K-ballo 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/
K-ballo has quit [Quit: K-ballo]
hkaiser has quit [Quit: bye]
<gonidelis[m]> ms[m]: ahh do we want these off formatted on purpose?
<gonidelis[m]> 2187136d6278dd708a149bbafdf26ba0cb9e25ac
jbjnr has joined #ste||ar
<jbjnr> Just a test message to see if Matrix/Element/Riot or whatever it is now is working
<gonidelis[m]> freenode_biddisco[m]: ack
<ms[m]1> gonidelis[m]: could you give a link to the source? that link is only for you
<ms[m]1> jbjnr: works
<gonidelis[m]> ms[m]: sorry
<jbjnr> thanks. And I got a small notification popup when you replied. very good.
<jbjnr> is the web interface good enough, or should I download the app thingy?
<gonidelis[m]> freenode_biddisco[m]: works for me (yes, even on ubuntu). notifications, enxcryption, do not disturb, anything
<ms[m]1> jbjnr: it's essentially the same app (if we're talking about element web vs the desktop app), so that's mostly a matter of preference
<gonidelis[m]> you might have to do some tweaking when you want to pm someone for the first time though
<ms[m]1> but there are some other clients as well which you may or may not prefer
<jbjnr> ok. I'll use the broswer for now
<jbjnr> I have a problem because my RMA memory stuff forces dependencies on serialization to the parcelport, so I think I need to move rma stufff to libs/memory
<jbjnr> but maybe it deserves its own lib
<ms[m]1> jbjnr: parcelport depends on serialization? that should be fine
<ms[m]1> or do you mean the other way around?
<jbjnr> other way around. I have special rma:: types that serialization uses
<jbjnr> so I think memory::rma is the place to put them now
K-ballo has joined #ste||ar
<zao> gonidelis[m]: Rather confusing when you mention people by a different name than what they show as on IRC proper, where they're a five-letter word starting with j.
<zao> (not your fault, just an observation)
<gonidelis[m]> zao: starting with j?
<gonidelis[m]> zao: i let riot use autocorrect accroding to that name actually
<gonidelis[m]> i don't know how it shows to irc
hkaiser has joined #ste||ar
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
akheir has quit [Quit: Leaving]
nanmiao has joined #ste||ar
K-ballo1 has joined #ste||ar
diehlpk_work has joined #ste||ar
K-ballo has quit [Ping timeout: 240 seconds]
K-ballo1 is now known as K-ballo
jehelset has joined #ste||ar
<srinivasyadav227> @hk
<srinivasyadav227> hkaiser: i have written an initial version of proposal, where can i share it for review?
<hkaiser> srinivasyadav227: I think you can submit it to the google submission site, there you should be able to update it later
<pedro_barbosa[m]> Do I have to follow the instructions on https://github.com/STEllAR-GROUP/hpxcl in order to use HPXCL or is it implemented on the main HPX repository?
<hkaiser> pedro_barbosa[m]: it has not been integrated with the main repo
<hkaiser> also, it may not compile because of changes in HPX, we have not tried in a long time - please report any issues
<srinivasyadav227> hkaiser: ok, i have couple of doubts regarding "Add vectorization to par_unseq" project, is using OpenMP simd as backend for vectorization fine? because compiler specific pragmas for loop vectorization are not flexible and some of them are not portable
<hkaiser> yes, as a first stept #pragma simd would be fine, many compilers support it nowadays
<srinivasyadav227> #pragma omp simd? or #pragma simd?
<hkaiser> #pragma simd
<hkaiser> (I think)
<hkaiser> that's something to investigate
<srinivasyadav227> yes i have searched regarding loop pragmas for most commonly used compilers
<srinivasyadav227> #pragma simd is for intel
<gonidelis[m]> [after 8 hours ;p]
<hkaiser> gonidelis[m]: it's cmake-format, no?
<ms[m]1> gonidelis[m]: ah, yeah, don't worry about that, it's broken on master
<ms[m]1> (should fix it)
<hkaiser> the clang-format is an issue on master
<srinivasyadav227> hkaiser: ok, is there any restriction that i should use only compiler specific pragmas or i can use openmp library for vectorization?
<ms[m]1> hkaiser: gonidelis[m] it's both format checks
<hkaiser> srinivasyadav227: as long as OpenMP doesn't try to do any parallelization or launches it's own threads, sure
<hkaiser> ms[m]1: yah, I caused this - sorry
<srinivasyadav227> hkaiser: yes, sure
<ms[m]1> no worries!
<gonidelis[m]> hkaiser ms[m] hmm ok...
<hkaiser> just ignore the issue, alternatively, cherry pick this commit: https://github.com/STEllAR-GROUP/hpx/pull/5266/commits/a6e2c9ab76f468f8e14e42b6b5e7bfeae2c4abba
<gonidelis[m]> Ok thanks!
hkaiser has quit [Quit: bye]
nanmiao has quit [Quit: Connection closed]
hkaiser has joined #ste||ar
<pedro_barbosa[m]> I get this error (https://gist.github.com/PADBarbosa/0f22f43e9c2b378213066878e83dfe6f) when trying to use cmake (command included in the gist) on the HPXCL folder. The problem seems to be in the examples since it runs successfully if I use the -DHPXCL_BUILD_EXAMPLES=Off flag.
<hkaiser> pedro_barbosa[m]: this has to be HPX_DIR nowadays
<hkaiser> ms[m]1: yt?
bita has joined #ste||ar
<pedro_barbosa[m]> What should I do to fix it?
<hkaiser> pedro_barbosa[m]: no idea, need to look
nanmiao has joined #ste||ar
<hkaiser> pedro_barbosa[m]: would you mind creating a ticket here: https://github.com/STEllAR-GROUP/hpxcl/issues?
<K-ballo> every use of boost.accumulators only accumulates a single thing at a time
<pedro_barbosa[m]> <hkaiser "pedro_barbosa: would you mind cr"> Ticket created
<diehlpk_work> ms[m]1, is there any way to build kokkos using hpx without using nvcc?
<diehlpk_work> gdaiss[m], ?
<gdaiss[m]> Sure! It should work with clang or you could just disable the CUDA backend
<diehlpk_work> gdaiss[m], /lustre/scratch/pdiehl/PowerTiger/src/kokkos/bin/nvcc_wrapper: line 520: nvcc: command not found
<diehlpk_work> gdaiss[m], Ok, I found the workaround
<gdaiss[m]> sounds like the wrong CMAKE_CXX_COMPILER
<gdaiss[m]> you might want to try using the latest commit, especially when building with clang:
<gdaiss[m]> try octotiger commit b43babc7141c0be931bb79321bea7143ccb1bdd4 and for the build scripts 8ea85f5b63ab3a4aed34cdae7441625bf34efdb4
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
parsa has joined #ste||ar
parsa has quit [Client Quit]
jehelset has quit [Ping timeout: 240 seconds]
parsa has joined #ste||ar
parsa has quit [Client Quit]
parsa has joined #ste||ar
<hkaiser> pedro_barbosa[m]: thanks
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
parsa has joined #ste||ar
parsa has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
parsa has joined #ste||ar
nanmiao has quit [Quit: Connection closed]