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
hkaiser has quit [Quit: bye]
K-ballo has quit [Remote host closed the connection]
K-ballo has joined #ste||ar
Hashmi has joined #ste||ar
mdiers_ has joined #ste||ar
mcopik has joined #ste||ar
mcopik has quit [Client Quit]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Ping timeout: 272 seconds]
nikunj has joined #ste||ar
Hashmi has quit [Quit: Connection closed for inactivity]
gonidelis has joined #ste||ar
hkaiser has joined #ste||ar
gonidelis has quit [Remote host closed the connection]
Pranavug has joined #ste||ar
Pranavug has quit [Client Quit]
wate123_Jun has joined #ste||ar
ibalampanis has joined #ste||ar
<ibalampanis> Hello to everyone! I wish to all candidates of GSoC good luck!
<ibalampanis> hkaiser: Do you know what time the results will be announced? Official GSoC website has not the target time.. I'm very anxious about this!
<ibalampanis> Thank you!
<hkaiser> ibalampanis: today 18:00UTC, I believe
<ibalampanis> In 3.5 hours approx!
<ibalampanis> Thank you for the info!
diehlpk_work has joined #ste||ar
ibalampanis has quit [Remote host closed the connection]
rtohid has joined #ste||ar
<diehlpk_work> ms[m], yet?
<diehlpk_work> Can you register here as my backup organizer (https://www.google.com/url?q=https://bit.ly/gsod-altadmin-reg&sa=D&ust=1588607372745000&usg=AFQjCNEnPePimrYQDxJYyewCaQVrsmP5Jg)?
<diehlpk_work> heller1, hkaiser parsa an you register as a mentor for Season of Docs?
<hkaiser> diehlpk_work: what's the project name this year?
gonidelis has joined #ste||ar
<diehlpk_work> hkaiser, The C++ Standard Library for Parallelism and Concurrency (HPX)
gonidelis has quit [Remote host closed the connection]
<ms[m]> diehlpk_work: will do
<ms[m]> thanks for registering us!
<ms[m]> is the application otherwise good to go now?
<ms[m]> I suppose I should register as a mentor as well?
bita_ has joined #ste||ar
<ms[m]> and we might need to update the list of mentors on the project page (maybe even remove them since it's the same mentors on all of them?)
<diehlpk_work> ms[m], Yes, please register as a mentor too
<diehlpk_work> Here is the application, we have so far
<diehlpk_work> I just updated the links to the 2020 stuff and added that we have experience
<hkaiser> diehlpk_work: I have registered as a mentor now
<diehlpk_work> hkaiser, Thanks
<diehlpk_work> parsa as well
<diehlpk_work> ms[m], I am in favor to keep the mentor at all projects. Applications might not see the contact information at the top
<ms[m]> diehlpk_work: sure, fair enough
<diehlpk_work> ms[m], parsa will edit the application and after that I will submit
nan_ has joined #ste||ar
<ms[m]> diehlpk_work, parsa thank you!
<hkaiser> Yorlik: yt?
<Yorlik> Yes
<hkaiser> I think I now understand why th eexception is not handled correctly
<Yorlik> Nice! What do you think is the reason?
<hkaiser> the difference is that you're using that wrapping executor, I think I need to go back and check that it properly propagates exceptions
<Yorlik> That's possible - because here we kinda detour from the standard usage
<hkaiser> right
<Yorlik> Development joys ;)
<hkaiser> I will write a test that uses that executor and make sure everything propagates things correctly
<Yorlik> Awesome !
<Yorlik> We had a long discussion yesterday. TLDR: We'll use vcpg for boost, hwloc and some other things
<hkaiser> Yorlik: is that time critical? are you blocked by this now?
<Yorlik> No
<Yorlik> I can work
<hkaiser> ok, good
<Yorlik> I need to help my codev build a tricky thing and integrate it into the build system
<Yorlik> We had the restrategize the networking we want to use
<Yorlik> And we'll use Valves gamenetsockets as a base layer for a longer while now
<Yorlik> Whiuch requires Protobif, openssl and other stuff.
<Yorlik> So - I'm currently working on that
<hkaiser> Yorlik: both, openssl and protobuf are in vcpkg as well
<Yorlik> I'm using them from there
<hkaiser> nod
<Yorlik> Not OpenSSL though - I'm using abinary install
<diehlpk_work> hkaiser, ms[m] heller1 jbjnr Season of Docs application is submitted
<hkaiser> diehlpk_work: thanks!
karame_ has joined #ste||ar
<ms[m]> diehlpk_work: I'm registered as mentor and admin now as well
Hashmi has joined #ste||ar
<diehlpk_mobile[m> Yeah, Google summer of code is officially announced
<hkaiser> diehlpk_mobile[m: will you set up the gsoc-students mailing list?
<diehlpk_work> hkaiser, Yes, later today or tomorrow morning
Pranavug has joined #ste||ar
<diehlpk_work> hkaiser, I had two deadlines today (GSoC and one report), need to work on the paper revision now
<hkaiser> diehlpk_work: sure, np
ibalampanis has joined #ste||ar
<ibalampanis> Thank you all for this experience! I wish the best for the selected guys and surely for this perfect organization! Hope to find time to contribute to this interesting project
<ibalampanis> Special thanks to bita_ for the contact and communication. All are about experience! Thank you all! Best wishes!
<bita_> ibalampanis, thank _you_
<ibalampanis> bita_: Could you send me a feedback mail to make myself better for the next time? Is that feasible?
<diehlpk_work> ibalampanis, Note that we had much more projects as we had slots from Google.
<ibalampanis> Yeah, I can imagine that!
akheir has joined #ste||ar
Pranavug has quit [Quit: Leaving]
wate123_Jun has quit [Remote host closed the connection]
ibalampanis has quit [Remote host closed the connection]
<Yorlik> Just read the accepted proposals. weilewei, jbjnr, ms[m] - is there any chance we could get a CTrie structure? There is a lack in good C++ implementations (https://en.wikipedia.org/wiki/Ctrie : See references at the bottom).
<Yorlik> Might not any other structures afterwards ;)
<Yorlik> A plus is, it might even be implemented as an efficient immutable structure for better concurrency management.
Hashmi has quit [Quit: Connection closed for inactivity]
Abhishek09 has joined #ste||ar
<Abhishek09> diehlpk_work: not selected , give feedback about my proposal?
<diehlpk_work> Abhishek09, We had so many proposals for the pip package. So many students wanted to do this project
<diehlpk_work> Other proposals were much stronger as yours. So we decided to use a much stronger one.
<diehlpk_work> So not too much feedback I can provide
<Abhishek09> i know @kale is friend of nikunj who have been selected
<diehlpk_work> I am sorry that it was not working out for you this year. GSoC is very competitive
<weilewei> Yorlik looks good, I will look into that later but maybe work on that after GSoC
<nikunj> Abhishek09: I did not consider him as a friend while rating his proposal. Also, every mentor rates the proposal not just me. And none of them knew if I was a friend or not. I'm sorry it didn't work out for you. But please don't defame me.
<Abhishek09> these community has well played , i know before only @kale will be selected , because you all supported him
<diehlpk_work> Abhishek09, Hold you horses~
<Abhishek09> fuckoff
Abhishek09 has quit [Remote host closed the connection]
<Yorlik> weilewei: Cool! I'm aware it might be beyond the current scope you have. But this datastructure is a beast and well worth having available, I believe. You'd have to use Garbage Collection and other techniques - surely quite a sizable project if done right.
<nikunj> Yorlik: are you talking about distributed data structures here?
<Yorlik> Not distributed: Lock Free and potentially immutable.
<nikunj> Distributed lock free is what I meant there
<Yorlik> I believe there is a bunch of potential in that CTrie approach that is not yet harvested
<Yorlik> I don't know enough about distributed lock free structures
<Yorlik> But it should be possible - but that's ust a gut feeling.
<Yorlik> From how I understand it from the paper and some talks I saw it is a very powerful concept.
<Yorlik> Especially if you could combine the concurrent properties with immutability. But I don't know if thatcould be done - this is beyond my paygrade ;)
<Yorlik> I'm pretty sure the C++ specialists here could totally judge that much better than I can.
<Yorlik> What I dare to say is: Have a look - there is something most likely worth looking at.
<weilewei> Yorlik noted, I am working on distributed application, so surely, that will be my interest, just at this point, I will be focusing on my projects
<Yorlik> Sure - I just thought I'd throw it in. :)
<Yorlik> weilewei: Just FYI: This is the reference implementation of that structure in scala from the author: https://github.com/axel22/Ctries
<weilewei> Yorlik oops, last commit 9 years ago..
<Yorlik> Yep - its kinda fizzled away
<Yorlik> Which is a shame, imo
<Yorlik> There is a pretty enthusiastic talk by Phil Nash from CppCon 2017 about the non concurrent version of this: https://www.youtube.com/watch?v=imrSQ82dYns
<Yorlik> I think one would have the check throroughly
<Yorlik> Not saying it's automagically awesome.
<Yorlik> I just try to imagine what this could do if you'd combine lock free access and immutability
<heller1> If it's not being mutated, why do you need a lock in the first place?
<bita_> It might be a stupid question, but I am using hpx:dev image and trying to install blaze and apparently it doesn't find Blas. What do I do wrong?
<bita_> CMake Error at /usr/share/cmake-3.13/Modules/FindPackageHandleStandardArgs.cmake:137 (message):
<bita_> Could NOT find BLAS (missing: BLAS_LIBRARIES)
<bita_> Was BLAS coming with hpx image before? because I am using the same script
<Yorlik> heller1: I think assignment might be an issue where you need locks. And to use such a structure in a meaningful way you need assignments and mutating functions which work on the snapshot you get, while leaving the original unharmed.
<Yorlik> Like c2 = c1.push_back(x)
<Yorlik> c2 and c1 could still share a lot of structure.
<Yorlik> but push back would not return a mutated c1 but a copy / copy_on_write thing
<zao> bita_: How long ago was it you used that image?
<bita_> I think I used it a month ago for the last time
<bita_> zao, what's on your mind?
<zao> Something like OpenBLAS might have been an accidental OS dependency.
<zao> Curious if you've used it before the image was upgraded to Ubuntu 19.10, but that was in February.
weilewei has quit [Remote host closed the connection]
<bita_> actually I have used it before February too
<bita_> thanks for your comment zao
weilewei has joined #ste||ar
<bita_> hkaiser, do you have a minute?
<hkaiser> sure
<bita_> the answer for both localities is list(3,3)
<bita_> from one hand I think it is shape not shape_d and that make sense
<bita_> from the other hand, we are propagating things, shouldn't it calculate the result for each locality?
bita_ has quit [Quit: Leaving]
bita_ has joined #ste||ar
<bita_> hkaiser, was my question unclear? what do you expect from shape having more than one locality?
<hkaiser> sorry
<hkaiser> shape will give you the extends of the local tile only
<hkaiser> bita_: ^^
<hkaiser> shape_d would give you the whole extend
<bita_> well, right now shape gives us the whole extend
<bita_> this, https://gist.github.com/taless474/9f9b5343a566f3ccb62bde16593db5ef, results in list(3,5) on both localities
<bita_> sorry, type list(3,3)
rtohid has left #ste||ar [#ste||ar]
nan_ has quit [Remote host closed the connection]
nan_ has joined #ste||ar
gonidelis has joined #ste||ar
bita_ has quit [Ping timeout: 260 seconds]
gonidelis has quit [Remote host closed the connection]