hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar-group.org | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | This channel is logged: irclog.cct.lsu.edu
tufei_ has joined #ste||ar
tufei__ has quit [Ping timeout: 276 seconds]
diehlpk_work has joined #ste||ar
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]
K-ballo has quit [Ping timeout: 246 seconds]
hkaiser has quit [Quit: Bye!]
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]
diehlpk_work has quit [Remote host closed the connection]
<ms[m]> dkaratza: sorry for not responding yesterday
<ms[m]> I can't answer your question because I don't understand the problem
<ms[m]> could you describe what you tried that didn't work along with maybe a screenshot of what result you get
<jedi18[m]> The is_sized_sentinel_for check passes for test_iterators wrapped around a bidirectional iterator
<jedi18[m]> How do I make it check for the base bidirectional iterator instead of the test iterator?
<ms[m]> dkaratza: all good with resolving the conflict?
Yorlik has joined #ste||ar
hkaiser has joined #ste||ar
tufei_ has quit [Remote host closed the connection]
tufei_ has joined #ste||ar
K-ballo has joined #ste||ar
Yorlik has quit [Ping timeout: 246 seconds]
hkaiser has quit [Quit: Bye!]
<gonidelis[m]> ms: we tried rebasing to the current master
<gonidelis[m]> with demetra
Yorlik has joined #ste||ar
<gonidelis[m]> and then pushed again
<gonidelis[m]> probably that was the wrong way (kinda my idea)
<ms[m]> gonidelis: tried and failed or tried and succeeded? what else would you have liked to rebase on? ;)
<ms[m]> rebase+push is exactly what I would've done
<gonidelis[m]> a shit
<gonidelis[m]> no it succedeed
<gonidelis[m]> here is the thing: when rebasing, demetra did not got any conflicts
<gonidelis[m]> although she had git fetched the upstream previously
<gonidelis[m]> (i am proxying for her since she is in a class rn)
<ms[m]> sometimes github thinks there are conflicts but git will think the conflicts are resolvable by itself
<gonidelis[m]> hm...
<ms[m]> but there's still a conflict (in case you actually pushed that attempt)
<gonidelis[m]> yup
<gonidelis[m]> we saw that
<ms[m]> if you pushed then you'll want to fetch master again
<gonidelis[m]> we thought we would take your advice then
<gonidelis[m]> ~.^ why?
<ms[m]> because if you had an up-to-date master there wouldn't be a conflict anymore
<ms[m]> is her origin STEllAR-GROUP or her own fork?
<ms[m]> in any case, try again ;)
<gonidelis[m]> origin is stellar
<gonidelis[m]> we rebased on origin
<gonidelis[m]> "demetra" is her own fork
<gonidelis[m]> did not rebase on that
<gonidelis[m]> thanks1
<gonidelis[m]> we 'll try again and report at night ;)
<gonidelis[m]> thanks!*
<ms[m]> gonidelis: her branch is branched from master about three weeks ago
<ms[m]> if I rebase it git definitely tells me there's a conflict
<ms[m]> note that `git fetch origin && git rebase master` is not enough, it has to be `git fetch origin && git rebase origin/master`
<gonidelis[m]> ms: thank you. git rebase origin/master is what she did. anywyas, since we aggree let us try once again
diehlpk_work has joined #ste||ar
<diehlpk_work> ms[m], yt?
<ms[m]> diehlpk_work: here
<diehlpk_work> akheir, Can you please send me the error log on rostam for rocprof?
<diehlpk_work> ms[m], I experience some issues with hpx-kokkos and some g++ unknown flags
<diehlpk_work> nvcc fatal : Unknown option '-soname,libhpx_hydrolib.so'
<ms[m]> diehlpk_work: ok, that looks unrelated to hpx-kokkos, but rather related to kokkos
<ms[m]> and probably nvcc_wrapper
<ms[m]> is this new?
<diehlpk_work> Yes, I could complile octotiger with kokkos on daint before
<diehlpk_work> I do not see the error on Summit
<ms[m]> and where do you see it?
<ms[m]> this is on daint?
<diehlpk_work> Yes
<ms[m]> gcc environment?
<diehlpk_work> I built octotiger with hpx-kokkos few months ago there and it worked
<diehlpk_work> Yes, gcc
<diehlpk_work> 9.0.3
<ms[m]> that looks like a linker option, and should most likely be preceded by `-Wl,`
<diehlpk_work> Yes, I googled that.
<diehlpk_work> However, I am not sure where to add the missing -Wl
<ms[m]> or stripped out by nvcc_wrapper
<ms[m]> do you add that somewhere in octotiger manually or is it added by cmake?
<diehlpk_work> I think it is added by cmake
<ms[m]> I don't know what the problem is, but it seems like nvcc_wrapper is at least aware of -Wl
<ms[m]> can you post the full log?
<diehlpk_work> -shared -Wl,-soname,libhpx_hydrolib.so
<diehlpk_work> At least in the log it looks good
<ms[m]> and what is the kokkos version?
<diehlpk_work> git checkout 3.4.00
<diehlpk_work> Gregor told me to use this version, since he used it for his development
<ms[m]> diehlpk_work: as far as I can tell that should be handled, but it may be buggy
<ms[m]> either find a way to disable setting soname, or open an issue on the kokkos repo
<ms[m]> I think they'll be able to tell you what to do much faster
<diehlpk_work> I think, I will do the second
<ms[m]> it may of course be fixed on later versions of kokkos but at least they'll be able to tell you if that's the case
tufei_ has quit [Remote host closed the connection]
tufei_ has joined #ste||ar
akheir has quit [Quit: Leaving]
diehlpk has joined #ste||ar
diehlpk has quit [Client Quit]
<jedi18[m]> I remember seeing a proposal about splitting hpx into distributed and local-only but never got to know what was decided
<jedi18[m]> So has it been decided to split hpx into two repositories?
<gonidelis[m]> jedi18: yes
<gonidelis[m]> the cscs group focuses on local hpx, the rest of us try to maintain a complete hpx (with the distributed aspects) on our end
<gonidelis[m]> the collaboration though is still on
<gonidelis[m]> i don't know if the fork happened yet though, ms
<jedi18[m]> Ohh ok thanks!
<diehlpk_work> ms[m], Yes, it seems to be some kokkos bug
Yorlik has quit [Ping timeout: 268 seconds]
tufei_ has quit [Quit: Leaving]
Yorlik has joined #ste||ar
tufei has joined #ste||ar
Yorlik has quit [Ping timeout: 264 seconds]
diehlpk has joined #ste||ar
diehlpk has quit [Client Quit]
hkaiser has joined #ste||ar