K-ballo 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/
<gnikunj[m]> nice! I'll try to create a calculator using Qi and Karma like in your talk. That should help me understand a bit about them. Then something like scheme.
<gnikunj[m]> wow, thanks!
<hkaiser> gnikunj[m]: sure, that's not the scheme compiler,, though
<gnikunj[m]> yeah
<hkaiser> gnikunj[m]: I found the scheme stuff on an old disk of mine, do you want to have it
<gnikunj[m]> definitely! Can you upload it on drive and share a link?
<gnikunj[m]> or may be upload it on github?
<hkaiser> I sent it, do with it what you like
<gnikunj[m]> thanks!!!
<hkaiser> you'll also need the utree code, I'll send that as well
<gnikunj[m]> btw, could you share the compiler from phylanx?
<gnikunj[m]> the file that takes care of converting code from ast to dataflow
<hkaiser> gnikunj[m]: the parser of the compiler?
<gnikunj[m]> <hkaiser "you'll also need the utree code,"> that will be very helpful, thanks!
<hkaiser> *or
<gnikunj[m]> the compiler, not the parser
<gnikunj[m]> thanks!
bita has quit [Ping timeout: 260 seconds]
nanmiao11 has quit [Remote host closed the connection]
bita has joined #ste||ar
hkaiser has quit [Quit: bye]
bita has quit [Ping timeout: 260 seconds]
shahrzad has quit [Quit: Leaving]
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
parsa has joined #ste||ar
parsa has quit [Client Quit]
parsa has joined #ste||ar
hkaiser has joined #ste||ar
hkaiser has quit [Quit: bye]
K-ballo1 has joined #ste||ar
K-ballo has quit [Ping timeout: 260 seconds]
K-ballo1 is now known as K-ballo
akheir has joined #ste||ar
K-ballo1 has joined #ste||ar
K-ballo has quit [Ping timeout: 265 seconds]
K-ballo1 is now known as K-ballo
hkaiser has joined #ste||ar
bita has joined #ste||ar
powderluv has joined #ste||ar
<powderluv> Hi - are there any examples of using NCCL collectives when using HPX + CUDA ? Thx
<powderluv> I thought I found an implementation of the collectives that allowed any backend library but haven’t been able to located it since
powderluv has quit [Remote host closed the connection]
<K-ballo> building with boost.program_options involves over 42k more includes than using hpx.program_options, how is that even possible?
powderluv has joined #ste||ar
powderluv has quit [Ping timeout: 260 seconds]
<hkaiser> K-ballo: boost program_options includes a lot of boost which is not done by our fork
<K-ballo> I guess if there's 42 TUs that include program options that makes boost only take an extra 1000 files, which wouldn't be as unreasonable
<K-ballo> there's 91, so it's only 500 extra files
<K-ballo> also, there's something severely wrong with core if we need to touch program options in 91 separate TUs
<hkaiser> K-ballo: fair point, are those direct or indirect includes?
<K-ballo> ..only 2 are direct? I must be doing it wrong
<K-ballo> no, it's not wrong, looks like command_line_handling.hpp gets included in the other 89 locations
<K-ballo> some of those should be genuine, but not all 89 of them
<hkaiser> K-ballo: I can't see the document, could you share it wiht my gmail account?
<K-ballo> should be public now
<K-ballo> it's possible other TUs have a direct #include too, but get the header via some indirect include first
<K-ballo> detail/partitioner seems to be leaking it.. why would a detail/partitioner expose command line argument?
<hkaiser> yah, that is somehow not right
<hkaiser> the command line handling should be included by 2 or 3 files, max
<hkaiser> directly, I mean
<K-ballo> there's 81 direct includes
<hkaiser> nod
<K-ballo> out of the 91 inclusions of program options
<K-ballo> no, I'm confusing things
<K-ballo> there's 81 TUs including command line handling
<hkaiser> it's really referred to from about 20 files or so, which makes perfect sense
<hkaiser> most of the includes are in tests
<K-ballo> that means it's just leaked everywhere then
<K-ballo> why is the partitioner the one parsing the command line?
<zao> Ho ho, you found that mess? :D
<zao> I have vague memories of having to deal with it back on BSD OSes, is it because it's so early in startup that it can't get it from anything constructed already?
<K-ballo> ..sounds backwards
<hkaiser> K-ballo: ask ms[m]
bita has quit [Ping timeout: 260 seconds]