hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar-group.org | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | This channel is logged: irclog.cct.lsu.edu | Everybody: please respond to the documentation survey: https://forms.gle/aCpNkhjSfW55isXGA
K-ballo has quit [Quit: K-ballo]
hkaiser has quit [Quit: Bye!]
<ms[m]>
Thanks K-ballo!
<rachitt_shah[m]>
Hey ms I'll be sending over a PR today with insegel, build instructions and my ideas for content rearrangement.
<rachitt_shah[m]>
Will look at the doxygen build later today by build a part of the docs.
<rachitt_shah[m]>
I was looking at the broader under-the-hood arrangement with sphinx to understand where I was wrong
<ms[m]>
You might have to check which ones actually still exist
<rachitt_shah[m]>
Yep, I'll try that, I was building the docs section by section, since building it completely didn't work
<rachitt_shah[m]>
Thanks for the heads-up!
hkaiser has joined #ste||ar
<rori[m]>
hkaiser ms just fyi, looks like the perftests report a slight degradation in performance for the C++17 PR, I might drag this a bit longer as I need to investigate it
<hkaiser>
interesting...
<rori[m]>
Yes I wasn't expecting that
<hkaiser>
rori[m]: what tests did slow down?
<rori[m]>
future_overhead_report i.e. create_thread_hierarchical_placement
<hkaiser>
is it a lot?
<rori[m]>
the report says < 5%
<hkaiser>
rori[m]: the PR doesn't even touch anything related :/
<rori[m]>
yes
K-ballo has joined #ste||ar
hkaiser has quit [Quit: Bye!]
<ms[m]>
rori: is the reference with c++14? If yes, I'd first check that it's really a regression compared to c++17 on master (we had a case before of the c++17 build being slower than 14)
<ms[m]>
I'd also check that it's repeatable and not just noise on daint
hkaiser has joined #ste||ar
<rori[m]>
ms: the reference is with C++17 and It's repeatable in the last few days at least cause it was failing the 4 last time I pushed on that PR
<hkaiser>
ms[m]: before it was our cache_line handling that was accidentally initialzing the whole cache_line without any need, that's not the case anymore
hkaiser has quit [Ping timeout: 252 seconds]
hkaiser has joined #ste||ar
<ms[m]>
rori: ok, then it's going to need some bisecting
<ms[m]>
But I'd suggest we do it in a separate pr
<rori[m]>
Yes I'm also in to do it in a separate PR 👌
<ms[m]>
hkaiser: yeah, I was wondering if this could be a similar (but different case) if the standards were different, but that seems to not be the case
<ms[m]>
rori: also, good thing we have the perftest ci now!
<hkaiser>
ms[m]: I agree - rori[m] many thanks for setting this up!
K-ballo has quit [Read error: Connection reset by peer]
K-ballo has joined #ste||ar
hkaiser has quit [Quit: Bye!]
hkaiser has joined #ste||ar
calla has joined #ste||ar
calla has quit [Remote host closed the connection]