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/
hkaiser has quit [Quit: bye]
rgoswami has joined #ste||ar
hkaiser has joined #ste||ar
hkaiser has quit [Quit: bye]
chuanqiu has joined #ste||ar
chuanqiu has quit [Quit: Connection closed]
chuanqiu has joined #ste||ar
chuanqiu has quit [Client Quit]
nanmiao has joined #ste||ar
nanmiao has quit [Quit: Connection closed]
<pedro_barbosa[m]> diehlpk_work: you there?
nanmiao has joined #ste||ar
chuanqiu has joined #ste||ar
chuanqiu has quit [Client Quit]
chuanqiu has joined #ste||ar
<YourNick[m]> Is there any reason for us not having `std::accumulate()`
<YourNick[m]> ?
hkaiser has joined #ste||ar
chuanqiu has quit [Quit: Connection closed]
chuanqiu has joined #ste||ar
chuanqiu has quit [Client Quit]
chuanqiu has joined #ste||ar
chuanqiu has quit [Quit: Connection closed]
<ms[m]> gonidelis[m]: accumulate must be sequential, we have reduce instead which is allowed to run in parallel
<YourNick[m]> ms[m]: cool thanks!
gonidelis has joined #ste||ar
<YourNick[m]> hkaiser: pm ;)
<YourNick[m]> Does anyone sees my nick as "YourNick" ?
YourNick[m] is now known as NewNickname[m]
<bering[m]> Nope.
<NewNickname[m]> bering[m]: thanks
gonidelis is now known as gonidelis2
<ms[m]> hkaiser: should we still wait for adrian to respond?
hkaiser_ has joined #ste||ar
<hkaiser_> ms[m]: probably not, let's send it out
<rachitt_shah[m]> Just remembered, we wouldn't be needing the user feedback form at the moment, if I'm correct?
hkaiser has quit [Ping timeout: 260 seconds]
hkaiser_ has quit [Client Quit]
hkaiser has joined #ste||ar
diehlpk_work has quit [Remote host closed the connection]
hkaiser has quit [Quit: bye]
gonidelis2 has quit [Quit: Connection closed]
<ms[m]> rachitt_shah: well, it's not strictly required, but it certainly wouldn't hurt either
<rachitt_shah[m]> <ms[m] "rachitt_shah: well, it's not str"> Let me float it then, I've ran the docs locally, I'll ping if I'm stuck
<ms[m]> 👍️
hkaiser has joined #ste||ar
diehlpk_work has joined #ste||ar
<diehlpk_work> pedro_barbosa[m], I am here now
<hkaiser> NewNickname[m]: are you sure your laptop has really 16 cores? or is it 8 cores with hyper-threading?
<NewNickname[m]> 8 cores with hyper threading
<NewNickname[m]> hkaiser why?
<hkaiser> because that would make a big difference
<hkaiser> if it was 8 cores with hyper-threading it would not make sense to scale out to 16 threads
<NewNickname[m]> really?
<NewNickname[m]> I didn't know that
<NewNickname[m]> why?
<hkaiser> because HTs are not real cores
<NewNickname[m]> yeah I know... but don't they work like logical cores? I mean, do we expect the samer result from an 8core system without hyperthreading?
<hkaiser> using the HTs usually slows things down, at least with HPX
<hkaiser> if you need real 16 or more cores, do it on rostam
<NewNickname[m]> wow
<NewNickname[m]> yeah I certainly will
<NewNickname[m]> this is only for me to be sure that things do scale
<hkaiser> ok
<pedro_barbosa[m]> diehlpk_work: , I was trying to convert this kernel launch cycle (GIST: https://gist.github.com/PADBarbosa/3d703ef121606f614c259239039e8443) to HPXCL and was trying to follow the suggestion you recommended a few days ago, but I'm not sure I understood it correctly
<diehlpk_work> Ok, how far were you able to go?
<diehlpk_work> Here, you cna find some nice example for dataflow
<hkaiser> NewNickname[m]: btw, for me parallel copy is faster than the sequential copy
<pedro_barbosa[m]> not much I tried this out (GIST: https://gist.github.com/PADBarbosa/b2b5519dd7c9e902bd8a7108d7b35426) cause I couldn't understand how to do it with dataflow
<NewNickname[m]> all our algorithms scale very well compared to their sequential versions
<NewNickname[m]> hkaiser: for 10m elements we even get sth close to x4.2 speedup
<NewNickname[m]> no hpx algo thus far is slower in parallel than sequential
<hkaiser> NewNickname[m]: your cilk copy runs a parallel version of std::copy, btw
<NewNickname[m]> hkaiser I pack up std algos with cilk because they have similar compilation rules
<NewNickname[m]> I print both std and cilk
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
hkaiser has quit [Quit: bye]