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/
<github>
[hpx] hkaiser deleted annotate_function at a93b210: https://git.io/v7I06
<github>
[hpx] hkaiser opened pull request #2781: Check for and optionally use [[deprecated]] attribute (master...deprecated_attribute) https://git.io/v7I0i
bikineev has quit [Remote host closed the connection]
ajaivgeorge_ has joined #ste||ar
pree has joined #ste||ar
<zao>
Smells like something of use for debugging or traces.
bikineev has joined #ste||ar
bikineev has quit [Ping timeout: 255 seconds]
diehlpk has joined #ste||ar
K-ballo has joined #ste||ar
diehlpk has quit [Ping timeout: 248 seconds]
mcopik has quit [Ping timeout: 268 seconds]
bikineev has joined #ste||ar
mars0000 has joined #ste||ar
parsa has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 246 seconds]
<github>
[hpx] K-ballo force-pushed iota_range from aba681f to 1860039: https://git.io/v7Ir3
<hkaiser>
K-ballo: the list became much shorter recently...
<K-ballo>
revisions reflect that
<K-ballo>
(the base revision is bogus, I forgot to include src/ and plugin/)
<K-ballo>
-50 preprocessor includes
<K-ballo>
-10 date_time includes
Matombo has joined #ste||ar
<hkaiser>
nice
<hkaiser>
boost/assign/std/vector.hpp can probably go, boost/exception/exception.hpp as well
<hkaiser>
and some others
<K-ballo>
there's a lot of low hanging fruit
<K-ballo>
with equally low impact
<jbjnr>
hkaiser: yes I don't care about hello_world on my branch - only making the matrix stuff run fast. That's why the commits that remove the other thread pools are flagged as WIP:stuff
<jbjnr>
hkaiser: feel free to take a branch point just before all the WIP commits (I rebased it all to make them on the tip so they can be removed easily)
<hkaiser>
jbjnr: ok, I created a PR against your branch, adding config options for this
<hkaiser>
and some other changes
<jbjnr>
hkaiser: ok great. I'm not sure we're ready for a PR yet, but you're the boss, so if it looks usable to you ...
<hkaiser>
no, I created a PR against your branch
<jbjnr>
ah, yes. sorry misunderstood
<jbjnr>
you can just rewrite any commits of mine you dislike, I will handle any conflicts
<hkaiser>
jbjnr: as said, I'd like to start using this to integrate with legacy mpi codes
<jbjnr>
but PR's work too
<jbjnr>
ok
<hkaiser>
jbjnr: expect lot'sa nitpicking ;)
<jbjnr>
if you need anything we've done exlained, then a skype call might be a good idea
<hkaiser>
ok, thanks
<hkaiser>
currently doing a formal once over to get the big picture and to satisfy my ocd
<hkaiser>
jbjnr: I really like it so far, thanks for all the work
<hkaiser>
jbjnr: do you still in contact with Shoshana?
<hkaiser>
are you*
<hkaiser>
(I also can look into implementing the perf-counters, btw
<jbjnr>
she is in boston for the summer, I have not heard from here recently, but when she gets back we might hear news etc. We have an idea that a multi-thread pool paper talking about executers and use cases would be nice
<hkaiser>
k
<jbjnr>
as a result of her internship
<hkaiser>
sounds cool
<jbjnr>
perf counters - yes please. any help welcome. need to decide on how to query idle rates etc on a per pool basis.
mars0000 has quit [Quit: mars0000]
<hkaiser>
ok
<jbjnr>
reall want to put io and timer pools into the RP and remove asio if poss
<hkaiser>
tcp needs asio
<jbjnr>
(until we get the libfabric tcp layer working :) )
<hkaiser>
but yah, for the io and timer pools we can use any thread pool
<hkaiser>
jbjnr: give my poor windows machine a chance
<jbjnr>
libfabric + windows + tcp = works
<jbjnr>
(not tested though)
<hkaiser>
ahh, ok - fine by me, then ;)
mcopik has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 240 seconds]
eschnett has quit [Quit: eschnett]
diehlpk has joined #ste||ar
<jbjnr>
hkaiser: actually libfabric will never replace tcp/asio, we would need something like pmi on all platforms, which won't happen -otherwise out of band data is probably still needed - which requires tcp - I must look into this at some point and see what could really be done
<hkaiser>
jbjnr: ok
diehlpk has quit [Ping timeout: 268 seconds]
<hkaiser>
jbjnr: are there some examples showing how to use the RM?