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/
hkaiser has joined #ste||ar
Coldblackice_ is now known as Coldblackice
nikunj has joined #ste||ar
hkaiser has quit [Quit: bye]
K-ballo1 has joined #ste||ar
K-ballo has quit [Read error: Connection reset by peer]
K-ballo1 is now known as K-ballo
nikunj has quit [Ping timeout: 250 seconds]
akheir has quit [*.net *.split]
daissgr has quit [*.net *.split]
jaafar has quit [*.net *.split]
diehlpk_work has quit [*.net *.split]
akheir has joined #ste||ar
daissgr has joined #ste||ar
jaafar has joined #ste||ar
diehlpk_work has joined #ste||ar
daissgr_ has joined #ste||ar
mdiers_ has quit [Quit: mdiers_]
mdiers_ has joined #ste||ar
daissgr_ has quit [Ping timeout: 258 seconds]
mdiers_ has quit [Remote host closed the connection]
mdiers_ has joined #ste||ar
daissgr_ has joined #ste||ar
mdiers_ has quit [Remote host closed the connection]
mdiers_ has joined #ste||ar
mdiers_ has quit [Remote host closed the connection]
hkaiser has joined #ste||ar
mdiers_ has joined #ste||ar
bibek has quit [Quit: Konversation terminated!]
bibek has joined #ste||ar
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
Karame has joined #ste||ar
nikunj97 has joined #ste||ar
daissgr_ has quit [Remote host closed the connection]
mdiers_ has quit [Remote host closed the connection]
<lsl88> hi!
<lsl88> I found some small typos in the documentation, would you like me to fix them while I read it?
<hkaiser> lsl88: yes, pls
Karame has quit [Ping timeout: 252 seconds]
<lsl88> hkaiser: ok :)
hkaiser has quit [Quit: bye]
akheir has quit [Quit: Leaving]
<diehlpk_work> SC 19 paper is considered for acceptance, if we address two major issues :)
<Yorlik> Disclaimer: Newbie Question:: I just discovered the existence of std::span<> and wonder if I should use it for my custom allocators of pods in an array like memory region. Opinions? Advice?
<Yorlik> (Instead of C pointer like arithmetic)
<zao> The GSL is cute, but I found it largely hurt my small plugins.
<zao> Spans are great, but they're greater as slices in Rust :P
<zao> I found that they were mostly useless for textual operations, as no existing algorithms work with pointer+len tuples.
<zao> (strstr, etc.)
<parsa> zao: wash says hi!
<zao> \o/
hkaiser has joined #ste||ar
wash[m] has joined #ste||ar
<zao> parsa: You're at some event or do you have nice visitors? :)
<zao> Yorlik: I didn't actually know that they'd pulled them into C++ proper in 20.
<Yorlik> zao: I think I'll rewrite my allocators to use spans. Looks more correct to me and safer.
<diehlpk_work> hkaiser, Check your inbox
<zao> Might get some more traction in algorithms then.
<parsa> zao: he's actually here
<Yorlik> In the moment I'm busy writing tests with catch2 and the catch2 test adapter for VS
<diehlpk_work> hkaiser, SC19 notifications are in
<Yorlik> Unfortunately catch2 doesn't yet support death tests, e.g. I couldn't test my segfaults in memory management, but I like the rest of it and ease of use a lot.
diehlpk_mobile has joined #ste||ar
<hkaiser> diehlpk_mobile: nice!
<diehlpk_work> yes, I think the two remarks are doable
<hkaiser> nod
<nikunj97> hkaiser: yt?
<hkaiser> nikunj97: here
<nikunj97> hkaiser: could you please go through the PR. I will benchmark the PR and generate the graphs before the meeting on Tuesday.
<hkaiser> will do over the weekend
<hkaiser> thanks
<nikunj97> also we still have to figure something out about the negative overheads with dataflow
<nikunj97> I simply can't understand what's going wrong in there
<nikunj97> but we definitely can't get negative overheads
<hkaiser> lol
<hkaiser> let's leave it in ;-)
<hkaiser> sounds cool anyways
<nikunj97> yes! very.. but I don't have any feasible way to explain the negative overheads xD
diehlpk_mobile has quit [Ping timeout: 258 seconds]
<hkaiser> youtime machine
<hkaiser> you invented a time machine
<nikunj97> the guys from Sandia will be pretty excited to see those
<hkaiser> let's see what's going on there
<nikunj97> maybe it's got something to do with caching.. coz otherwise I simply can't explain that
<hkaiser> yah, superscalar behavior is mostly caused by caching effects
<nikunj97> let me create separate benchmarks and see if that's the case
<nikunj97> until then, we're gonna show negative overheads to them
diehlpk_mobile has joined #ste||ar
<hkaiser> ok
diehlpk_mobile has quit [Read error: Connection reset by peer]
daissgr_ has joined #ste||ar
<hkaiser> diehlpk_work: : have you seen this problem on Fedora before: https://travis-ci.com/laristra/flecsi-third-party/jobs/208179764?
<hkaiser> looks like an -latomic issue again :/
<nikunj97> hkaiser: it's most likely cache issues. I just ran them separately and they took good amount of extra time. I'll do the benchmarks for them running all the codes separately.
<nikunj97> maybe that's also the reason why we see validate to take lot lower than normal.
<nikunj97> I'll split everything into different fragments and then benchmarks just to get true results. I'll run the benchmarks over the weekend and generate the final sets of graphs before the meeting
<daissgr_> Ah, the SC paper did not get rejected in the second phase! That's the kind of news I like to receive when arriving at home :-)
<daissgr_> nice!
nikunj97 has quit []
diehlpk_mobile has joined #ste||ar
diehlpk_mobile has quit [Ping timeout: 252 seconds]