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 quit [Quit: Zzzzzzzzzzzz]
daissgr1 has quit [Quit: WeeChat 1.4]
tjtn has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
tjtn has joined #ste||ar
tjtn has quit [Client Quit]
thundergroudon[m has joined #ste||ar
jaafar has joined #ste||ar
autrilla has joined #ste||ar
taeguk[m] has joined #ste||ar
<parsa[w]>
hkaiser: what's the difference between hpx::performance::counters::server::raw_values_counters and hpx::performance::counters::server::raw_counters?
<parsa[w]>
raw_counter*
jaafar has quit [Ping timeout: 255 seconds]
tjtn has joined #ste||ar
tjtn has quit [Client Quit]
<hkaiser>
parsa[w]: good question
<hkaiser>
parsa[w]: one returns one scalar value, the other one returns a vector of values (like the histogram counter)
K-ballo has quit [Quit: K-ballo]
parsa has joined #ste||ar
jaafar has joined #ste||ar
hkaiser has quit [Quit: bye]
diehlpk has quit [Ping timeout: 260 seconds]
jaafar has quit [Ping timeout: 255 seconds]
parsa has quit [Quit: Zzzzzzzzzzzz]
nanashi55 has quit [Ping timeout: 240 seconds]
nanashi55 has joined #ste||ar
EverYoung has quit [Ping timeout: 240 seconds]
EverYoung has joined #ste||ar
zbyerly has quit [Ping timeout: 246 seconds]
zbyerly has joined #ste||ar
thundergroudon[m has quit [Ping timeout: 250 seconds]
jbjnr has quit [Remote host closed the connection]
david_pfander has quit [Remote host closed the connection]
david_pfander has joined #ste||ar
jbjnr has joined #ste||ar
mcopik has quit [Ping timeout: 240 seconds]
jbjnr has quit [Remote host closed the connection]
jbjnr has joined #ste||ar
david_pfander has quit [Remote host closed the connection]
jbjnr has quit [Remote host closed the connection]
david_pfander has joined #ste||ar
jbjnr has joined #ste||ar
thundergroudon[m has joined #ste||ar
autrilla has joined #ste||ar
taeguk[m] has joined #ste||ar
<heller>
simbergm: closing in...
<heller>
I'm looking at the after 588 failures today
<heller>
After the throttle fixes
<simbergm>
heller: nice :) should I interpret that as you will look at the throttle fixes?
<simbergm>
it looked like there were 6 or 7 tests with the same error as 588 so hopefully they'll all be fixed in one go
<simbergm>
I think probably 3039 would be good to look at first if you have the time
Smasher has quit [Remote host closed the connection]
Smasher has joined #ste||ar
<heller>
simbergm: alright, will have a look
<simbergm>
heller: thanks
<heller>
simbergm: yes, allscale is waiting on the throttle fixes ;)
<heller>
simbergm: and yes, there are a few tests exposing the same problem
<simbergm>
heller: okay, cool :) do they only need the thread suspension or full runtime suspension?
<simbergm>
note to self, I didn't clean up the throttle example or throttling scheduler yet, but that will come when once you're happy with the changes...
<heller>
simbergm: only thread suspension
<simbergm>
so it's really the throttling use case (measuring power usage and suspending/resuming based on that)?
<heller>
simbergm: #3039 has lots of changes intertwined :/
<heller>
simbergm: yes, more or less
<heller>
simbergm: the throttling test... doesn't it miss the elasticity flag?
<heller>
or do I miss something?
<simbergm>
heller: well, I do set it in hpx_main as I think it's not possible to set it via the rp at the moment, but this is definitely something that could be nice to change
<github>
hpx/master 141890e Thomas Heller: Merge pull request #3039 from STEllAR-GROUP/fixing_3027...
<simbergm>
heller: I'm confident it doesn't break any new tests now, not 100% confident I've covered everything with the added tests but I hope we'll see that soon enough
<simbergm>
let me first try to move enabling the elasticity to the setup though
<simbergm>
jbjnr: did you already change whatever you found was causing the 30 second timeout in pycicle?
<heller>
simbergm: ok
<heller>
simbergm: I'd to merge it once the executor PR merge cycled ;)
<heller>
simbergm: I am fine with having that in another PR ;)
<simbergm>
how urgently would you want it merged? :)
<simbergm>
ok :)
<heller>
pretty urgently :P
Smasher has quit [Remote host closed the connection]
<simbergm>
your call, can't you work with the branch...?
Smasher has joined #ste||ar
<heller>
simbergm: I can, I guess
<heller>
my collaborators might get confused...
<simbergm>
I'll do it in a separate PR then, I should remove the set_scheduler_mode methods as well then so that it can't be changed at runtime
<simbergm>
understandable, it should do the basics already
<simbergm>
besides, it's good to get it tested with real code
<heller>
yeah
<heller>
so far, the tests I did where positive
<heller>
no hangups!
<simbergm>
yeah, I'm quite sure the current tests pass well because it's been passing with pycicle on every merged PR this week
<simbergm>
if there are problems with certain configurations we'll see it once buildbot gets its hands on it
eschnett has quit [Quit: eschnett]
<simbergm>
heller: one thing, I left it so that you can do either remove_processing_unit or suspend_processing_unit, but I removed the tests with remove_processing_unit because I thought one should only use suspend/resume...
<jbjnr>
simbergm: sorry : I updated the 30 second timeout locally, but didn't do it on greina. just restarted pycicle there and forced a rebuild of all prs
<jbjnr>
sorry it took so long. holiday here, was out for a few hours.
<jbjnr>
back to work now :)
<simbergm>
jbjnr: no worries, just a friendly reminder :)
<simbergm>
ah, well you can also enjoy a day off maybe... ;)
<jbjnr>
no such thing
<jbjnr>
only days when I work a bit less
<jbjnr>
does anyone know what hk was on about when he said that some tests that are supposed to fail are being marked as fails
<jbjnr>
there are compilation tests that should fail, but I disabled them anywa as they annoy me
<jbjnr>
perhaps it was those ...
<heller>
should be good
<heller>
the compilation tests that should fail, should only be compiled when running ctest
<07IAB4WI0>
[hpx] sithhell pushed 75 new commits to master: https://git.io/vb8BH
<07IAB4WI0>
hpx/master aa681ca Mikael Simberg: Unrevert reverted throttle changes
<07IAB4WI0>
hpx/master 14f0572 Mikael Simberg: Fix selection of thread number in create_thread
<07IAB4WI0>
hpx/master 831f93b Mikael Simberg: Make throttle test work with any (>= 2) number of threads
<jbjnr>
all this pushing to master and merging is really giving pycicle a good work out. I've disabled daint today as I'm trying to get the clang build running and the scripts are broken, but there's a serious amount of results building on the dashboards the last few days. We need a cdash upgrade because you can hide smoe stuff and reduce the info (might please heller)
<heller>
jbjnr: yay!
<heller>
ok ... the problem with clang seems to be that the order of the global dtors isn't fixed :/
<heller>
and the mutex used for unloading the dlls is what's causing problems :/
<heller>
fixed!
aserio has joined #ste||ar
EverYoung has joined #ste||ar
aserio has quit [Quit: aserio]
jakemp has quit [Ping timeout: 248 seconds]
mcopik has joined #ste||ar
aserio has joined #ste||ar
hkaiser has joined #ste||ar
david_pfander has quit [Ping timeout: 240 seconds]
aserio has quit [Ping timeout: 246 seconds]
<github>
[hpx] hkaiser created fixing_exceptions_seq (+1 new commit): https://git.io/vb4J1
<github>
hpx/fixing_exceptions_seq 78ed5c7 Hartmut Kaiser: Fix exception handling for execution::seq
aserio has quit [Read error: Connection reset by peer]
aserio has joined #ste||ar
aserio has quit [Ping timeout: 260 seconds]
<github>
[hpx] K-ballo created workaround-3058 (+1 new commit): https://git.io/vb44V
<github>
hpx/workaround-3058 3d36f0b Agustin K-ballo Berge: Workaround for minimal_timed_async_executor_test compilation failures, attempts to copy a deferred call (in unevaluated context)