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/ | GSoC2018: https://wp.me/p4pxJf-k1
hkaiser has quit [Quit: bye]
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Client Quit]
eschnett has joined #ste||ar
stmatengss has joined #ste||ar
diehlpk has quit [Ping timeout: 256 seconds]
K-ballo has quit [Quit: K-ballo]
stmatengss has quit [Quit: Leaving.]
V|r has quit [Ping timeout: 265 seconds]
nikunj97 has joined #ste||ar
Anushi1998 has joined #ste||ar
Anushi1998 has quit [Ping timeout: 276 seconds]
Anushi1998 has joined #ste||ar
nikunj97 has quit [Ping timeout: 260 seconds]
nikunj97 has joined #ste||ar
nikunj1997 has joined #ste||ar
nikunj97 has quit [Ping timeout: 240 seconds]
stmatengss has joined #ste||ar
stmatengss has quit [Ping timeout: 260 seconds]
Anushi1998 has quit [Ping timeout: 268 seconds]
Anushi1998 has joined #ste||ar
stmatengss has joined #ste||ar
anushi has joined #ste||ar
nikunj1997 has quit [Remote host closed the connection]
anushi has quit [Client Quit]
Anushi1998 is now known as anushi
Anushi1998 has joined #ste||ar
jbjnr_ has quit [Read error: Connection reset by peer]
jbjnr has joined #ste||ar
V|r has joined #ste||ar
V|r has quit [Changing host]
V|r has joined #ste||ar
<jbjnr> heller: yt?
<jbjnr> I tried the simple libfabric test you made and on the laptop it runs with two copies, using "bin/libfabric 12345 localhost 12345" from one terminal, and the same from another. Fine. However, when I try "mpiexec -n 2 bin/libfabric 12345 127.0.0.1 12345" then it does nothing and locks up before int main. I'm puzzled. Any idea why it would do that?
<jbjnr> or localhost instead of 127.0.0.1
<heller> I'm surprised that the first even works
<heller> I only tried with two localities having a different port so far
<heller> jbjnr: ^^
<jbjnr> works might be overstating it, a connection is made, then it hangs, but that is at least something
<jbjnr> node A: bin/libfabric 12345 127.0.0.1 54321
<jbjnr> node b: bin/libfabric 54321 127.0.0.1 12345
<jbjnr> better?
V|r has quit [Remote host closed the connection]
<jbjnr> now I at least get a segfault instead of a hang
<heller> ok
<heller> yeah, better
<heller> give me a second please
<heller> the initial connection takes a while
<heller> erm, I might have a pending change
V|r has joined #ste||ar
V|r has quit [Changing host]
V|r has joined #ste||ar
<heller> jbjnr: just pushed
<jbjnr> k
<heller> the kokkos build system is ... interesting...
<heller> also, they won't let me on their slack
<jbjnr> you will need to ask Christian for access, he added cscs.ch to the "allowed" list
<jbjnr> I can ask on slack for you if you want ...
<heller> I can ask him myself, I guess
<jbjnr> ok, new version sends something and then sits there waiting I guess.
<jbjnr> no segfault now
<jbjnr> Is that what we expect?
<heller> yes
<heller> you should see the exchanged endpoints etc
<heller> it's not complete yet or anything...
<jbjnr> yup. got 3
<jbjnr> I understand. I just wanted to see what you'd done and if we can use it
<heller> I think we can use it
<heller> but takes a bit more work
hkaiser has joined #ste||ar
hkaiser_ has joined #ste||ar
hkaiser has quit [Ping timeout: 240 seconds]
hkaiser_ has quit [Client Quit]
hkaiser has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
twwright_ has joined #ste||ar
nikunj97 has joined #ste||ar
anushi has quit [Ping timeout: 240 seconds]
anushi has joined #ste||ar
<M-ms> nikunj97: yt?
<nikunj97> M-ms: yes
<M-ms> did you want to talk about the documentation?
<nikunj97> M-ms: yes I wanted to
<nikunj97> M-ms: How can I contribute to the documentation and porting process?
<M-ms> I think your ideas are really valid for beginners and I'd like to see improvements there
<M-ms> in principle all that information is there, but it may need some (or a lot of) restructuring to make it more logical
<M-ms> basically my hope is that the quick start and building hpx sections should cover your first but if you'd like to reorganize those feel free to do so (might be good to branch off from my branch)
<nikunj97> M-ms: The documentation given is really good to begin with. But for a beginner, if we could add examples then it can become easier for him/her to understand the hpx library in a better manner.
<M-ms> for the conversion itself I added a checklist of things that need to be done
<M-ms> did you see that?
<nikunj97> I for one was looking for example based explanations.
<nikunj97> M-ms: no I did not. I will see today
<M-ms> ok, so you could pick one of those things and try to do it manually or automatically (I'm doing the tables at the moment)
<nikunj97> I'm in the sphinx branch, where have you added the checklist?
<M-ms> and yes, more examples would be good
<M-ms> todo.rst
<M-ms> I would like to see more single node examples, but examples with components and actions are also welcom
<M-ms> e
<nikunj97> M-ms: we have plenty to begin with in the examples directory. If we could document a few of them explaining the how everything works. Then it will be a better idea, from what I think.
<M-ms> agreed
<M-ms> do you know how to build it? I don't rememeber if I put the instructions somewhere...
<nikunj97> M-ms: no I do not. It would make things easier if you could help me with it
<M-ms> just a sec, there are some instructions but I'll add some more information there
<nikunj97> M-ms: thank you
hkaiser has quit [Read error: Connection reset by peer]
<M-ms> nikunj97: ok, pushed the instructions, let me know if something doesn't work
<M-ms> (there are still a lot of warnings)
<nikunj97> M-ms: I've understood the process. Thank you
<nikunj97> M-ms: How would you like me to proceed?
<M-ms> I think best might be to try to convert the markup first, and then think about content
<M-ms> there are the fix code block markup, fix ... etc, you could try to convert those first
<M-ms> or one of those
<M-ms> btw, are you familiar with rst from before?
<nikunj97> Yes that's right. I'll try to convert the markups.
<nikunj97> Yes I'm familiar with rst
<M-ms> good
<M-ms> thanks for helping out with this
<nikunj97> M-ms: I will learn a lot in the process, so I should be the one thanking you for letting me help :)
<M-ms> we need all the help we can get and everyone is welcome to help ?
<jbjnr> M-ms: how is your opencv student doing?
<M-ms> jbjnr: he's doing well
<M-ms> the backend is working but there are still some details to work out regarding that
<M-ms> opencv does its own chunking (sometimes)
<jbjnr> ok
<M-ms> the HPX backend is a lot slower than their pthreads and the TBB backend, but this is hopefully just because of bad chunking
<jbjnr> you mean tht it isn't all nicely located in one place?
<jbjnr> ^messages crossed
<M-ms> he should start with the webcam capture and the qt stuff soonish
<jbjnr> getting the opencv to match tbb is probably #1 important though
<M-ms> no, it's all there, but it wasn't the clearest in terms of when and how chunking is done (depends on backend as well), I think it's pretty clear now though
<M-ms> yep, it should really be able to match it for such a simple thing as a parallel for loop
<jbjnr> if not, we have big problems
<M-ms> yep...
hkaiser has joined #ste||ar
<nikunj97> hkaiser: yt?
<hkaiser> nikunj97: here
<nikunj97> hkaiser: I was able to make things work for static binaries as well
<hkaiser> ohh, nice
<nikunj97> It works perfectly fine for gcc and clang
<nikunj97> and other similar compilers
<hkaiser> very nice
<hkaiser> you will run out of work at this pace ;-)
<nikunj97> The method that I used can be extended to non-static executables as well
<nikunj97> hkaiser: I can always make things more robust ;)
<nikunj97> Besides I'm really interested in the project and the library itself
<hkaiser> glad to hear
<nikunj97> I had one doubt though, regarding the hpx runtime system
<hkaiser> shoot
<nikunj97> You are only allowed to initiate the runtime system once and provide with a function
<hkaiser> you can run it once only, but you don't need to supply a startup function
<nikunj97> Now in my implementation if I were to make that happen, I will have to do everything at __libc_start_main, which is not possible (as everything will begin to initialize itself)
<hkaiser> what do you mean by 'if I were to make that happen'?
<nikunj97> This is because __libc_start_main is the one calling every other functions
<hkaiser> happen - what?
<nikunj97> By that I mean, if I use the current method to initiate the runtime system, then I'll have to implement at __libc_start_main
<hkaiser> ok
<nikunj97> This is especially not possible
<hkaiser> why?
<nikunj97> Because __libc_start_main is responsible for generating the stack, other elf segments, constructor, destructors and only after their initialization it calls main
<nikunj97> None of the std c++ code will work here
<hkaiser> k
<nikunj97> And trying to make things work will most likely result in a seg fault
<hkaiser> what do you suggest
<nikunj97> hkaiser: If there were a way to initiate it twice, or register a function directly with the thread that provides full hpx functionaliy, then I could make things work for global objects as well
<hkaiser> i'm not sure I really understand what youhave in mind
<nikunj97> Basically the current problem is to make hpx functionality available a global object. If I could register a function (responsible to call the global constructors), then every global scoped object will run on hpx threda
<nikunj97> *thread^^
<hkaiser> ok
<nikunj97> hkaiser: Also, I noted one thing. If you initialize the hpx runtime system and then return to the same function (from which you initiated the hpx runtime system). Then that function is still running on the standard thread
<hkaiser> yes
<nikunj97> I see, so it is an expected behavior. I thought of it as a bug, so I thought I'd let you know
<hkaiser> I think that's expected
<nikunj97> hkaiser: you could take a look at this example to understand my implementation for static executable (it is for a std c++ program not hpx), link: https://github.com/NK-Nikunj/GSoC-experimental-codes/tree/master/normal_code/dl_wrap
<hkaiser> will do, thanks
<nikunj97> you only need to run `make` to see the results
<hkaiser> k
<nikunj97> hkaiser: I am trying to find a solution for global scope objects. I will report it to you, once I'm able to make it happen.
<hkaiser> thanks a lot!
<nikunj97> One last thing, did you get in contact with msvc team?
mcopik has joined #ste||ar
K-ballo has joined #ste||ar
stmatengss has quit [Quit: Leaving.]
<heller> nikunj97: does it also work on windows?
<hkaiser> heller: unfortunately not
<hkaiser> gtg
hkaiser has quit [Quit: bye]
<heller> grr, Kokkos ranges aren't proper C++ ranges :/
twwright_ has quit [Quit: twwright_]
<jbjnr> heller: kokkos doesn't handle iterators, only indices. Much is missing from their stuff.
<heller> jbjnr: i guess it is more complicated than initially anticipated to support all of kokkos
<heller> especially the team stuff is confusing to me right now
eschnett has quit [Quit: eschnett]
Anushi1998 has quit [Remote host closed the connection]
ITassin has joined #ste||ar
akheir has joined #ste||ar
<jbjnr> heller: yes. My plan was to do the GPU version of kokkos first - because then we don't have to realy worry about their thread model and teams etc. We just push it through unchanged, it's only when we try to run that on our cpu version that we start getting deep down
mcopik has quit [Ping timeout: 276 seconds]
mcopik has joined #ste||ar
Anushi1998 has joined #ste||ar
eschnett has joined #ste||ar
rtohid has joined #ste||ar
stmatengss has joined #ste||ar
<diehlpk_work> jbjnr, Could you finish the GSoC evaluation for Ahmed
<jbjnr> diehlpk_work: I'm waiting till tomorrow.
<jbjnr> Up until now, he has been working on understanding fflib, libfabric, hpx and not writing code. I've asked for something by tomorrow to help decide
<jbjnr> (NB. tomorrow evening!)
<jbjnr> diehlpk_work: do not worry - I will not miss the review
<diehlpk_work> Sure, I just do not want to lose money :)
<jbjnr> correct
<diehlpk_work> And from previous experiences it is necessary to remind mentors
bibek has joined #ste||ar
<Anushi1998> https://github.com/STEllAR-GROUP/hpx/blob/master/cmake/HPX_UpdateGitDocs.cmake#L42 I think here the path should be instead ${CMAKE_BINARY_DIR}/share/hpx-${HPX_VERSION
<Anushi1998> Because share folder is available in ${CMAKE_BINARY_DIR}
<Anushi1998> I have also tested it in my system and it works fine
diehlpk has joined #ste||ar
<nikunj97> heller: do you mean for msvc?
Anushi1998 has quit [Quit: Bye]
anushi has quit [Read error: Connection reset by peer]
david_pfander1 has joined #ste||ar
diehlpk has quit [Ping timeout: 255 seconds]
Anushi1998 has joined #ste||ar
anushi has joined #ste||ar
Anushi1998 has quit [Read error: Connection reset by peer]
bibek has quit [Quit: Konversation terminated!]
david_pfander1 has quit [Ping timeout: 255 seconds]
bibek has joined #ste||ar
anushi has quit [Ping timeout: 256 seconds]
jakub_golinowski has joined #ste||ar
jakub_golinowski has quit [Client Quit]
twwright_ has joined #ste||ar
Vir has quit [Ping timeout: 265 seconds]
anushi has joined #ste||ar
Anushi1998 has joined #ste||ar
twwright_ has quit [Quit: twwright_]
anushi has quit [Read error: Connection reset by peer]
Anushi1998 has quit [Ping timeout: 260 seconds]
Anushi1998 has joined #ste||ar
bibek has quit [Quit: Konversation terminated!]
<M-ms> Anushi1998: is ${CMAKE_BINARY_DIR} the root build directory for you? I would expect it to be build_dir/bin, but I might be wrong...
<M-ms> also, if you're building the docs keep this in mind: https://github.com/STEllAR-GROUP/hpx/issues/3264#issuecomment-395437189
<K-ballo> ${CMAKE_BINARY_DIR} is indeed the root build directory
<M-ms> ok, then it looks off
galabc has joined #ste||ar
diehlpk has joined #ste||ar
<Anushi1998> I am building in hpx_directory/build.
<Anushi1998> Moreover cmake is writing share folder in build directory directly (in your case I think share must now be located in build_dir/bin) so I still think it to be wrong but I can try by making bin folder as well in build
<Anushi1998> M-ms: ^^
<M-ms> ah, I didn't actually check what my ${CMAKE_BINARY_DIR} is
<M-ms> I would just have expected it to be the bin directory based on the name, and share is for me in the root build dir, and then the line you linked would have been correct
<M-ms> but my guess seems to have been wrong
<M-ms> what's the symptom of that line being wrong?
diehlpk has quit [Ping timeout: 260 seconds]
Anushi1998 has quit [Ping timeout: 240 seconds]
Anushi1998 has joined #ste||ar
<Anushi1998> Sorry I got disconnected. :(
<Anushi1998> M-ms: When I try to add include("${CMAKE_SOURCE_DIR}/cmake/HPX_UpdateGitDocs.cmake") in hpx/docs/CMakeLists.txt it returned an error
<Anushi1998> CMake Error at cmake/HPX_UpdateGitDocs.cmake:47 (file):
<Anushi1998> file COPY cannot find "/home/anushi/HPX/hpx/build/../share/hpx-1.2.0/docs".
<K-ballo> that looks like it should be include(HPX_UpdateGitDocs), not that it would make a difference with that other error though
bibek has joined #ste||ar
<Anushi1998> K-ballo: Okay, I will use that in future :)
<Anushi1998> M-ms, K-Ballo: If the error is confirmed, should I create a PR for this?
* K-ballo has no idea what this is about
galabc has quit [Ping timeout: 248 seconds]
nikunj97 has quit [Ping timeout: 256 seconds]
galabc has joined #ste||ar
<M-ms> tbh I'm not 100% sure either (the reason I'm skeptical is that our documentation builder at least produces some sort of output and I would expect errors if that line was wrong, but it might just lead to broken links or something)
<M-ms> what are you trying to do?
<Anushi1998> I am just trying to solve issue #3217
nikunj has joined #ste||ar
<Anushi1998> I don't think it could be the reason for broken documentation because if it were than there were no updates in gh-pages branch but there are some :/
<Anushi1998> *there should be no updates
nikunj has quit [Ping timeout: 256 seconds]
nikunj has joined #ste||ar
<M-ms> I don't know if this is a good lead but some examples don't seem to be updated at all: https://github.com/STEllAR-GROUP/hpx/tree/gh-pages/docs/html/code/examples/quickstart
<M-ms> maybe the docs git script doesn't know to push changed files, only new files? simplest_hello_world_1 and 2 seem to have been updated
<M-ms> but again, keep in mind that we might change to sphinx and then this would not be needed
<M-ms> so judge yourself how important it is to fix it (it's nonetheless appreciated if you fix it)
stmatengss has quit [Quit: Leaving.]
<M-ms> *most likely change to sphinx
twwright has quit [Quit: twwright]
<Anushi1998> Okay then I think it is not needed :(
<M-ms> Anushi1998: have you worked on the associative containers stuff at all? I know you haven't really received any help with that...
<Anushi1998> I am quite sorry I have worked one month ago on that and I got busy in some stuffs but now that I am settled at new place I was starting all my left-out work back
<Anushi1998> I started with this issue but since it is not needed further I will switch back to containers one :)
<M-ms> sorry... but as you can see it's not really nice to debug, so it would be nice to get rid of it
<M-ms> no worries, I was just wondering
twwright has joined #ste||ar
<M-ms> and happy to hear that you're still interested in working on that
<Anushi1998> Yeah if we are moving to sphinx then surely it is not needed and further it's my bad I can't get that on time :(
<Anushi1998> Yeah sure I will try at best to complete that :)
<M-ms> I can't promise to give you good help but feel free to ask questions (someone will most likely be able to help you), but you'll have to be prepared to do some research yourself :)
<Anushi1998> For sure. I will do my best :)
<Anushi1998> M-ms: Thanks a lot for all that support :)
<M-ms> you're welcome
anushi has joined #ste||ar
Anushi1998 has quit [Remote host closed the connection]
jakub_golinowski has joined #ste||ar
anushi has quit [Read error: Connection reset by peer]
anushi has joined #ste||ar
Anushi1998 has joined #ste||ar
anushi_ has joined #ste||ar
Anushi1998 has quit [Read error: Connection reset by peer]
anushi_ is now known as Anushi1998
anushi has quit [Remote host closed the connection]
jakub_golinowski has quit [Quit: Ex-Chat]
<github> [hpx] hkaiser force-pushed ready_future from 7a82123 to 343eaf1: https://git.io/vhrDS
<github> hpx/ready_future 343eaf1 hkaiser: Adding direct data value to future to avoid allocation for make_ready_future....
K-ballo has quit [Read error: Connection reset by peer]
<K-ballo> I don't like futures stashing values
K-ballo has joined #ste||ar
<jbjnr> K-ballo: we're going to have to have new future types soon...
<K-ballo> for some definition of "soon"
<jbjnr> heller: I added an extra f_enable(endpoint) and now the libfabric PP runs on TCP!
<jbjnr> K-ballo: soon = whenever we need them:)
<jbjnr> cuda_future, remote_future?, every executor can have it's own future type
<K-ballo> that has been "soon" for a couple years now
* K-ballo yawns
<K-ballo> but fair point, I don't like hpx::lcos::future<T> stashing T
<jbjnr> K-ballo: I was not aware of it until now though
<jbjnr> heller: bootup is still done via asio, but the main LF PP runs with TCP (at least my mini test on one node)
ITassin has quit [Quit: Leaving]
anushi has joined #ste||ar
Anushi1998 has quit [Ping timeout: 276 seconds]
Anushi1998 has joined #ste||ar
anushi has quit [Ping timeout: 264 seconds]
<jbjnr> heller: I lied. it doesn't work
<heller> jbjnr: buhuu
<heller> jbjnr: Re kokkos cuda. If I'm not mistaken it should run out of the box
nikunj97 has joined #ste||ar
nikunj has quit [Ping timeout: 245 seconds]
Vir has joined #ste||ar
<jbjnr> I can't get two hello_world's to work on the same node, using the tcp PP alone.
eschnett has quit [Quit: eschnett]
anushi has joined #ste||ar
Anushi1998 has quit [Ping timeout: 264 seconds]
<galabc> Hi I have a question about file modification with ubuntu
<galabc> I want to use bash script to automativally uncomment/comment a cpp file
<galabc> i usually do sed -i 'ns/A/B/' file.txt to replace A with B in the nth line of the file file.txt
<galabc> but it doesnt work when trying to replace // with " " in order to comment uncomment
<galabc> doesnt anyone have any advice?
<galabc> it doesnt work because I cant write sed -i 'ns//// /' file.txt
<galabc> I have also tried sed -i /"//"/" "/' file.txt
<parsa[w]> use | or # instead of / maybe? as in s#A#B
<galabc> thanks is worked :D
<galabc> it **
akheir has quit [Quit: Leaving]
mcopik has quit [Ping timeout: 264 seconds]
galabc has quit [Quit: Leaving]
galabc has joined #ste||ar
galabc has quit [Client Quit]
<K-ballo> is hkaiser traveling?
rtohid has left #ste||ar ["Leaving"]
hkaiser has joined #ste||ar