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/
zbyerly_ has joined #ste||ar
josef__k has joined #ste||ar
josef_k has quit [Ping timeout: 240 seconds]
denis_blank has quit [Quit: denis_blank]
diehlpk has joined #ste||ar
josef__k has quit [Ping timeout: 240 seconds]
hkaiser has quit [Quit: bye]
josef__k has joined #ste||ar
diehlpk has quit [Ping timeout: 246 seconds]
ajaivgeorge_ has joined #ste||ar
ajaivgeorge has quit [Read error: Connection reset by peer]
K-ballo has quit [Quit: K-ballo]
josef__k has quit []
wash is now known as washplanez
ajaivgeorge_ has quit [Ping timeout: 246 seconds]
pree has joined #ste||ar
pree has quit [Read error: Connection reset by peer]
pree has joined #ste||ar
bikineev has joined #ste||ar
Matombo has joined #ste||ar
bikineev has quit [Remote host closed the connection]
bikineev has joined #ste||ar
pree has quit [Ping timeout: 246 seconds]
david_pf_ has joined #ste||ar
pree has joined #ste||ar
bikineev has quit [Remote host closed the connection]
<pree>
hkaiser : Hi Did you see the gist yesterday i have sent to you ? Please tell whether it is correct ? or else i have to reimplement it and want to write tests
<hkaiser>
pree: I saw it but have not closely looked yet
<hkaiser>
still trying to understand by reading up on this in the chapel docs
<pree>
okay. I hope that i explained well
<pree>
chapel implemented differently.Because of in-built domains support
<pree>
I have also mailed them about this
<pree>
Because of domains, the notion has become easy for implementing operations such as slicing etc
<hkaiser>
ok
<hkaiser>
what is a domain in chapel?
<pree>
:)
<pree>
An index
Matombo has quit [Ping timeout: 255 seconds]
<hkaiser>
could you elaborate, pls?
<pree>
index set which specify the size and shape of arrays .
<hkaiser>
is that something we should implement?
<pree>
I actually thought it. But it leads to reimplement many things in hpx i hope so
<hkaiser>
what would you reimplement?
<pree>
wait a sec
<pree>
domain is a type in chapel. It's a first class index set which specifies the shape and size of arrays and also support iterations over arrays.
<pree>
It's supports data parallelism in chapel
<pree>
It's pure chapel language feature
hkaiser has quit [Read error: Connection reset by peer]
Matombo has joined #ste||ar
<pree>
Domain maps have ownership of domain indices and array elements.. Layouting a domain maps means having it in a locality, and distributing a domain map means distributind it based on policies like cyclic-block-distribution over a set of localities.
<pree>
It hides a notion of directly dealing with underlying array.
<pree>
we can have one or more arrays over a domain
bikineev has quit [Remote host closed the connection]
bikineev has joined #ste||ar
hkaiser has joined #ste||ar
hkaiser has quit [Read error: Connection reset by peer]
K-ballo has joined #ste||ar
denis_blank has joined #ste||ar
ajaivgeorge has joined #ste||ar
bikineev has quit [Remote host closed the connection]
bikineev has joined #ste||ar
hkaiser has joined #ste||ar
<pree>
hkaiser did you receive my last 2 messages ? I sent it after "connection reset by peer"
<hkaiser>
I saw them on the logs, yes
<pree>
okay.
<pree>
what about the gist ?
<hkaiser>
what about it?
<pree>
Did you understand the idea of cyclic_block_distribution_policy ?
<hkaiser>
pree: pls, give me a day or so to figure things out
hkaiser has quit [Read error: Connection reset by peer]
bikineev has quit [Remote host closed the connection]
hkaiser has joined #ste||ar
<pree>
okay !
<pree>
For now. I will write tests , for what I have implemented
ajaivgeorge_ has joined #ste||ar
ajaivgeorge has quit [Ping timeout: 240 seconds]
bikineev has joined #ste||ar
<github>
[hpx] K-ballo force-pushed throw_with_info from 8faf4e5 to 2c91a80: https://git.io/vHKTJ
<hkaiser>
any idea how such a thing could be implemented?
<K-ballo>
wow, as part of the proposal's interface? :|
<K-ballo>
void_t<decltype(op(args...))>, or within a guarded namespace introduce a fallback fallback_t op(...); and compare decltype(op(args...)) against fallback_t ?
denis_blank has joined #ste||ar
<hkaiser>
the args are not specified as part of the traits interface
<K-ballo>
oh, that's even dumber
<hkaiser>
really just op(exec, ...) has to be valid
<hkaiser>
right
<K-ballo>
ok, introduce the fallback, take its address, if there are more overloads it will fail to be deduced
<hkaiser>
kindof has_xxx_member, just for free functions
<K-ballo>
oh, but `exec`
<K-ballo>
nah, that makes no sense
<hkaiser>
the idea is that an executor should be have the possiblility to expose its functionality either using exec.op(...) or alternatively using op)exec,...=
<K-ballo>
what does "of single cardinality" mean?
<hkaiser>
op(exec, ...) that is
<hkaiser>
cardinality is either just one invocation or bulk invocation
<K-ballo>
I can't imagine a) how this is actually useful b) how this is implementable, even with holes in it
<K-ballo>
the traits, I mean
<hkaiser>
right
<hkaiser>
same here
<K-ballo>
doesn't seem to be a sound design, I can only imagine is trying to reflect a bogus implementation
<hkaiser>
well, I can see the usefulness of giving implementors of executors the choice whether to expose th eexecutors functionality via member functions or free functions
<hkaiser>
kindof general call syntax for executors
<K-ballo>
that doesn't make the traits useful
<hkaiser>
well, sure
<K-ballo>
there's no bag of has_member traits for allocators, for instance
<hkaiser>
just wanted to make sure I don't miss anything - I don't believe this is implementable as specified
<zao>
Simply extend the compiler enough :)
<hkaiser>
the whole proposal is about a badly designed interface
<hkaiser>
I tried to convince them of alternative designs exposing the same functionality - well, who am I ...
<K-ballo>
even if it were implementable via approximation (using fixed arguments, archetypes, or something of the sort) the lookup would be incorrect
<github>
hpx/P0443R2 353e4aa Hartmut Kaiser: Refining C++ language mode detection for MSVC...
<github>
hpx/P0443R2 845495e Hartmut Kaiser: Renaming `<op>_bulk_execute` to `bulk_<op>_execute`
david_pf_ has quit [Quit: david_pf_]
Matombo has quit [Remote host closed the connection]
<denis_blank>
Is there a trait in the codebase which checks whether a callable object is callable with the given arguments? Actually I thought that's what is_callable is for, however it behaves strange (and doesn't even compile)...