heller> jbjnr_: see pm pleased
<jbjnr_> wrong jbjnr
hkaiser has joined #ste||ar
aserio has quit [Ping timeout: 264 seconds]
<heller> That'd be the idea
<heller> Creating the extra data lazily, adding asserts if they haven't been handled
<heller> That way, you don't pay if you don't have pointers, id_types or futures...
<nikunj> diehlpk_work, ok. Thanks!
aserio has joined #ste||ar
aserio has quit [Ping timeout: 240 seconds]
K-ballo has quit [Quit: K-ballo]
K-ballo has joined #ste||ar
<zao> Found a hint to why I crash trying to do PAPI in hello_world_2, I enter this loop with an empty list of still_unregistered_options: https://github.com/STEllAR-GROUP/hpx/blob/master/src/runtime/components/server/runtime_support_server.cpp#L1088
<zao> It's weird. --hpx:help doesn't list the PAPI options, then when I run in a debugger the options processing runs twice, once earlier and once in a late HPX thread.
<zao> The second time it kind of considers --hpx:papi-event-info an unrecognized option, but also not one.
aserio has joined #ste||ar
<zao> Gah, I have no idea how this is supposed to work and GDB isn't helping much :D
aserio has quit [Ping timeout: 250 seconds]
aserio has joined #ste||ar
<hkaiser> zao: yah, command line processing runs twice
<hkaiser> once right away and once after all plugins were loaded
<zao> Not sure how many times it has run now, but it takes its fair time getting there.
<zao> In any way, at the point it blows up towards the end, there's a whole "unrecognized command line", but none of the options are considered unrecognized.
<hkaiser> zao: ok, I'm planning to have a look
<zao> So for sanity's sake... what do I need to actually use PAPI? Is there some specific build step for a target to use it, or can I use --hpx:papi-* with any target?
<zao> I'm using hello_world_2 for now, as-is.
<zao> Shouldn't blow up regardless, of course, just wondering.
<hkaiser> true
<hkaiser> you should be able to use it with --hpx:print-counter=<some papi counter name>
<zao> Not sure if I've spelled the counter right, of course.
<zao> $ bin/hello_world_2 --hpx:print-counter="/papi{locality#*/total}"
<zao> Typing in something bogus gets me that the only one is called "/papi", which gets me that kind of output again.
<zao> (the exception as linked above)
<hkaiser> that's part of the problem, you should be able t olist the available counters to find out the proper name, which doesn't work ;-)
<hkaiser> I can't remember what the correct naming is, even more as different systems expose different PAPI counters, potentially
<zao> --hpx:list-counters works, --hpx:print-counter=/blargh gets another list of counters.
<hkaiser> zao does list-counters list the PAPI counters as well?
<hkaiser> yah, that's not helpful
<hkaiser> it's supposed to list all available papi counters
<hkaiser> try --hpx:print-counter=/papi/*
<zao> {what}: counter type /papi/* does not match any known type, known counter types:
<hkaiser> grr
<hkaiser> zao: the guy who contributed the papi counters has disappeared long time ago
<hkaiser> we might have to rip them out
<zao> Note that this is running from the build tree, I have yet to install anything.
<zao> Also, PAPI 5.7.0, if the version matters.
K-ballo has quit [Quit: K-ballo]
<hkaiser> so it doesn't even work in the build tree - good to know
K-ballo has joined #ste||ar
nikunj has quit [Remote host closed the connection]
hkaiser has quit [Ping timeout: 245 seconds]
jbjnr_ has quit [Quit: WeeChat 2.5]
jaafar has joined #ste||ar
hkaiser has joined #ste||ar
aserio has quit [Quit: aserio]
nikunj has joined #ste||ar
jbjnr has quit [Read error: Connection reset by peer]
weilewei has joined #ste||ar