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/ | GSoC: https://github.com/STEllAR-GROUP/hpx/wiki/Google-Summer-of-Code-%28GSoC%29-2020
bita_ has joined #ste||ar
heller1 has quit [Ping timeout: 244 seconds]
neill[m] has quit [Ping timeout: 244 seconds]
ms[m] has quit [Ping timeout: 244 seconds]
diehlpk_mobile[m has quit [Ping timeout: 244 seconds]
alexandros[m] has quit [Ping timeout: 260 seconds]
kimbo[m] has quit [Ping timeout: 244 seconds]
mathegenie[m] has quit [Ping timeout: 244 seconds]
fred[m] has quit [Ping timeout: 244 seconds]
noise[m] has quit [Ping timeout: 244 seconds]
toralf[m] has quit [Ping timeout: 260 seconds]
ms[m] has joined #ste||ar
diehlpk_mobile[m has joined #ste||ar
mathegenie[m] has joined #ste||ar
neill[m] has joined #ste||ar
heller1 has joined #ste||ar
noise[m] has joined #ste||ar
alexandros[m] has joined #ste||ar
fred[m] has joined #ste||ar
kimbo[m] has joined #ste||ar
toralf[m] has joined #ste||ar
nikunj97 has joined #ste||ar
bita_ has quit [Ping timeout: 260 seconds]
hkaiser has quit [Quit: bye]
nikunj97 has quit [Remote host closed the connection]
bita_ has joined #ste||ar
bita_ has quit [Ping timeout: 244 seconds]
Yorlik has joined #ste||ar
weilewei has quit [Ping timeout: 245 seconds]
nanmiao11 has quit [Ping timeout: 245 seconds]
karame_ has quit [Ping timeout: 245 seconds]
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
Yorlik has quit [Read error: Connection reset by peer]
mariella[m] has left #ste||ar ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
joe[m]1 has left #ste||ar ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
oleg[m] has left #ste||ar ["Kicked by @appservice-irc:matrix.org : Idle for 30+ days"]
Yorlik has joined #ste||ar
nikunj97 has joined #ste||ar
nikunj97 has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
nk__ has joined #ste||ar
nk__ has quit [Remote host closed the connection]
<gonidelis[m]> what's the difference between `hpx::util::invoke_r` and `hpx::util::invoke`?
<gonidelis[m]> plus
parsa has joined #ste||ar
<ms[m]> freenode_gonidelis[m]: it's a "trailing return type" see (2) here: https://en.cppreference.com/w/cpp/language/function
<ms[m]> goes together with the auto
<gonidelis[m]> ms[m]: Yeah great! I just recalled. It seems like I have asked about this before... Thank you very much
nikunj97 has quit [Quit: Bye]
parsa[m] has joined #ste||ar
<diehlpk_mobile[m> "ms[m]" (https://matrix.to/#/@freenode_ms[m]:matrix.org) When is the next release planned?
<diehlpk_mobile[m> We run in many issues with 1.4 in Fedora and I have to apply patches for more recent versions of compilers and libraries.
<diehlpk_mobile[m> So if we will have a release candidate soon, I would stop fixing 1.4 and wait for the new release
<ms[m]> freenode_diehlpk[m]: first rc on monday
<ms[m]> hkaiser, jbjnr, rc postponed to Monday due to logistical reasons :p (just in case you wonder where the tag is...)
<K-ballo> gonidelis[m]: with the _r variant you specify the expected return type, and you get a conversion if needed
<K-ballo> it's as if static_cast<T>(invoke(...)), but only for implicit conversions
<gonidelis[m]> hmm... yeah that makes sense!
<gonidelis[m]> K-ballo: Thanks a lot!
hkaiser has joined #ste||ar
<gonidelis[m]> hkaiser: Just posted #4855. As you suggested, I made an early-adaptation by changing the end-begin iterators but I have not utilized CPO's yet. Once this PR is approved I will extend the adaptation in order to create all the `transform` versions dictated by c++20
<gonidelis[m]> K-ballo: You could take a look too, it would help a lot :) (mainly on the `projectin_identity` specialization)
nikunj has quit [Ping timeout: 244 seconds]
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
<rori> <diehlpk[m] ""ms[m]" (https://matrix.to/#/@fr"> the release candidate will probably be on monday
<rori> diehlpk_work: sorry for the delay
<hkaiser> gonidelis[m]: ok, will have a look
<hkaiser> gonidelis[m], rori: do we meet today?
<gonidelis[m]> hkaiser: When I have a `hpx::util::tagged_tuple<tag::in1(FwdIter1), tag::in2(FwdIter2),
<gonidelis[m]> tag::out(FwdIter3)>` do I convert it like `hpx::util::in_out_result(FwdIter1, FwdIter2, FwdIter3)`
<gonidelis[m]> ?
<gonidelis[m]> or like `hpx::util::in_out_result(FwdIter1, hpx::util::in_out_result( FwdIter2, FwdIter3))`
<gonidelis[m]> ?
<hkaiser> I think it should be hpx::util::in_out_result<FwdIter1, FwdIter2, FwdIter3>
<hkaiser> I think it should be hpx::util::in_in_out_result<FwdIter1, FwdIter2, FwdIter3>
<rori> hkaiser: yes
<gonidelis[m]> yeah yeah sorry
<gonidelis[m]> hkaiser: we meet :)
<gonidelis[m]> hkaiser: in_in ? is that right?
akheir has joined #ste||ar
<gonidelis[m]> hkaiser: oh ok! I think that need to be implemented... we don't have it
<hkaiser> gonidelis[m]: yes
<hkaiser> rori: meeting?
<rori> coming
weilewei has joined #ste||ar
<weilewei> distributed G4 stuff finally in DCA master branch now!
karame_ has joined #ste||ar
<hkaiser> weilewei: congrats!
<weilewei> hkaiser thanks!
nanmiao11 has joined #ste||ar
<hkaiser> weilewei: yt?
<weilewei> hkaiser yes
<hkaiser> I'm looking at the HPX/libcds integration
<hkaiser> currently the setuplibcds refers to your repository
<weilewei> yes
<hkaiser> and to a special branch
<hkaiser> we're going to release HPX next week or so, that means that the repository and branch have to be stable for eternity
<hkaiser> could we do two things, please
<weilewei> sure, what do you need from me
<hkaiser> a) transfer the repository to the STEllAR-GROUP organization (I'd give you full access to that repo), and b) create a real bit tag to refer to the proper commit
<hkaiser> I also would like to have some changes for the libcds integration, I'll create a PR for that separately
<hkaiser> I meant: real *git* tag
<weilewei> I see, that's reasonable. I remember transferring git repo to STE||AR needs some special trick, like last time we did with DCA
nikunj97 has joined #ste||ar
<hkaiser> weilewei: you need to transfer it to me (hkaiser) first, I'll move it to the orgs afterwards
<weilewei> let me do it now
<hkaiser> thanks a lot!
<weilewei> hkaiser ok, have you received transfer request?
<hkaiser> yes, thanks
<weilewei> ok, good
<hkaiser> weilewei: pls change your upstream link: git@github.com:STEllAR-GROUP/libcds.git
<hkaiser> you should have full admin access to that repository
<weilewei> hkaiser thank you! I saw it now
<hkaiser> weilewei: should this stay commented out: https://github.com/STEllAR-GROUP/libcds/blob/hpx-thread/src/init.cpp#L40-L43?
<weilewei> Yes, it should be commented out
<hkaiser> so everything on this branch still uses kernel thread local data?
<weilewei> hkaiser it uses hpx thread data, for hpx support, it does not need that two lines
<hkaiser> but further down it still defines the thread_local variables
<weilewei> hkaiser sorry I haven't really cleaned up hpx-thread branch yet
<weilewei> hkaiser where?
nikunj97 has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
nikunj97 has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
nikunj97 has quit [Read error: Connection reset by peer]
<weilewei> Right, the logic wasn't right, but I did detect hpx thread is used, let me see how to fix it
<rori> hkaiser: should I completely remove VC support for this release?
<rori> ah ok actually no I'll just remove the opion
<rori> option*
nikunj97 has joined #ste||ar
weilewei has quit [Remote host closed the connection]
weilewei has joined #ste||ar
<hkaiser> rori: not sure if the OctoTiger people still rely on it
<hkaiser> rori: Patrick is on vacation today, so we can't ask
<hkaiser> not sure if gdaiss[m] is available, though
<rori> ok thanks for the info
<rori> but if so why did we deprecate it in 1.4.0 ^^'
<hkaiser> I anticipate to need it once the standard has settled on datapar and friends
<rori> ah ok!
<hkaiser> can we leave the DATAPAR stuff in (even if disabled)?
<rori> yes
<hkaiser> thanks
<rori> it was what I was thinking
bita_ has joined #ste||ar
nikunj97 has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
<weilewei> hkaiser if you would like to try hpx support based on Alex's branch, you can follow these instructions https://gist.github.com/weilewei/a10e03cd883c3b5ce4d707812a0a91cd
<weilewei> oh, one more thing, apparently, you might want to build hpx with -DHPX_WITH_LIBCDS=ON
sidhu has joined #ste||ar
<weilewei> will come back later after lunch...
<sidhu> Hey everyone!I want to contribute to ste||ar can anyone help me from where I can get started and start contributing by setting up my development environment?
<zao> Hi there!
<sidhu> hi
<zao> There's some documentation on the prerequisites you need to build HPX, if you're interested in working on the codebase.
<zao> If you're on Windows, vcpkg tends to be a good source for the dependencies, if on Linux the distribution packages may be sufficient.
<zao> Clone the repository, configure with CMake and build :)
<sidhu> sure , i will follow that right now
<nikunj97> hkaiser, is the cpos work done?
<nikunj97> I pulled your code. Working on it right now. Just wanted to make sure if I need to pull/rebase before committing my code.
<zao> sidhu: Is there something in particular that draws you to our projects? Any particular project itself or are you considering using HPX or something for an application of yours?
<sidhu> zao The basic idea of need of advance and smoother parallel computing draws me to this organization as more modern machines get , the software would get way advance and complex and would surely require better parallel process to reduce the processing time. x
<hkaiser> nikunj97: for now, yes
<nikunj97> cool!
<sidhu> which extensions shall I download with visual studio?
<hkaiser> sidhu: whichever you'd like to use ;-)
<zao> Regular Visual Studio doesn't tend to need much, maybe clang-format and editorconfig if they're not built-in nowadays.
<zao> If VS Code, C++ dev, maybe some debugger stuff, never really used it for serious work.
<K-ballo> clang-format is builtin these days
diehlpk__ has joined #ste||ar
<sidhu> i had to install clan-format it wasn't inbuilt
<sidhu> clang*
<hkaiser> editorconfig is built in as well
<sidhu> i had to install that too , even in latest build
<K-ballo> are you sure you are on the latest visual studio?
<sidhu> yes , just installed it from official site
<K-ballo> do you have the actual visual studio, or vscode?
<K-ballo> they were not built-in in vscode last I checked, only in visual studio
<sidhu> actual one
<K-ballo> then what made you have to install them?
<sidhu> sorry ,did not get u
<K-ballo> what happen that led you to install them?
<sidhu> zao told me too, as i asked which one would be useful for contributing in this organization
<K-ballo> but why couldn't you simply use the built-in ones?
<sidhu> I think they weren't inbuilt because they were being displayed in to-install extensions
<zao> It's kind of sneaky, as built-in functionality isn't extensions anymore :D
<zao> In any way, the end goal there is to ensure that the code and fixes you prepare to contribute upstream is formatted according to the style, which we enforce with clang-format.
<sidhu> yes
<sidhu> which primary language do we use here?
<zao> Fancy C++, with a bit of Python for build infrastructure and CMake for the build system.
<zao> Some projects are Python-heavy like Phylanx.
<sidhu> My primary language is python so I would love to contribute in Phylanx
diehlpk__ has quit [Ping timeout: 244 seconds]
<weilewei> hkaiser I saw seg_fault with your latest commit in libcds_integration branch
<hkaiser> ok
<hkaiser> weilewei: have backtrcae?
<hkaiser> it worked for me now, but I'm on windows
<hkaiser> there was an assertion before the segfault
<hkaiser> weilewei: are you using the libcds changes I pushed to #3?
nikunj97 has quit [Quit: Bye]
<weilewei> oh, I am not, let me pull it and build it again
<hkaiser> weilewei: that this pointer at the point of failure looks wrong
<hkaiser> but something else went wrong before that
<weilewei> correct... I am trying to understand why
<hkaiser> weilewei: I think this is something for Alex to look at
<hkaiser> I did add more executor tests to the benchmark, i.e. there are 1e6 more attached hpx threads than before
<hkaiser> I think it's a problem in libcds, they have never had so many threads to worry about
<hkaiser> pls feel free to comment out the new executor tests to see if that circumvents the problem
<hkaiser> or simply try creating less futures (--futures=50000)
sidhu has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
<weilewei> tpe_nostack is causing problem, number of futures is not
<hkaiser> weilewei: that makes sense, no wthat you say it
<hkaiser> do you want to remove it or should I?
<weilewei> hkaiser can you comment it for now, pls?
<weilewei> comment it out with // so we can look into it later
<hkaiser> weilewei: I have removed it
<weilewei> ok, thanks
<hkaiser> I know why it breaks
<hkaiser> the nostack executor is equivalent to a normal kernel thread based implementation
<weilewei> oh okay
<hkaiser> that also means that we should throw an exception on the set/get functions you added to the nostack thread data
<hkaiser> instead of silently doing something that is not going to work
nikunj97 has quit [Remote host closed the connection]
karame_ has quit [Remote host closed the connection]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
Yorlik has quit [Ping timeout: 260 seconds]
<gonidelis[m]> sidhu: Hello!!! and welcome to HPX :D
nikunj has quit [Ping timeout: 240 seconds]
nikunj has joined #ste||ar
Yorlik has joined #ste||ar
<weilewei> apex annoted_function problem is solved, it is my problem not syncing code between my local machine and Summit
<zao> :D
<hkaiser> weilewei: good, I was sure we had fixed this before
<weilewei> hkaiser do you mean cmake formatting is off?
<weilewei> I have a question about variadic template function: how can I pass a string to enqueue() so that hpx::util::annotated_function(f, "name_place_holder")
<hkaiser> no, look at the logic
<hkaiser> wrap the function: futures.emplace_back(pool.enqueue(annotated_function(&ThisType::startWalker, "..."), this, i));
<weilewei> hkaiser the LIBCDS_INSIDE_HPX is defined if libcds is built inside hpx
<weilewei> hkaiser I see, let me try the annotated function
akheir has quit [Quit: Leaving]
<hkaiser> sure, there is an if(cond) ... else() if(cond) ... endif() ...endif() which looks wierd
<hkaiser> ahh, I see it now, nvm
<weilewei> I see, might be the blank spaces are not correct
<weilewei> Ok, i apex profile a longer run DCA++ (1 mins): https://imgur.com/a/YvSSfjM
<weilewei> The walker is clearly takes much much longer time than accumulator
bita_ has quit [Ping timeout: 244 seconds]
<hkaiser> well, the run_helper annotation is clearly wrong ;-)