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/ | GSoC: https://github.com/STEllAR-GROUP/hpx/wiki/Google-Summer-of-Code-%28GSoC%29-2020
hkaiser has quit [Quit: bye]
nan_ has quit [Remote host closed the connection]
akheir has quit [Quit: Leaving]
<heller1> Yorlik: i see. so in this case, the "commit" operation is made explicit
<heller1> Yorlik: (assuming that you have a mental model of atomic operations that undergo a modify/commit/rollback mechanism)
nikunj has quit [Ping timeout: 240 seconds]
nikunj has joined #ste||ar
diehlpk_work has quit [*.net *.split]
Vir has quit [*.net *.split]
jaafar has quit [*.net *.split]
freifrau_von_ble has quit [*.net *.split]
K-ballo has quit [*.net *.split]
Yorlik has quit [*.net *.split]
Amy1 has quit [*.net *.split]
karame_ has quit [*.net *.split]
weilewei has quit [*.net *.split]
Guest8932 has quit [*.net *.split]
gdaiss[m] has quit [*.net *.split]
diehlpk_mobile[m has quit [*.net *.split]
ms[m] has quit [*.net *.split]
rori has quit [*.net *.split]
pfluegdk[m] has quit [*.net *.split]
nikunj has quit [*.net *.split]
tiagofg[m] has quit [*.net *.split]
heller1 has quit [*.net *.split]
tarzeau has quit [*.net *.split]
parsa has quit [*.net *.split]
mdiers[m] has quit [*.net *.split]
kordejong has quit [*.net *.split]
mdiers_ has quit [*.net *.split]
jbjnr has quit [*.net *.split]
zao has quit [*.net *.split]
wash[m] has quit [*.net *.split]
bobakk3r has quit [*.net *.split]
nikunj has joined #ste||ar
tiagofg[m] has joined #ste||ar
Guest8932 has joined #ste||ar
jaafar has joined #ste||ar
mdiers_ has joined #ste||ar
jbjnr has joined #ste||ar
pfluegdk[m] has joined #ste||ar
Vir has joined #ste||ar
Yorlik has joined #ste||ar
rori has joined #ste||ar
karame_ has joined #ste||ar
ms[m] has joined #ste||ar
weilewei has joined #ste||ar
K-ballo has joined #ste||ar
diehlpk_work has joined #ste||ar
mdiers[m] has joined #ste||ar
freifrau_von_ble has joined #ste||ar
kordejong has joined #ste||ar
parsa has joined #ste||ar
gdaiss[m] has joined #ste||ar
Amy1 has joined #ste||ar
bobakk3r has joined #ste||ar
heller1 has joined #ste||ar
zao has joined #ste||ar
tarzeau has joined #ste||ar
wash[m] has joined #ste||ar
diehlpk_mobile[m has joined #ste||ar
karame_ has quit [Remote host closed the connection]
gonidelis has joined #ste||ar
<gonidelis> Hello @channel. First of all I would like to thank you in my turn for all your support. You have proven to be a really interesting and helping group. I would also like to express my gratitude for trusting and accepting me for the Summer Period to work with you. I promise I will do my best. Is there any place where we (summer students) could talk
<gonidelis> specifics with the mentors in order not to spam on this very fine channel?
<zao> gonidelis: Welcome!
<gonidelis> zao Thank you!!!
<heller1> gonidelis: hey there
<heller1> gonidelis: this channel is just fine. We want to setup a zoom call with you however in the next few days
mcopik has joined #ste||ar
mcopik has quit [Client Quit]
<Yorlik> heller1: I do think, that this datastructure has the potential to be used in many different ways, could be implemented as template parameters: lock-free / mutable vs. immutable (w. lock-free under the hood for said operations). Rollback could probably be orthogonal to that.
<gonidelis> heller1 Oh ok I get it. Thank you. I am looking forward to the meeting... :D
<zao> GSoC is nice, livens up the channel.
gonidelis has quit [Ping timeout: 245 seconds]
gonidelis has joined #ste||ar
hkaiser has joined #ste||ar
gonidelis has quit [Ping timeout: 245 seconds]
gonidelis has joined #ste||ar
gonidelis has quit [Ping timeout: 245 seconds]
karame_ has joined #ste||ar
gonidelis has joined #ste||ar
gonidelis has quit [Remote host closed the connection]
gonidelis has joined #ste||ar
<heller1> Yorlik: yt?
<Yorlik> heller1: I'm here now.
<Yorlik> Just was in the PC shop an ordered an AMD 3900 based machine. Fed up with the the compile wait. :)
<zao> Welcome to team Red.
<Yorlik> I'll keep around my Trusty old 4790k though ...
<Yorlik> 64 GB memory will also allow some virtual clustering
<Yorlik> zao: I guess technically I'm team purple now :)
<hkaiser> lol
nikunj has quit [Ping timeout: 240 seconds]
nikunj has joined #ste||ar
rtohid has joined #ste||ar
<hkaiser> nikunj: that was an angry GSoC student yesterday - what did you do to him?
<K-ballo> link..?
<diehlpk_work> hkaiser, Abishek was just angry becuase he was not accepted
<hkaiser> I think I even told him to select a different project to avoid being outsmarted
<diehlpk_work> He was angry because we selected a friend of Nikunj and assumed that we just did it because he knows Nikunj
<nikunj> hkaiser: I didn't do anything :/
<diehlpk_work> His proposals was much better, but we all supported Kale because of whatever reason
<nikunj> Fwiw, I helped everyone that came on irc to the best I could. That includes Abhishek as well.
<hkaiser> nikunj: sure
<K-ballo> nikunj: obviously, you shouldn't have friends
<weilewei> lol
<gonidelis> is there any way to prevent such behavior from the channel? Sth like ban from administration or sth?
<heller1> Yorlik: do you have that trace with the memory leak still handy?
nikunj has quit [Ping timeout: 260 seconds]
<Yorlik> I would have to create a new one I think - let me check
nikunj has joined #ste||ar
<Yorlik> heller1: I can make a new one if you want to look into it - the old one is gone.
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
<heller1> That would be cool
<jbjnr> nikunj: did you finish your work on the stencil codes and simd stuff?
<nikunj> jbjnr: yes
<nikunj> Finalizing my report these days
<jbjnr> great. look forward to seeing it
akheir has joined #ste||ar
<nikunj> jbjnr: i will upload the results to arxiv. Unfortunately it's not novel to get a workshop paper :/
<hkaiser> nikunj: you could try to get it working with hpx algorithms and a datapar execution policy ;-)
<nikunj> Is that a new execution policy?
<nikunj> And which hpx algorithm?
<nikunj> I use parallel_for_loop already
bita_ has joined #ste||ar
<Yorlik> heller1: That specific leak seems to be gone in the moment, probably it was realted to that error with the lock checking in debug mode.
<Yorlik> There are some other leaks not related to hpx.
gonidelis36 has joined #ste||ar
<Yorlik> I suggest, when the big error is fixed, I revisit everything and see if there are remaining leaks
gonidelis has quit [Ping timeout: 245 seconds]
<Yorlik> There are still leaks going through hpx though - I wonder if it would be premature to visit them right now
<Yorlik> I htink I need to brush my code more
<diehlpk_work> hkaiser, operation bell meeting
<diehlpk_work> Will you join?
<diehlpk_work> An Optimum Tea Fermentation Detection Model Based on Deep Convolutional Neural Networks
<diehlpk_work> G Kimutai, A Ngenzi, RN Said, A Kiprop, A Förster - Data, 2020
<diehlpk_work> Our first Phylanx citation
<Yorlik> join #valve
<Yorlik> woops
gonidelis36 is now known as gonidelis
<gonidelis> Are you guys have some sort of way to log in at the IRC channel automatically without filling name and channel id? Also the usernames that end with '[m]' are users from their mobile phones?
<jbjnr> we are using some truly awful thing called matrix/riot that addm to our names and takes 30 seconds to post every comment.
<jbjnr> `/rant off`
<gonidelis> Is there any way I can keep my username stored and not have to reenter it every time? I mean, what if someone else enters with your names?
<heller1> Yorlik: ok, I'm really interested. I'm running into an asan segfault at shutdown when it wants to print the leaks...
<heller1> gonidelis: nickserv
<jbjnr> nikunj: the comment "great. look forward to seeing it" that just appeared, was sent by me about an hour ago, it just appeared in the chat now!
<Yorlik> heller1: I'm using a trial version of deleaker in the moment - I might just pay for a private license later or use vld - I still have to evaluate the free versions and how good they are (on windos)
<gonidelis> 'nickserv'?
<jbjnr> The mirroring between HPX and #stellar in this riot thingy is messing me up.
<Yorlik> They started adding ASAN support, but in the moment it's 32 bit only - whish sucks, but hopefully that will change
<nikunj> jbjnr: woah. That was ages ago
<jbjnr> I'm having trouble with tis irc stuff. this riot/matrix interface just doesn't seem to work. I gave up for a few weeks, but it thought I was still logged in and I missed messages
<jbjnr> I don't know if I should post to #stellar or HPX,
nan_ has joined #ste||ar
<nikunj> jbjnr: try using hexchat?
<nikunj> Have always worked nicely for me
* Yorlik uses hexchat
<heller1> biddisco: works for me usually
<jbjnr> nikunj: I am trying to avoid having too many application windows ope, so went for this riot thingy cos it is web browser based and everyone said it was great.
<Yorlik> Discord FTW (#it_is_2020) ;)
<nikunj> I'd love to see stellar on discord as well
<jbjnr> nikunj: and I wanted somethig that would record history when I was shut down. I think hexchat doesn't pickup old messages ...
<nikunj> Or everything on slack
<nikunj> jbjnr: hexchat does log messages as long as it is open. So if you are not logged in with Hexchat, it will not log those messages.
<Yorlik> Slack or Discord, I'd agree to that.
<nikunj> Btw if you're a phone person, you can install revolution irc.
<nikunj> I'm always logged onto stellar through phone. nikunj is through revolution irc and nikunj97 through laptop using hexchat.
<gonidelis> could you tell me who is Auriane Reverdell ? He is one of my two mentors...
<gonidelis> he or she*
<jbjnr> She works at CSCS and is one of our HPX team (we are 3 people)
<gonidelis> You mean you are 3 people that work at HPX from CSCS?
<hkaiser> gonidelis: Auriane's nick here is rori
<gonidelis> perfect... thank you!
Vir has quit [Excess Flood]
Vir has joined #ste||ar
<jbjnr> gonidelis: yes. There's me (biddisco on github), Mikalel (@ms or simbergm) and Aurianne/Rori. We are developing HPX, but there are others who work on applications that are "using" hpx, when they have problems, we do our best to fix them.
<jbjnr> ^Mikael not Mikalal
<gonidelis> That's great! Thank you
<gonidelis> I like learning about team's members relations
<jbjnr> Currently we have people developing linear algebra libraries using hpx, n-body simulations (proof of concept), potentially (including the work that weilewei is doing at LSU) monte carlo material structure simulations. etc etc
<weilewei> Hi there :)
gonidelis has quit [Ping timeout: 245 seconds]
gonidelis has joined #ste||ar
<hkaiser> ms[m]: yt?
<hkaiser> nikunj: 'which algorithm'? let's start with for_each or copy
<hkaiser> also datapar is an execution policy from the parallelism TS V2
<hkaiser> not sure if it's still called datapar, you'll figure it out
<hkaiser> ;-)
<nikunj> hkaiser: I don't use copy. Will have to look at datapar though
<hkaiser> what algorithm do you use?
<hkaiser> nikunj: ?
<nikunj> I use the tactic you have in stencil
<hkaiser> also, we do have some support for this, but built around VC
<nikunj> Where you don't swap
<nikunj> Also I use fill
<nikunj> Also I need to make sure if I use hpx algorithms, they work with nsimd
<heller1> hkaiser: do you remember which global had the memory leak in Yorlik's trace?
<zao> gonidelis: I’m Lars Viklund, HPC sysadmin at Umeå University, and I do nothing at all with HPX but keep these people entertained :)
<gonidelis> '=D '=D Actually you have answered lots of my questions so you are one of the top mentors as far as I am concerned
<hkaiser> heller1: some global in the C runtime system
<heller1> ok
<gonidelis> I have literally friends working on LSU (physics department), Sweden (Lund - biomedical engineering) and ETH (some PHD CS guy). The world is so smal...
<heller1> and surely not inside HPX?
<hkaiser> heller1: afair no
<heller1> hmm, what a bummer ... the leak sanitizer is segfaulting when it is trying to report the leaks...
<heller1> so I am guessing that this is something inside of HPX
<heller1> but I can't debug this...
<Yorlik> heller1: I will get this reproduced, since I know how it happened - I'm juts busy with this help my codev needs right now.
<heller1> sure
<heller1> no worries
<heller1> Yorlik: which jemalloc version were you using?
<Yorlik> Latest
<Yorlik> Latest tagged to be precise
<Yorlik> But when I investigated the leaks I had it off
<heller1> ok
<heller1> it looks like it is trying to process a potentially leaked thing from a TLS
gonidelis has quit [Ping timeout: 245 seconds]
diehlpk_work has quit [Ping timeout: 260 seconds]
rtohid has quit [Ping timeout: 245 seconds]
bita_ has quit [Ping timeout: 240 seconds]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Ping timeout: 260 seconds]
nikunj has joined #ste||ar