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/
EverYoung has quit [Ping timeout: 240 seconds]
bikineev has quit [Remote host closed the connection]
<github> [hpx] hkaiser pushed 1 new commit to master: https://git.io/vQEnp
<github> hpx/master 25d5634 Hartmut Kaiser: Merge pull request #2693 from STEllAR-GROUP/P0443R2...
K-ballo has quit [Quit: K-ballo]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 276 seconds]
hkaiser has quit [Quit: bye]
eschnett has quit [Quit: eschnett]
eschnett has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 276 seconds]
parsa has quit [Quit: Zzzzzzzzzzzz]
Matombo has joined #ste||ar
Matombo has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 246 seconds]
david_pfander has joined #ste||ar
pree has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 276 seconds]
pree has quit [Read error: Connection reset by peer]
mcopik_ has joined #ste||ar
denis_blank has joined #ste||ar
<github> [hpx] StellarBot pushed 1 new commit to gh-pages: https://git.io/vQEQF
<github> hpx/gh-pages af43293 StellarBot: Updating docs
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 246 seconds]
<denis_blank> Is there an existing util function in the codebase like make_range(begin_iter, end_iter) which returns a container like object that returns the given begin and end iter?
mcopik_ has quit [Ping timeout: 255 seconds]
<heller> denis_blank: Look at boost range
bikineev has joined #ste||ar
parsa[[w]] is now known as parsa[w]
hkaiser has joined #ste||ar
Matombo has joined #ste||ar
Matombo has quit [Ping timeout: 240 seconds]
heller has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
heller has joined #ste||ar
K-ballo has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 246 seconds]
denis_blank has quit [Ping timeout: 260 seconds]
denis_blank has joined #ste||ar
bikineev has quit [Ping timeout: 268 seconds]
bikineev has joined #ste||ar
Matombo has joined #ste||ar
bikineev has quit [Ping timeout: 276 seconds]
taeguk has joined #ste||ar
ajaivgeorge has joined #ste||ar
bikineev has joined #ste||ar
Matombo has quit [Ping timeout: 276 seconds]
denis_blank2 has joined #ste||ar
denis_blank has quit [Ping timeout: 248 seconds]
pree has joined #ste||ar
denis_blank2 has quit [Ping timeout: 276 seconds]
pree has quit [Read error: Connection reset by peer]
hkaiser has quit [Quit: bye]
RostamLog has joined #ste||ar
denis_blank2 has joined #ste||ar
denis_blank has quit [Ping timeout: 248 seconds]
hkaiser has joined #ste||ar
<github> [hpx] hkaiser pushed 1 new commit to master: https://git.io/vQuGJ
<github> hpx/master 93f4496 Hartmut Kaiser: Merge pull request #2723 from STEllAR-GROUP/fixing_2722...
<github> [hpx] hkaiser pushed 1 new commit to fixing_2729: https://git.io/vQuZs
<github> hpx/fixing_2729 3953ac6 Hartmut Kaiser: Moving invocation of hook to before objects becomes unmarked
parsa has quit [Quit: Zzzzzzzzzzzz]
parsa has joined #ste||ar
denis_blank has joined #ste||ar
denis_blank2 has quit [Ping timeout: 240 seconds]
denis_blank has quit [Quit: denis_blank]
denis_blank has joined #ste||ar
<github> [hpx] hkaiser pushed 1 new commit to fixing_2699: https://git.io/vQu8M
<github> hpx/fixing_2699 7df83a4 Hartmut Kaiser: Fix option description in cmake
bikineev has quit [Ping timeout: 240 seconds]
aserio has joined #ste||ar
vamatya has joined #ste||ar
Matombo has quit [Remote host closed the connection]
jbjnr_ has joined #ste||ar
<jbjnr_> hkaiser: yt? quick question?
<jbjnr_> (bad train internet)
<hkaiser> jbjnr_: sure
<jbjnr_> when a future becomes ready and a task that depends on it is waiting - does the task get changed to pending/or ready or whaever immediatly - or are they done in batches of N at a time. We are seeing a delay between when a task is theoretically ready and when it actually runs, that could be explained by this
bibek_desktop has quit [Ping timeout: 255 seconds]
<jbjnr_> (lag between when it is ready, and when it goes into a queue and is then run)
<pree> jbjnr_ : may I add my suggestion
<jbjnr_> please do
<pree> I think "the task does not changed to pending/ready". When the task again get's scheduled by the scheduler it executes without blocking for the future (.get())
<pree> I think hpx::dataflow can do what you want exactly ,
<jbjnr_> pree: I'm debugging scheduler code, it is low level and dataflow is not quite applicable to this scenario
<jbjnr_> (dataflow sits on top of the schedulers)
<jbjnr_> (indirectly)
<pree> I don't know exactly , Please wait until expert "HPXers" reply : )
<jbjnr_> hkaiser: ran away :(
<pree> jbjnr_ : Okay .
<pree> jbjnr_ :thanks for making dataflow more clear : )
bibek_desktop has joined #ste||ar
<K-ballo> suddenly the github review UI feels clumsy
<K-ballo> there's not enough context to know what comments are about, and somehow there's no way to see the question in full context?
<K-ballo> (other than just scanning the entire diff, that is, something like a link to comment in place, or ability to load more context before/after)
<K-ballo> I think it used to show about 10 lines of context, instead of just 2, didn't it?
<github> [hpx] atrantan opened pull request #2737: Fixing 2735 (master...fixing_2735) https://git.io/vQugS
mcopik has joined #ste||ar
<hkaiser> jbjnr_: I'm back
zbyerly_ has joined #ste||ar
<hkaiser> tasks become ready immediately
<K-ballo> heller: I'll get to answer your range comments eventually, may take me some time, I don't remember the details anymore so I'll have to refresh them when I have the time
thundergroudon[m has quit [Read error: Connection reset by peer]
taeguk[m] has quit [Read error: Connection reset by peer]
thundergroudon[m has joined #ste||ar
zbyerly_ has quit [Ping timeout: 240 seconds]
jbjnr_ has quit [Ping timeout: 248 seconds]
<hkaiser> K-ballo: does the standard say anything about the memory layout of initializer lists?
taeguk[m] has joined #ste||ar
<K-ballo> maybe, depending which one you mean
<hkaiser> {0, 42, 43]/
<hkaiser> ?
<K-ballo> that could be one of many "initializer lists" still
<K-ballo> std::initializer_list<T> will point to an underlying array of T
<K-ballo> does that answer the question?
<hkaiser> C-array?
<K-ballo> the layout itself of `std::initializer_list` is unspecified, but the underlying array is a plain C array, yes
<hkaiser> ok
<hkaiser> that answers the question, thanks
<K-ballo> as for std::initializer_list, msvc and gcc/clang are not compatible
<K-ballo> one does pair of pointers, the other does pointer+size
<K-ballo> although I believe STL wanted to change the msvc one eventually to match gcc/clang
<hkaiser> K-ballo: I was more interested in the underlying data representation
bikineev has joined #ste||ar
jbjnr_ has joined #ste||ar
ajaivgeorge has quit [Quit: ajaivgeorge]
jbjnr_ has quit [Ping timeout: 258 seconds]
bikineev has quit [Ping timeout: 246 seconds]
bikineev has joined #ste||ar
zbyerly_ has joined #ste||ar
aserio has quit [Ping timeout: 246 seconds]
david_pfander has quit [Ping timeout: 246 seconds]
zbyerly_ has quit [Quit: Leaving]
zbyerly_ has joined #ste||ar
zbyerly_ has left #ste||ar [#ste||ar]
<K-ballo> I just rememberd!
zbyerly_ has joined #ste||ar
mcopik has quit [Remote host closed the connection]
aserio has joined #ste||ar
mcopik has joined #ste||ar
ajaivgeorge has joined #ste||ar
mcopik has quit [Remote host closed the connection]
mcopik has joined #ste||ar
pree has quit [Ping timeout: 255 seconds]
<zbyerly_> K-ballo, thats some serious shade
pree has joined #ste||ar
EverYoung has joined #ste||ar
zbyerly_ has quit [Remote host closed the connection]
zbyerly_ has joined #ste||ar
jbjnr_ has joined #ste||ar
denis_blank2 has joined #ste||ar
denis_blank has quit [Ping timeout: 276 seconds]
mbremer has joined #ste||ar
zbyerly_ has quit [Ping timeout: 255 seconds]
zbyerly_ has joined #ste||ar
<hkaiser> zbyerly_: any success on the ticket (1d_hydro)?
<mbremer> Hi, could someone help me build hpx on Stampede2. I seem to be getting compiler errors involving boost.
<zao> mbremer: I believe that's due to a Boost incompatibility with C++17. You can force HPX to use C++14 with some CMake parameter.
<zao> 2017-06-29.log:[13:41:35] <hkaiser> zao: configure with HPX_WITH_CXX14=On, that works around the issue
<mbremer> Great! That would make a lot of sense. It seems to be build with c++17 right now.
<mbremer> Let me give it a try
zbyerly_ has quit [Ping timeout: 258 seconds]
hkaiser has quit [Quit: bye]
denis_blank2 has quit [Ping timeout: 240 seconds]
bikineev has quit [Read error: Connection reset by peer]
bikineev has joined #ste||ar
pree has quit [Remote host closed the connection]
bikineev has quit [Remote host closed the connection]
<mbremer> zao: yup seems like everything has build correctly now
<mbremer> thanks
<zao> Great!
parsa has quit [Quit: Zzzzzzzzzzzz]
mcopik has quit [Ping timeout: 246 seconds]
ajaivgeorge has quit [Ping timeout: 246 seconds]
ajaivgeorge has joined #ste||ar
mcopik has joined #ste||ar
hkaiser has joined #ste||ar
zbyerly_ has joined #ste||ar
<hkaiser> zbyerly_: yt?
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
<zbyerly_> hkaiser, hey sorry, my IRC client doesn't make any noise for some
<zbyerly_> zbyerly, didn't see your ping. just replied to your email
<hkaiser> zbyerly: thanks
<hkaiser> zbyerly: _: I don't know who this is
<zbyerly_> hkaiser, interesting. I think I wrote that example 1,000 years ago
<hkaiser> zbyerly_: should we remove it?
<zbyerly_> hkaiser, probably, but let me look at it first
<zbyerly_> hkaiser, i think the 1-d stencil does everything it does
<hkaiser> ok, pls prepare a PR if you'd like to remove it
<hkaiser> zbyerly_: ^^
<zbyerly_> hkaiser, okay, there is one element that is shown in that example that isn't in 1D stencil.
<zbyerly_> hkaiser, it's a technique that wash[m] and I wrote a few years ago
<zbyerly_> hkaiser, i'm happy to help with anyone who has problems with it until it becomes too much of a problem, or whatever everyone wants to do
bikineev has joined #ste||ar
<hkaiser> zbyerly_: sure, I'd appreciate that
zbyerly_ has quit [Ping timeout: 240 seconds]
<github> [hpx] hkaiser created appveyor (+1 new commit): https://git.io/vQzJ7
<github> hpx/appveyor abb18fe Hartmut Kaiser: Adding support for AppVeyor CI...
akheir has quit [Remote host closed the connection]
<github> [hpx] hkaiser force-pushed appveyor from abb18fe to efd6a0a: https://git.io/vQzUf
<github> hpx/appveyor efd6a0a Hartmut Kaiser: Adding support for AppVeyor CI...
aserio has quit [Quit: aserio]
<mbremer> I'm still having some trouble getting hpx to run on the KNL nodes. Is there anything specific I need to be doing? I've attached the error. It's a little strange, because the error is for the 1d_stencil_1 example, which doesn't even seem to be using hpx really.
<hkaiser> mbremer: have you used the KNL toolchain file?
<mbremer> I have not
<mbremer> I looked at the one for Cori
<mbremer> but it's not a cray cluster as far as I know
bikineev has quit [Read error: Connection reset by peer]
<hkaiser> this is the important setting: set(HPX_WITH_MAX_CPU_COUNT "512" CACHE STRING "")
bikineev_ has joined #ste||ar
<hkaiser> probably this as well: set(HPX_WITH_RDTSCP ON CACHE BOOL "")
<mbremer> Should I try using it and commenting out the cray-like portitions? (e.g. using CC)
<hkaiser> mbremer: you'd end up with the two setting I listed above ;)
<hkaiser> but pls try it, I have no idea what's needed on stampede
<hkaiser> leave the parcelport stuff out as well, just start with mpi
<mbremer> Kk, will do. I'll let you know how it goes
<mbremer> hkaiser: I was building with jemalloc should I switch to tbbmalloc?
<hkaiser> jemalloc should be fine
<taeguk> I think f3's parameter naming is wrong when using scan_partitioner.
<taeguk> In above codes, 'next' may imply that 'next' is 'curr' of next partition. But, it doesn't.
<github> [hpx] hkaiser force-pushed appveyor from efd6a0a to bef6864: https://git.io/vQzUf
<github> hpx/appveyor bef6864 Hartmut Kaiser: Adding support for AppVeyor CI...
<hkaiser> taeguk: feel free to change the naming if you believe this will improve readability
<taeguk> When the size of workitems is 2, f2 and f3 seems to be called twice. Is it okay? As I think, it doesn't.
<taeguk> 'the size of workitems is 2' means that there is only one partition. So, I think that f2 and f3 is called once.
<hkaiser> taeguk: are you sure?
<taeguk> I'm not certain.
<taeguk> I'm just now analyzing scan_partitioner. And so far as I understand it, that part does not make sense.
<hkaiser> get_bulk_iteration_shape() may create a workitem
<taeguk> But, when parameters_traits::has_variable_chunk_size() is std::true_type, get_bulk_iteration_shape() don't create a workitem.
parsa has joined #ste||ar
<hkaiser> taeguk: right, that's why the check for workitems.size() == 2
<hkaiser> if it's == 2, then get_bulk_iteration_shape() has created one
<taeguk> Ah, I got it all.
<taeguk> Sorry, that's my misunderstanding. There is no problem.
<mbremer> hkaiser: That seems to have done the trick
<mbremer> I was able to run stencil 1,4,8 (8 with 1 and 2 localities on 1 node)
<hkaiser> mbremer: cool
<hkaiser> mbremer: would you mind documenting how to build things on stampede?
mcopik has quit [Ping timeout: 260 seconds]
<mbremer> Should I try building with libfabric first? It seems like they have it on a module.
<github> [hpx] hkaiser force-pushed appveyor from bef6864 to e56431e: https://git.io/vQzUf
<github> hpx/appveyor e56431e Hartmut Kaiser: Adding support for AppVeyor CI...
jbjnr_ has quit [Ping timeout: 255 seconds]
bikineev_ has quit [Remote host closed the connection]
parsa has quit [Quit: Zzzzzzzzzzzz]
<mbremer> hkaiser: Can you see this? https://github.com/zbyerly/dgswem-hpx/wiki/Building-HPX-on-Stampede2 This is how I've been building hpx on stampede2
<hkaiser> thank mbremer
parsa has joined #ste||ar
<mbremer> Also for the libfabric it seems that HPX can't find PMI. I think the network is Intel Omni-Path (OPH) network, so it's probably just a matter of setting the configuration up right. Not sure if there's a point atm
<github> [hpx] hkaiser force-pushed appveyor from e56431e to eb11282: https://git.io/vQzUf
<github> hpx/appveyor eb11282 Hartmut Kaiser: Adding support for AppVeyor CI...
mbremer has left #ste||ar [#ste||ar]