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/
parsa has joined #ste||ar
parsa has quit [Quit: Zzzzzzzzzzzz]
parsa has joined #ste||ar
parsa has quit [Client Quit]
parsa has joined #ste||ar
parsa has quit [Quit: Zzzzzzzzzzzz]
parsa has joined #ste||ar
EverYoung has quit [Ping timeout: 258 seconds]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 258 seconds]
jaafar has quit [Ping timeout: 240 seconds]
vamatya_ has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
vamatya_ has joined #ste||ar
vamatya_ has quit [Ping timeout: 248 seconds]
K-ballo has quit [Quit: K-ballo]
taeguk has joined #ste||ar
parsa has quit [Quit: Zzzzzzzzzzzz]
vamatya_ has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 246 seconds]
jaafar has joined #ste||ar
vamatya_ has quit [Ping timeout: 248 seconds]
jaafar has quit [Ping timeout: 255 seconds]
parsa[[w]] has joined #ste||ar
jbjnr_ has joined #ste||ar
jbjnr has quit [Ping timeout: 246 seconds]
jbjnr_ is now known as jbjnr
parsa[w] has quit [Ping timeout: 246 seconds]
<zao> Heh, ninja -j 16 on HPX... I forgot that tests eat memory like candy.
<zao> Hello swap my old friend.
troska has joined #ste||ar
<zao> This was surprising...
<zao> 98% tests passed, 12 tests failed out of 550
<zao> Better than expected :D
<heller> zao: which platform?
<zao> x86_64, Arch Linux, GCC 7.2.0, hwloc, Boost 1.65.1 C++14.
<zao> Ryzen.
<heller> nice
<heller> I have a patch for the failing 12 tests
<zao> Evaluating if the NAS is stable enough on this OS to also be used to build evil C++.
<heller> ;)
<heller> looks good. the tests failing are known
<zao> I mean, put 32G of memory in it and those 16 threads ought to fly.
K-ballo has joined #ste||ar
parsa[[[w]]] has joined #ste||ar
parsa[[w]] has quit [Ping timeout: 246 seconds]
taeguk has quit [Quit: Page closed]
hkaiser has joined #ste||ar
<heller> hkaiser: somehow, for_loop is broken :/
<heller> it doesn't seem to run in parallel for random access iterators
<heller> at least the tutorial stencil example is performing very poorly, live demo
parsa has joined #ste||ar
<hkaiser> heller: ok
<hkaiser> need to look
<heller> hkaiser: found a way forward
<heller> so no hurry
<hkaiser> what's wrong there?
<heller> no idea :(
<hkaiser> ok
<hkaiser> pls create a ticket
<heller> the for_loop body seems to be executed sequentially
<heller> according to the APEX profile and the idle rates
<hkaiser> how to reproduce?
<hkaiser> heller: ^^
<heller> hkaiser: working on it
<hkaiser> k
<heller> I am still in the tutorial
<hkaiser> sure, just pls don't forget it
<K-ballo> what happened to the VS builders?
<heller> I won't ;)
<hkaiser> K-ballo: isn't appveyor running?
<K-ballo> I meant the buildbot ones, were they replaced by appveyor?
<hkaiser> yes
<hkaiser> I didn't set them up again after the machine died
heller has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
<K-ballo> do I need to open a PR for appveyor to build a branch?
<hkaiser> K-ballo: don't think so, it should build even from non-pr branches
<K-ballo> mmh, indeed, I see the std-regex branch in there
<hkaiser> it will just stop building the branch as soon as it's a pr
<K-ballo> but I don't see the sfinae-expression-complete one for some reason
<hkaiser> K-ballo: yah, you're right - don't see it either :/
heller has joined #ste||ar
<K-ballo> there we go
<github> [hpx] K-ballo force-pushed sfinae-expression-complete from 69161c9 to 2b0c219: https://git.io/v5oJd
<github> hpx/sfinae-expression-complete 3d3bd7d Agustin K-ballo Berge: Drop special case config macro for "complete" SFINAE support
<github> hpx/sfinae-expression-complete a7673b0 Agustin K-ballo Berge: Remove empty on-framework feature test cmake functionality
<github> hpx/sfinae-expression-complete 2b0c219 Agustin K-ballo Berge: Remove duplicated reserve_if_possible utility
eschnett has joined #ste||ar
<github> [hpx] hkaiser force-pushed fixing_2916 from 90dc03d to e29ace6: https://git.io/vdnfL
<github> hpx/fixing_2916 e29ace6 Hartmut Kaiser: Adding test to trigger problem reported in #2916...
aserio has joined #ste||ar
aserio has quit [Ping timeout: 248 seconds]
aserio has joined #ste||ar
<hkaiser> heller: I found the issue, stupid me
<heller> hkaiser: what's wrong?
<jbjnr> meaning - what was the bug ...
<hkaiser> heller: the for_loop
<github> [hpx] hkaiser created fixing_for_loop (+1 new commit): https://git.io/vd0pR
<github> hpx/fixing_for_loop a0b484e Hartmut Kaiser: Making sure parallel::for_loop is executed in parallel if requested
<hkaiser> ^^
<github> [hpx] hkaiser opened pull request #2933: Making sure parallel::for_loop is executed in parallel if requested (master...fixing_for_loop) https://git.io/vd0po
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
<github> [hpx] hkaiser closed pull request #2933: Making sure parallel::for_loop is executed in parallel if requested (master...fixing_for_loop) https://git.io/vd0po
<github> [hpx] hkaiser deleted fixing_for_loop at a0b484e: https://git.io/vdEvF
rod_t has joined #ste||ar
EverYoung has joined #ste||ar
eschnett has quit [Quit: eschnett]
<github> [hpx] K-ballo opened pull request #2934: Remove the need for "complete" SFINAE checks (master...sfinae-expression-complete) https://git.io/vdET1
hkaiser has quit [Quit: bye]
parsa has quit [Quit: Zzzzzzzzzzzz]
eschnett has joined #ste||ar
hkaiser has joined #ste||ar
pree has joined #ste||ar
pree has quit [Ping timeout: 240 seconds]
hkaiser has quit [Read error: Connection reset by peer]
rod_t has left #ste||ar [#ste||ar]
aserio has quit [Ping timeout: 246 seconds]
hkaiser has joined #ste||ar
<zao> Some tests take an awful lot of time to run - test.unit.host_.block_allocator took 56.31s last run, parallel_sort took 76s.
<zao> 789 seconds for test suite, that's over 12 minutes of just running :(
<zao> (whole build and test run took 156 min, but that's still like 10% of a build)
<zao> (take figures with fistful of salt, as I hit swap something bad :D )
<hkaiser> zao: your machines would just run idle without HPX
<zao> hkaiser: I'm weighing if it's worth the money to fit it with 16G more RAM, just to be able to say I can compile HPX without swapping :P
<hkaiser> :-P
aserio has joined #ste||ar
rod_t has joined #ste||ar
<jbjnr> hkaiser: thanks for fixing the for_loop - got the patch just in time to rerun our demo and show a better task plot!
mbremer has joined #ste||ar
<zao> 14 tests failed out of 550
<zao> :D
<zao> tests.unit.components.distributed.tcp.migrate_component and tests.unit.resource.throttle went timeout this time, yay.
<heller> Ohoh
<heller> this is probably related to the shutdown bug
<heller> zao: release or debug build?
<zao> RelWithDebInfo
<zao> Difference being that it has whatever for commits came in after lunch.
<zao> Could of course just be luck.
<heller> Luck, of course
<jbjnr> hkaiser: heller what namespace do I put my numa_allocator into - answer must not include hpx:compute - I like the idea of hpx::memory::host ....
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 255 seconds]
EverYoung has joined #ste||ar
EverYoun_ has quit [Ping timeout: 240 seconds]
<hkaiser> jbjnr: hpx::util?
<hkaiser> for simplicity's sake?
parsa has joined #ste||ar
eschnett has quit [Quit: eschnett]
<heller> hkaiser: im against util
<heller> What's wrong with target? I think it would make sense to consolidate the stuff we have in there with the resource partitioner, opening another namespace sounds wrong
EverYoung has quit [Ping timeout: 258 seconds]
<jbjnr> you mean hpx::target::host::numa_allocator - tht's ok. But I object to hpx::compute::xxxx
parsa has quit [Client Quit]
<hkaiser> jbjnr: just trying to avoid overly cluttering the namespaces
eschnett has joined #ste||ar
<aserio> hkaiser: are you working on a fix to Phylanx?
<hkaiser> is it broken?
<aserio> CircleCI is broken
<hkaiser> k
<hkaiser> it's that hpx problem again
<aserio> I was about to rebase and then I thought twice :D
<hkaiser> no, go ahead, it's an unrelated hpx problem
<aserio> hkaiser: but the tests are failing
<aserio> not the build
<hkaiser> aserio: the tests are failing because of an hpx problem
<aserio> hkaiser: ah, the build is using top of master
<aserio> ok
<heller> jbjnr: i'm fine with the hpx::host namespace
<heller> I'd even be ok with having it in hpx:: directly
hkaiser has quit [Quit: bye]
eschnett has quit [Quit: eschnett]
<jbjnr> heller: I had to add a commit to your fixing_thread_executors branch - can I rebase your branch onto current master and force push it?
jaafar has joined #ste||ar
hkaiser has joined #ste||ar
aserio has quit [Quit: aserio]
diehlpk_work has quit [Quit: Leaving]
rod_t has left #ste||ar [#ste||ar]
eschnett has joined #ste||ar
rod_t has joined #ste||ar