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 [Ping timeout: 252 seconds]
hkaiser has quit [Quit: bye]
EverYoung has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoung has quit [Ping timeout: 265 seconds]
EverYoun_ has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
daissgr1 has quit [Ping timeout: 240 seconds]
hkaiser has joined #ste||ar
EverYoung has quit [Ping timeout: 245 seconds]
mcopik has joined #ste||ar
hkaiser has quit [Quit: bye]
EverYoung has joined #ste||ar
EverYoun_ has joined #ste||ar
EverYoung has quit [Read error: Connection reset by peer]
EverYoun_ has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
gedaj has quit [Remote host closed the connection]
Anushi1998 has joined #ste||ar
<Anushi1998>
Why this throws an error "g++ -I /home/anushi/boost/ -I home/anushi/hpx_build/ fibonacci.cpp"
<Anushi1998>
error: ‘unique_future’ is not a member of ‘hpx'
EverYoung has quit [Ping timeout: 265 seconds]
<K-ballo>
unique_future? that was from my time
<K-ballo>
I took it out in late 2013, when we aligned our futures with the concurrency ts
<K-ballo>
Anushi1998: which version of HPX are you using?
gedaj has joined #ste||ar
<Anushi1998>
K-ballo: I have forked it in 2018 only
<K-ballo>
alright, let's figure out where that `unique_future` comes from then
<K-ballo>
Anushi1998: would you file an issue please?
<K-ballo>
the curious part is that the embedded snippets below, which shuold also generated based on that file, display the correct content
<Anushi1998>
Yes you are right, my cloned repo has this updated file without unique_future
<Anushi1998>
K-ballo: Sure
<K-ballo>
I took unique_future out back in 2013 :) so the docs have been broken at least that long :/
<Anushi1998>
K-ballo: I just copied this new code to my file and again tried the same
<Anushi1998>
but it returned an error: ld returned 1 exit status
<K-ballo>
understandably, you don't link to hpx in your command line
<K-ballo>
btw if you built hpx already then you can build this example with `make example.quickstart.fibonacci` or si
<K-ballo>
`make examples.quickstart.fibonacci`, it's there in the tutorial as well
<K-ballo>
if you want to build manually, from the command line, then you need to take care of linking hpx, linking boost, linking hwloc, defining some compilation options, etc
<K-ballo>
all the stuff that cmake does
<K-ballo>
we are supposed to have a command line build example, one sec
<zao>
There's a bunch of ways to publish to that URL, but we're using the branch method.
<Anushi1998>
Okay
<nikunj>
Can someone explain me what the following error means: "what(): null thread id encountered (is this executed on a HPX-thread?): HPX(null_thread_id)"
<heller_>
jbjnr: you can absolutely play with it ... but I am pretty certain that it won't fix your problem
K-ballo has joined #ste||ar
<zao>
Would there be any point to trying the MPI parcelport on Windows to try to narrow my problems down?
<zao>
Seems like OpenMPI has given up on providing binaries long ao.
<heller_>
ask hartmut, he uses the MPI parcelport on windows
<jbjnr>
nikunj: yes. if you execute an hpx related task (part of the internals) on an OS thread, then you'll have a null thread id and get this error
<nikunj>
@jbjnr: I see, so in that case I'll have to register that function on hpx thread?
<jbjnr>
what is it you're calling that triggers the error
<jbjnr>
this sort of stuff usually only happend deep inside the scheduling/parcelport or other nasty hpx internals
<nikunj>
@jbjnr: I am trying to create a mini-poject of what I will carry out during GSoC period. Currently I'm using init_globally.cpp to initialize the hpx runtime system and trying to run factorial.cpp
<nikunj>
@jbjnr: This way I'm trying to learn what initialization sequences will I have to worry about and fix during GSoC period
<jbjnr>
aha. then I gues somehow you're not properly starting the runtime, or calling a function that should startup the runtime on a raw os thread and bypassing some init (or something like that)
<jbjnr>
I am not familiar with your project
<nikunj>
@jbjnr: Exactly.
<jbjnr>
what are you hoping to achieve that we can't already do with hpx::init etc
<nikunj>
@jbjnr: So I'm trying to replace the current hpx_main.hpp way of executing hpx runtime with another initialization mechanism
<jbjnr>
ok. I suspect you should try to get simbergm involved (he's not online - he's on a course today). He has been working on some stuff to start stop the runtime easily - this might be needed for lifting certain functions into hpx threads.
<jbjnr>
it says difficulty easy-medium - that's balls - this is hard surely. perhaps hkaiser has some other ideas ...
<zao>
:D
<nikunj>
@jbjnr: Yes I discussed with hkaiser about this project
<nikunj>
I discussed about 3 different methods to implement the runtime system. And I finally settled with hkaiser's implementation of runtime system
<jbjnr>
ok. he should be online soon, you'd best talk to him, cos I have to disappear shortly for a bit
eschnett_ has joined #ste||ar
<nikunj>
Only 1 problem though, I will have to correct initialization sequences everywhere. So I am currently making a list of functions that needs sequential correctness
<nikunj>
@jbjnr: ok, thanks for the help :)
eschnett has quit [Ping timeout: 260 seconds]
<zao>
Gaaaah, my VM guest with the rare debugger state from yesterday has keeled over in Virtualbox again :(
<jbjnr>
zao: constantly inflicting pain upon himself.
<zao>
I guess I have to do actual work instead then.
anushi has quit [Quit: Leaving]
galabc has joined #ste||ar
CaptainRubik has joined #ste||ar
anushi has joined #ste||ar
galabc has quit [Read error: Connection reset by peer]
<anushi>
How ste||ar bot commits automatically in gh-pages branch?
<anushi>
Actually the broken documentation is because stellarbot is not updating code in gh-pages,as far as I can understand
<zao>
Do we have any known problems with OpenMPI 3.0?
<diehlpk_work>
zao, Where?
<zao>
diehlpk_work: Nowhere, I'm looking at building for MSVC and need to pick a version.
<zao>
2.1 would be the safe bet, of course.
<diehlpk_work>
I was using 3.0 on linux and it waa working
<diehlpk_work>
and gcc
<heller_>
no known issues
<zao>
Ok!
<heller_>
an conforming MPI implementation should work
<heller_>
any
<heller_>
I'd be interested if the new MS MPI implementation works ;)
<zao>
Gonna chicken out and start with OpenMPI :)
jakub_golinowski has joined #ste||ar
<heller_>
because you are in for the fun :P
<zao>
Normally I'd go bananas, but I want to get some directed experiments going to dig into the shutdown issues.
<heller_>
yeah
jaafar has joined #ste||ar
<anushi>
diehlpk_work: thanks :)
aserio has joined #ste||ar
CaptainRubik has quit [Ping timeout: 260 seconds]
sharonhsl has joined #ste||ar
sharonhsl has quit [Read error: Connection reset by peer]
sharonhsl has joined #ste||ar
akheir has joined #ste||ar
eschnett_ has quit [Quit: eschnett_]
anushi has quit [Quit: Leaving]
parsa has joined #ste||ar
daissgr has joined #ste||ar
gedaj_ has quit [Read error: Connection reset by peer]
gedaj__ has joined #ste||ar
anushi has joined #ste||ar
<diehlpk_work>
Hi sharonhsl, I just saw your e-mail
<sharonhsl>
hi there, sorry for the messed up format
daissgr has quit [Ping timeout: 256 seconds]
akheir has quit [Remote host closed the connection]
<diehlpk_work>
In the project description, the neighbor search was not included in the distributed implementation
<diehlpk_work>
The focus was more on domain decompostion and load balancing
daissgr has joined #ste||ar
<sharonhsl>
so you mean that the original code to be extended is something else?
gedaj__ has quit [Remote host closed the connection]
gedaj_ has joined #ste||ar
<diehlpk_work>
Yes, you have the spatial points (1000) and you have to compute the force
galabc has joined #ste||ar
<diehlpk_work>
So when using two nodes you have to split them 500 and 500 on each node
<diehlpk_work>
But some points interact between nodes
<diehlpk_work>
So you have to exchange these information
<K-ballo>
aserio: where's hkaiser?
<diehlpk_work>
K-ballo, On a conference
<aserio>
yep
<diehlpk_work>
sharonhsl, The code with the neighbor search is not the code to be extended
<sharonhsl>
aha I get it, so I do not have to concern about finding neighbors , but only yo extend the part to compute the force
<anushi>
diehlpk_work: just being curious, which workshop hkaiser is attending is it something related to c++?
<diehlpk_work>
sharonhsl, Forfinding neighbors you can use a simple n^2 algorithm
anushi has quit [Quit: Leaving]
<sharonhsl>
that's nice to know
Anushi1998 is now known as anushi
diehlpk has joined #ste||ar
aserio has quit [Ping timeout: 245 seconds]
<sharonhsl>
I'm quite overwhelmed by the rigorous math in your paper
<sharonhsl>
from what I understand, a point affects it's neighbors within a radius
aserio has joined #ste||ar
<sharonhsl>
and there are several formulas aka different types of interactions to govern the displacement of a point?
<heller_>
diehlpk: nbody?
aserio has quit [Remote host closed the connection]
aserio has joined #ste||ar
sharonhsl has left #ste||ar [#ste||ar]
shalam has joined #ste||ar
shalam has left #ste||ar [#ste||ar]
sharonhsl_ has joined #ste||ar
sharonhsl_ has quit [Client Quit]
sharonhsl_ has joined #ste||ar
galabc has quit [Ping timeout: 268 seconds]
sharonhsl_ has quit [Client Quit]
parsa has quit [Quit: Zzzzzzzzzzzz]
parsa has joined #ste||ar
parsa has quit [Client Quit]
galabc has joined #ste||ar
galabc has quit [Client Quit]
parsa has joined #ste||ar
aserio has quit [Ping timeout: 260 seconds]
daissgr has quit [Quit: WeeChat 1.4]
daissgr has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Remote host closed the connection]
EverYoung has joined #ste||ar
aserio has joined #ste||ar
tianyi has joined #ste||ar
eschnett has joined #ste||ar
<diehlpk>
heller_, Not n-body, more like molecular dynamics
<heller_>
I see
<heller_>
So short range forces?
<diehlpk>
Yes
<heller_>
An octree with a space filling curve should give nice results there, no?
<diehlpk>
Yes
diehlpk has quit [Remote host closed the connection]
<diehlpk_work>
heller_, or whatever a student suggests
<heller_>
Yes
gedaj__ has joined #ste||ar
gedaj_ has quit [Read error: Connection reset by peer]
CaptainRubik has joined #ste||ar
aserio has quit [Ping timeout: 240 seconds]
nikunj has quit [Ping timeout: 260 seconds]
gedaj__ has quit [Read error: Connection reset by peer]
gedaj_ has joined #ste||ar
eschnett has quit [Ping timeout: 264 seconds]
eschnett has joined #ste||ar
CaptainRubik has quit [Ping timeout: 260 seconds]
aserio has joined #ste||ar
bibek has quit [Remote host closed the connection]
RostamLog has joined #ste||ar
tianyi has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
akheir has joined #ste||ar
eschnett_ has joined #ste||ar
eschnett has quit [Ping timeout: 264 seconds]
<diehlpk_work>
Is rostam down?
<diehlpk_work>
akheir, ?
hkaiser has joined #ste||ar
<akheir>
diehlpk_work: just noticed that. login node is down. I'm working on it
<diehlpk_work>
Thanks
eschnett_ has quit [Ping timeout: 264 seconds]
gablab has joined #ste||ar
nikunj has quit [Quit: Page closed]
<gablab>
hkaiser could I ask you a question about the feature extraction done to generate de data for hpxML?
<gablab>
the data**
<hkaiser>
ask away
<gablab>
from my understanding matrix multiplicaiton loops were used to generate the input features
<hkaiser>
ok
<gablab>
and then different policies were applied on these loops
<gablab>
and then the policy with minimal execution time is chosen as target value
<hkaiser>
yes
<gablab>
But
gedaj__ has joined #ste||ar
<gablab>
In a matrix multiplication operation there are several loops
<gablab>
do you try every policy on every loop?
<hkaiser>
uhh
<hkaiser>
I think she was working with the outermost loop only - but I might be wrong
<gablab>
mmh well I realise that one of the features is loop level
<gablab>
maybe she tried different loops for different tests
gedaj_ has quit [Ping timeout: 260 seconds]
<gablab>
Her loop level values are 1,2,3 so she probably chose different loop levels for different tests
<gablab>
I will ask her on e-mail
<hkaiser>
ok
gedaj__ has quit [Read error: Connection reset by peer]
gedaj_ has joined #ste||ar
<gablab>
hkaiser thank you again for your time
<hkaiser>
:D
gedaj__ has joined #ste||ar
gedaj_ has quit [Read error: Connection reset by peer]
<akheir>
diehlpk_work: rostam is up again
<diehlpk_work>
akheir, Thanks
<akheir>
diehlpk_work: no problem
nikunj has joined #ste||ar
bibek has quit [Quit: Leaving]
<jbjnr>
I've got tests that han on termination. did someone break master or is my stuff knackered
akheir has quit [Remote host closed the connection]
<hkaiser>
jbjnr: that's a question for simbergm
<aserio>
hkaiser: I like that my cousin's are all over the world :p
<aserio>
hkaiser: btw, I am looking at the checkpoint stuff and I am confused. Do the trait extensions that you worked on allow me to overload the serialization << operator?
Smasher has joined #ste||ar
gablab has quit [Quit: Leaving]
mcopik has quit [Ping timeout: 268 seconds]
<hkaiser>
aserio: did I work on this?
<hkaiser>
aserio: I looked at tis stuff during the flight, but couldn't come to a decision ;)
Smasher has quit [Read error: Connection reset by peer]
mcopik has joined #ste||ar
hkaiser has quit [Quit: bye]
aserio has quit [Quit: aserio]
aserio has joined #ste||ar
mcopik has quit [Ping timeout: 245 seconds]
aserio has quit [Quit: aserio]
gedaj__ has quit [Remote host closed the connection]