aserio 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/
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 265 seconds]
EverYoun_ has quit [Ping timeout: 276 seconds]
EverYoung has joined #ste||ar
<github> [hpx] hkaiser force-pushed fixing_async from 0405dc8 to e1929df: https://git.io/vNjwO
<github> hpx/fixing_async e1929df Hartmut Kaiser: Minor changes to how actions are executed. This mostly improves consistency of different APIs.
<github> [hpx] hkaiser pushed 1 new commit to master: https://git.io/vNjwl
<github> hpx/master 4a99629 Hartmut Kaiser: Adding missing command line option enabling the application logging channel
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
mcopik has quit [Ping timeout: 240 seconds]
daissgr has quit [Quit: WeeChat 1.4]
jaafar_ has quit [Remote host closed the connection]
jaafar has joined #ste||ar
jaafar_ has joined #ste||ar
jaafar has quit [Ping timeout: 260 seconds]
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 240 seconds]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 260 seconds]
gedaj has quit [Remote host closed the connection]
gedaj has joined #ste||ar
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
hkaiser has quit [Quit: bye]
ct_clmsn has joined #ste||ar
ct_clmsn has quit [Client Quit]
nanashi55 has quit [Ping timeout: 248 seconds]
nanashi55 has joined #ste||ar
<heller_> simbergm: hey, in the sanitizer builds, you should turn of the stack overflow detection inside of HPX
<heller_> jbjnr: and I did notice! also the dashboard isn't as nice as it used to be!
<github> [hpx] sithhell closed pull request #3141: Fixing small_vector compatibility with older boost version (master...completion_handler) https://git.io/vNhLt
<github> [hpx] sithhell deleted completion_handler at 2394aac: https://git.io/vNjAT
gedaj has quit [Remote host closed the connection]
gedaj has joined #ste||ar
<simbergm> heller_: I think I do... HPX_WITH_STACKOVERFLOW_DETECTION=OFF
<heller_> hmm
<heller_> then something strange goes on
<simbergm> what did you see that's strange?
<heller_> ah no, nevermind
<heller_> misinterpreted the output
<simbergm> ok
<simbergm> but yeah, the sections are gone from the dashboard
<simbergm> I tried to put the santizers separately...
<heller_> *nod*
<heller_> now our PRs are all red again ;)
<heller_> do you want to get the failures fixed for the release?
<simbergm> yeah, release delayed by 6 months
<heller_> woot?
<simbergm> :P
<heller_> we should surely do a release this spring
<simbergm> joking, I didn't mean for us to do the sanitizer failures for the release
<heller_> ok ;)
<simbergm> just wanted to get them up and running
<heller_> just be aware that this generates noise, which makes it hard to spot "real" errors
<simbergm> do you know something about this migrate component failure?
<zao> They were green? :)
<heller_> no
<simbergm> heller_: yeah, I know, and I'm considering turning them off until the release even
<heller_> It failed previously, sometimes, but not that often
<simbergm> but if they'd be in their own section that would be good as well
<heller_> I know that migrate_component was buggy before
<simbergm> hmm, ok
* jbjnr is very disappointed
<simbergm> did you notice by the way that clang 4 in debug times out on the thread_stacksize test? the open PR is unlikely to fix that, do you think it's something similar with too small stack overhead?
<simbergm> jbjnr: why?
<heller_> simbergm: I will look into it, no idea
<heller_> jbjnr: just because noone complained, doesn't mean noone noticed ;)
<heller_> maybe we just didn't want to bug you?
<heller_> because we assumed you were cooking something up to make pycicle even better?
* zao huggles jbjnr
<jbjnr> I'm disappointedbecause last night I found a bug in my guided_pool_executor and fixed it. I set jobs going and was realy hoping that the results I got this morning would fix the small block size performance drop - but they didn't
<jbjnr> I was singing "we are the champions" in my head on the bus, but now I'm back to loser songs.
<heller_> oh no!
<jbjnr> Now, it's radiohead. Creep.
david_pfander has joined #ste||ar
gedaj has quit [Remote host closed the connection]
gedaj has joined #ste||ar
gedaj has quit [Remote host closed the connection]
gedaj has joined #ste||ar
<github> [hpx] sithhell created lazy_thread_init (+1 new commit): https://git.io/vAeed
<github> hpx/lazy_thread_init d562990 Thomas Heller: Changing the coroutine implementations to do a lazy init...
<github> [hpx] sithhell force-pushed lazy_thread_init from d562990 to 05fef46: https://git.io/vAeeA
<github> hpx/lazy_thread_init 05fef46 Thomas Heller: Changing the coroutine implementations to do a lazy init...
<github> [hpx] sithhell opened pull request #3146: Changing the coroutine implementations to do a lazy init (master...lazy_thread_init) https://git.io/vAeve
gedaj has quit [Remote host closed the connection]
gedaj has joined #ste||ar
gedaj_ has joined #ste||ar
<heller_> hwloc 2 is released
gedaj has quit [Ping timeout: 276 seconds]
<zao> Anything fun/breaking?
gedaj_ has quit [Remote host closed the connection]
gedaj_ has joined #ste||ar
<heller_> jbjnr: exceeded the quota?
<simbergm> heller_: that one is actually from me, I'll have a look
<heller_> simbergm: should we merge the stacksize fix right away?
<simbergm> heller_: if you think it's okay, but it looked like hello_world was hanging on appveyor so I restarted that one
<simbergm> was going to wait for it
<heller_> ah, right
<heller_> you can restart appveyor builds?
<simbergm> git commit --amend :)
<simbergm> with nothing changed
<heller_> lol
<heller_> ok
<simbergm> i.e. not really
<jbjnr> simbergm: Your PR on pycicle - what does it change? The regex gets the name - but the "origin' that is set goes onto the github PR status - does it change the link back to the dashboard or something?
<jbjnr> aha. just the name displayed on the PR ststus. I see
<jbjnr> ok fine
<simbergm> jbjnr: basically it just adds the compiler name to the PR status
<simbergm> before it was just the compiler version
<jbjnr> yup. Took me a minute, but I got there
<simbergm> well, maybe it wasn't so clear
<jbjnr> merged now
<simbergm> ok, nice
<simbergm> jbjnr: are you then restarting your pycicle for PRs? as it seems I have failed in doing so...
<jbjnr> I'm restarting with the new improved pycicle.
<simbergm> very nice
<jbjnr> just testing now on 3146
<jbjnr> want to make sure hpx still works
<simbergm> so you've added better support for multiple projects? more config options?
<github> [hpx] msimberg pushed 1 new commit to master: https://git.io/vAeqZ
<github> hpx/master 9fd1463 Mikael Simberg: Merge pull request #3142 from STEllAR-GROUP/fix-config-typo...
<jbjnr> yes. In future, once this new version is merged, you will need to start pycicle with
<jbjnr> python ./pycicle.py -P hpx - .....
<jbjnr> where the "-P hpx" tells it to look in the hpx config folder and use those settings
<jbjnr> and other projects can be supported in other config folders
<jbjnr> python ./pycicle.py -P dca
<jbjnr> etc etc
<jbjnr> simbergm: I do not approve of you filling my lovely dashboard with all your redness
<jbjnr> PS. heller_ next week we will upgrade cdash (probably)
gedaj_ has quit [Remote host closed the connection]
<heller_> jbjnr: yeha
gedaj_ has joined #ste||ar
<simbergm> jbjnr: I will not stop
<simbergm> (I might for now, but in general not)
<jbjnr> my first PR test of the new pycicle seems to have worked ok, so I will restart it for all PRs again in a short while
<jbjnr> why so much red? does it come from some sanitizer flags
<jbjnr> should I add some flags to my builds
<simbergm> jbjnr: good, I've disabled the non-sanitizer builds on my pycicle
<jbjnr> if you are finding real bugs/problems, then carry on
<simbergm> some are just me messing up the config, but the ones that actually finish properly are because of sanitizer complaining
<jbjnr> fine
<jbjnr> keepon then
<simbergm> I've tried to put them in their own section, will see with the next batch
<simbergm> will see if it worked
<jbjnr> next major pycicle upgrade will need to be to add option setting for many different possible flags etc...
gedaj_ has quit [Remote host closed the connection]
gedaj_ has joined #ste||ar
gedaj_ has quit [Remote host closed the connection]
hkaiser has joined #ste||ar
gedaj_ has joined #ste||ar
mcopik has joined #ste||ar
<jbjnr> hkaiser: yt?
<hkaiser> here
<jbjnr> see pm please
<github> [hpx] StellarBot pushed 1 new commit to gh-pages: https://git.io/vAelU
<github> hpx/gh-pages 1c60917 StellarBot: Updating docs
hkaiser has quit [Read error: Connection reset by peer]
<github> [hpx] msimberg opened pull request #3147: Move yield_while out of detail namespace and into own file (master...refactor-yield_while) https://git.io/vAe8Z
hkaiser has joined #ste||ar
K-ballo has quit [Ping timeout: 240 seconds]
daissgr has joined #ste||ar
gedaj_ has quit [Remote host closed the connection]
gedaj_ has joined #ste||ar
<github> [hpx] hkaiser pushed 3 new commits to master: https://git.io/vAeRB
<github> hpx/master 2053e4c Mikael Simberg: Revert "Revert "Adapting stack overhead limit for gcc 4.9""...
<github> hpx/master c099201 Mikael Simberg: Factor out debug defines into separate file
<github> hpx/master f48ac14 Hartmut Kaiser: Merge pull request #3143 from msimberg/fix_stack_overhead_2...
<jbjnr> simbergm: hopefully you will have noticed that pycicle is now running again and handling PRs once more. Seems to be stable. I will merge my latest changes tomorrow if pycicle runs overnight without any errors or strange behaviour
<simbergm> jbjnr: yes, it looks very good
<simbergm> thanks for setting it up again
<simbergm> I'm trying to set the track to "Pull Requests (Sanitizers)" for my builds but they seem to end up in Experimental
<simbergm> are the allowed tracks also fixed?
<simbergm> or should that work?
Smasher has quit [Ping timeout: 240 seconds]
gedaj_ has quit [Remote host closed the connection]
gedaj_ has joined #ste||ar
<jbjnr> simbergm: yes, if you want an extra track, I have to config thre dashboard for it
<jbjnr> to be honest - I'd say that they are fine in the PR track, as long as sanitizer is in the name alongside boost/clang/etc
<simbergm> jbjnr: ah, okay
<jbjnr> but if you want a track, just ask and I'll add it
<simbergm> I wanted to avoid cluttering up the normal results especially now most tests fail with sanitizers
<simbergm> I'm happy either way
<jbjnr> if they fail -then we must fix them
<simbergm> yes, but I see that as unlikely for the release
<jbjnr> what we should do is add 'options' to the build so we can just fo sanitizer builds as part of the normal setup
<simbergm> yeah, that would be nice
<jbjnr> why "unlikely"
<jbjnr> too much work?
<simbergm> I think so, yes
daissgr has quit [Quit: WeeChat 1.4]
daissgr has joined #ste||ar
<hkaiser> at leastth eaddress sanitizer failures seem to be caused by one and the same code
<jbjnr> hkaiser: does a 'dataflow' statement create a task, or just a continuation thingy
hkaiser has quit [Quit: bye]
<jbjnr> (not the lambda that is caled, but the delayed futures bit)
<jbjnr> balls. he left
<heller_> jbjnr: depends on the launch policy ;)
<jbjnr> that will control the execution of the lambda - I guess that's the best I can do
daissgr has quit [Quit: WeeChat 1.4]
<jbjnr> the actual futures will become ready when they do ...
<jbjnr> I've used this in some places "return dataflow(launch::sync,
<jbjnr> ..." but not everywhere
<jbjnr> and now I don't remember why
daissgr has joined #ste||ar
aserio has joined #ste||ar
<heller_> jbjnr: dataflow itself creates a shared state
<heller_> which is not a task
<heller_> just the shared state to signal completion to the caller
<jbjnr> ok. that answers my question. I figured it was like that. Thanks
<jbjnr> didn't know the right terminology there.
diehlpk has joined #ste||ar
galabc has joined #ste||ar
hkaiser has joined #ste||ar
eschnett has joined #ste||ar
<simbergm> it allocates 12 Gb on rostam
<simbergm> twice
<hkaiser> lol
<hkaiser> yah, that should be changed, sorry for that
<simbergm> :)
<simbergm> will you change it?
<hkaiser> ok
<simbergm> either way, not sure exactly what you want to test there except the reinit function
<hkaiser> yah
galabc has quit [Ping timeout: 260 seconds]
<simbergm> I can change it as well, anything fixed, > 1, and smaller than 120000000 will do?
<github> [hpx] hkaiser pushed 1 new commit to master: https://git.io/vAeDM
<github> hpx/master 78d0d88 Hartmut Kaiser: Fixing test note insane amounts of memory
<simbergm> ah, thanks :)
<hkaiser> simbergm: done
<hkaiser> sorry for the garbled commit message, half of it got lost in the ether
galabc has joined #ste||ar
galabc has quit [Ping timeout: 260 seconds]
<heller_> lol. Cosmic rays are the worst
<hkaiser> heller_: most true if they hit the brain ;)
<hkaiser> heller_: btw, nice outcome ODR example!
<hkaiser> you clearly have too much time at your hands
<heller_> In telcos all day...
<hkaiser> fun
<heller_> And this stupid procrastination problem
<heller_> I don't know why Niall is constantly triggering me...
<hkaiser> heller_: 'someone on the internet is wrong'
eschnett has quit [Read error: Connection reset by peer]
<heller_> Yeah...
eschnett has joined #ste||ar
diehlpk has quit [Quit: Leaving]
<zao> heller_: At least you're not me, I was arguing in youtube comments the other day.
<heller_> Even better ;)
<heller_> I hope it was a video about the flat earth?
EverYoung has joined #ste||ar
Smasher has joined #ste||ar
<zao> Gods no, it was something way dumber.
galabc has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
david_pfander has quit [Ping timeout: 240 seconds]
galabc has quit [Ping timeout: 260 seconds]
hkaiser has quit [Quit: bye]
bibek has quit [Quit: Leaving]
aserio1 has joined #ste||ar
bibek has joined #ste||ar
aserio has quit [Ping timeout: 240 seconds]
aserio1 is now known as aserio
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
Smasherr has joined #ste||ar
aserio has quit [Ping timeout: 252 seconds]
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
aserio has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
vamatya has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoun_ has quit [Remote host closed the connection]
EverYoung has quit [Ping timeout: 260 seconds]
EverYoung has joined #ste||ar
Smasher has quit [Remote host closed the connection]
Smasherr is now known as Smasher
jaafar_ is now known as jaafar
aserio has quit [Ping timeout: 240 seconds]
hkaiser has joined #ste||ar
daissgr has quit [Ping timeout: 276 seconds]
hkaiser has quit [Read error: Connection reset by peer]
aserio has joined #ste||ar
hkaiser has joined #ste||ar
K-ballo has joined #ste||ar
eschnett has quit [Quit: eschnett]
eschnett has joined #ste||ar
aserio has quit [Read error: Connection reset by peer]
aserio has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 260 seconds]
EverYoun_ has quit [Ping timeout: 276 seconds]
EverYoung has joined #ste||ar
daissgr has joined #ste||ar
eschnett has quit [Quit: eschnett]
daissgr has quit [Ping timeout: 240 seconds]
shahrzad has joined #ste||ar
shahrzad has quit [Client Quit]
shahrzad has joined #ste||ar
daissgr has joined #ste||ar
shahrzad has quit [Client Quit]
shahrzad has joined #ste||ar
eschnett has joined #ste||ar
hkaiser has quit [Quit: bye]
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
hkaiser has joined #ste||ar
<github> [hpx] hkaiser deleted fixing_2996 at f49573f: https://git.io/vAvCC
mbremer has joined #ste||ar
EverYoung has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 252 seconds]
<mbremer> Would some one mind giving me feed back on a serialization idea?
<mbremer> I have a class that looks effectively like this: https://gist.github.com/bremerm31/2281f99c4d76796db49520ded97ef7e6
<mbremer> It's a silly example, but encapsulates what I'd like to do. Namely, assume that T is expensive to copy around, but also that it's a little tricky to know exactly where T* should point to. SO I want to serialize both the pointer and the object
<mbremer> Right now I'm thinking of writing an observer type pointer to replace the raw pointer ( http://en.cppreference.com/w/cpp/experimental/observer_ptr ) and then moving varr into a unique pointer
<mbremer> And then tracking all of the pointers. My hope is that this will keep the pointers pointing to the correct objects.
<mbremer> Would this type of approach work for the pointer tracking serialization?
aserio has quit [Quit: aserio]
<hkaiser> mbremer: not sure
<mbremer> Is that clear? I also looked more around. It seems want essentially raw pointers and unique pointers, but I want them to be tracked.
<mbremer> I guess I'll write up a unit test and see how it goes :)
<hkaiser> what do you mean by 'tracked'?
<mbremer> using hpx::serialization::detail::serialize_pointer_tracked
<mbremer> instead of ------//-------_untracked
<hkaiser> marco: that would mean that serializing the same pointer more than once in the same archive would a) transfer the data once and b) recreate the two pointers to the same data on the other end
<hkaiser> mbremer
<hkaiser> ^^
<mbremer> yes!
<hkaiser> I don't think that pointer tracking works with a raw pointer
<hkaiser> it does not work with a unique_ptr either as it does not has to (unique_ptr's are unique by design)
<hkaiser> pointer tracking will kick in for shared_ptr's only, iirc
<hkaiser> mbremer: heller_ might know better
<mbremer> Yeah, I guess the idea, is connecting pointers on the other end is a little tricky, so if the archive could keep them connected for me I would be very happy
<hkaiser> should work for shared_ptr's
<mbremer> Ya, I was kind of hoping to avoid those. It's not really consistent with who owns the data, and I was hoping to avoid the costs of shared pointer
<mbremer> @hkaiser: Thanks, anyway! I might think about it some more. The order is pretty problem specific. So I might just implement my own form tracking...(should just be a (unordered_)/map)
<hkaiser> lol
<hkaiser> mbremer: talk to heller_, he will know more about possible custmization points
<mbremer> Sure thing, I'll try pinging him tomorrow. Thanks!
mbremer has quit [Quit: Page closed]
<github> [hpx] hkaiser created refactor_base_action (+1 new commit): https://git.io/vAvro
<github> hpx/refactor_base_action 6498068 Hartmut Kaiser: Refactoring component_base and base_action/transfer_base_action to reduce number of instantiated functions and exported symbols
<github> [hpx] hkaiser opened pull request #3148: Refactoring component_base and base_action/transfer_base_action (master...refactor_base_action) https://git.io/vAvrQ