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
FunMiles has quit [Quit: Textual IRC Client: www.textualapp.com]
K-ballo has quit [Quit: K-ballo]
hkaiser has quit [Quit: Bye!]
tufei_ has joined #ste||ar
tufei__ has quit [Remote host closed the connection]
Yorlik_ has joined #ste||ar
<ms[m]> dkaratza: thanks! I've just merged the pr 🎉
<ms[m]> and sorry, I had completely forgotten about the rebasing by the end of the meeting...
<ms[m]> did you figure out what you did differently? or was it just a matter of rebasing the same way as the first time?
Yorlik_ has quit [Ping timeout: 252 seconds]
<dkaratza[m]> <ms[m]> "dkaratza: thanks! I've just..." <- great! haha finally
<dkaratza[m]> <ms[m]> "and sorry, I had completely..." <- np, actually auriene had sent me some instructions on how to do it. So, I followed these steps and it worked 😀
Yorlik_ has joined #ste||ar
Yorlik_ is now known as Yorlik
Yorlik has quit [Ping timeout: 265 seconds]
K-ballo has joined #ste||ar
Yorlik has joined #ste||ar
hkaiser has joined #ste||ar
Yorlik has quit [Ping timeout: 252 seconds]
diehlpk_work has joined #ste||ar
tufei_ has quit [Quit: Leaving]
<hkaiser> gonidelis[m]: see pm, pls
Yorlik has joined #ste||ar
Yorlik has quit [Ping timeout: 250 seconds]
tufei has joined #ste||ar
Yorlik has joined #ste||ar
<gnikunj[m]> hkaiser hey!
<hkaiser> gnikunj[m]: here
tufei_ has joined #ste||ar
<gnikunj[m]> hkaiser free for a call sometime today?
<hkaiser> yes, now?
<gnikunj[m]> Plenty of things to catch up on ;)
tufei has quit [Remote host closed the connection]
<gnikunj[m]> My coffee is brewing. In 5min?
<hkaiser> sure, I'm in - ready whenever you are
<diehlpk_work> ms[m], Have you compiled kokkos and hpx-kokkos on Daint with clang?
<diehlpk_work> Since g++ (cray) does not work with nvcc, I switched to cray
<diehlpk_work> clang
<ms[m]> diehlpk_work: not recently, but when I tried it was generally a less painful experience than gcc+nvcc
<ms[m]> I think gdaiss has done that quite recently as well (though not necessarily on daint)
<ms[m]> Octotiger needed some changes for it but otherwise it was fine as far as I understood
Yorlik has quit [Ping timeout: 240 seconds]
diehlpk_work has quit [Remote host closed the connection]
diehlpk_work has joined #ste||ar
<diehlpk_work> boost and clang combined are so painful
<K-ballo> I'd say boost is painful on its own
<diehlpk_work> Yes, and adding the cray compiler makes it even more painful
<diehlpk_work> Yes, boost 1.77 and 1.76 are not compiling with clang 11 at all
<diehlpk_work> Only 1.75 is
<diehlpk_work> ms[m], Octotiger is not the issue on Daint.
<diehlpk_work> Kokkos is killing me
<ms[m]> diehlpk_work: what is the problem exactly?
<ms[m]> Boost and cray clang definitely didn't go well together before 1.77
<ms[m]> Are you compiling all of boost?
<diehlpk_work> For 1.77 it failed to compile b2 with some linker error
<diehlpk_work> Same error for 1.76
<diehlpk_work> 1.75 boost compiled
<diehlpk_work> Here, Kokkos fails with the __float 128 not defined
<diehlpk_work> For 1.77 and 1.76 I got this error
<diehlpk_work> undefined symbol: __cray_sset_detect
<ms[m]> Yeah, that's not a kokkos problem that's a boost problem
<ms[m]> The float128 thing is fixed in 1.77
<diehlpk_work> Yes
<ms[m]> The cray thing I haven't seen
<diehlpk_work> Ok, so I need to get 1.77 to compile
<ms[m]> If you want a workaround you can take the boost config we have in hpx for cray clang
<ms[m]> That disables float 128 with it
<diehlpk_work> Ok, so I could just copy that file to boost include?
<diehlpk_work> I think that would be the easiest way to proceed
<diehlpk_work> Thanks
<ms[m]> The important part is BOOST_USER_CONFIG
<ms[m]> It's up to you where you put the config file as long as it's found
<diehlpk_work> Ok, I will try that later
<diehlpk_work> Moved to Summit and try to fix things there
<hkaiser> fun, I was not aware we do that
<diehlpk_work> hkaiser, I hate the cray compiler wrappers
<diehlpk_work> How can one get gcc to be broken and not conform with the plain gcc
<ms[m]> diehlpk_work: you're not the only one wondering...