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/
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
<wash[m]>
hkaiser, patg, aserio: small issue found while proofreading
<hkaiser>
k
hkaiser has quit [Read error: Connection reset by peer]
hkaiser has joined #ste||ar
<wash[m]>
On page 4, we state that ChaNGa and SWIFT "have demonstrated scalability up to 500K cores with a parallel efficiency of 93.4% and 100K cores with a parallel efficiency of 60%"
<wash[m]>
Im confused by that
<wash[m]>
Ohhhh
<wash[m]>
It says respectively afterwards, I understand now
hkaiser has quit [Quit: bye]
hkaiser has joined #ste||ar
<github>
[hpx] hkaiser force-pushed gentryx-master from a2c6708 to 4d7a55f: https://git.io/v71ZM
<github>
hpx/gentryx-master 4d7a55f Hartmut Kaiser: Touching up on configuration settings...
eschnett has joined #ste||ar
hkaiser has quit [Quit: bye]
vamatya has quit [Ping timeout: 240 seconds]
zbyerly_ has quit [Ping timeout: 255 seconds]
K-ballo has quit [Quit: K-ballo]
EverYoung has quit [Ping timeout: 240 seconds]
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 248 seconds]
patg has joined #ste||ar
<patg>
wash wash[m] yt?
patg has quit [Read error: Connection reset by peer]
patg has joined #ste||ar
patg is now known as Guest19663
vamatya has joined #ste||ar
Guest19663 has quit [Quit: See you later]
bikineev has joined #ste||ar
vamatya has quit [Ping timeout: 240 seconds]
taeguk has joined #ste||ar
david_pfander has joined #ste||ar
bikineev has quit [Remote host closed the connection]
Matombo has joined #ste||ar
pree has joined #ste||ar
Matombo has quit [Remote host closed the connection]
<github>
[hpx] hkaiser closed pull request #2773: Add sync_images() method to spmd_block class (master...spmd_block_with_sync_images) https://git.io/v7J2K
Kiril_ has joined #ste||ar
<Kiril_>
Hi guys. I am looking at the hpx/examples/allgather. Does anyone have any benchmark plots comparing the HPX vs MPI allgather as implemented there?
<hkaiser>
Kiril_: allgather is obsolete, don't look at it, it's to be removed soon
<hkaiser>
an we don't have any benchmarks like this
<hkaiser>
the idea is to avoid allgathers not replace them
<Kiril_>
okay -- but do you have cases you need to do e.g. a bcast in HPX, or you are saying -- don't think at all in message passing ?
<hkaiser>
sure there are cases for broadcasts, but not in terms of message passing (i.e. everbody blocks while waiting), but in the sense of distributing data
<hkaiser>
Kiril_: but yah, we're still missing some of the logocal collective operations in hpx
<Kiril_>
so in allgather we have non-blocking point-to-point, sometimes followed by wait_all -- are you saying we have a similar scheme, but without the wait_all ?
<hkaiser>
we have gather and broadcast, but no allgather
<hkaiser>
the allgather example is a (failed) attempt at creating one using hpx
<Kiril_>
can you point me to a bcast? I spent quite a while implementing my own bcast based on allgather
<hkaiser>
but it's failed because the person who wrote the example tried to 100% reproduce the mpi semantics
<Kiril_>
and I see two variants there -- one where everyone gets a data, and one where the root sets the data everywhere
<mbremer_>
@hkaiser: Thanks, so then presumably I need to pass the serialized version to the channel?
<hkaiser>
no
<hkaiser>
channel<vector<datapar>> should just work
<mbremer_>
How convenient :)
mbremer_ has quit [Quit: Page closed]
bikineev has quit [Remote host closed the connection]
aserio has quit [Ping timeout: 276 seconds]
bikineev has joined #ste||ar
<K-ballo>
hkaiser: there?
<hkaiser>
K-ballo: here
eschnett has quit [Quit: eschnett]
<K-ballo>
hkaiser: see pm
aserio has joined #ste||ar
<zbyerly>
error: ‘unwrapped’ is not a member of ‘hpx::util’
<hkaiser>
zbyerly: has been renamed
<hkaiser>
what do you use it for?
<zbyerly>
ok i will change LibGeoDecomp
<zbyerly>
it's in the HPX send & recv
<hkaiser>
either unwrap() or unwrapping(), depending on use case
aserio has quit [Client Quit]
<hkaiser>
I thought however that we left it on for backwards compatibility
<hkaiser>
denis_blank ^^
<zbyerly>
hkaiser, it looks like the dataflow simulator isn't using it
aserio has joined #ste||ar
<denis_blank>
hkaiser, zbyerly: Mh, the backwards compatibility should work, zbyerly did you rerun cmake?
<denis_blank>
It's also possible that unwrapped was only included through another header implicitly, in that case the include was changed and it would cause a build error.
<zbyerly>
denis_blank, sorry, hpx is a couple of days old
<zbyerly>
denis_blank, it does suggest "unwrapping" as an alternative
<zbyerly>
denis_blank, okay cool i will add those and check the bw compatibility
<zbyerly>
denis_blank, we don't use that code anyway
<denis_blank>
zbyerly: Ok, thanks. However I suggest you to upgrade the usage of unwrapped to unwrapping and unwrap respectively since unwrapped was deprecated: hpx::util::unwrapped(futures) -> hpx::util::unwrap(futures), hpx::util::unwrapped(idsFuture) -> hpx::util::unwrap(idsFuture), hpx::util::unwrapped(std::move(updateGroupCreationFutures)) -> hpx::util::unwrap(std::move(updateGroupCreationFutures))
akheir has quit [Remote host closed the connection]