hkaiser 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/
hkaiser has joined #ste||ar
jbjnr_ has quit [Ping timeout: 268 seconds]
jbjnr_ has joined #ste||ar
jbjnr__ has quit [Ping timeout: 268 seconds]
jbjnr__ has joined #ste||ar
jbjnr_ has quit [Ping timeout: 268 seconds]
jbjnr_ has joined #ste||ar
jbjnr__ has quit [Ping timeout: 268 seconds]
jbjnr__ has joined #ste||ar
jbjnr_ has quit [Ping timeout: 268 seconds]
jbjnr_ has joined #ste||ar
jbjnr__ has quit [Ping timeout: 268 seconds]
jbjnr__ has joined #ste||ar
jbjnr_ has quit [Ping timeout: 268 seconds]
mdiers_ has quit [Ping timeout: 268 seconds]
mdiers_ has joined #ste||ar
jbjnr_ has joined #ste||ar
jbjnr__ has quit [Ping timeout: 268 seconds]
nikunj97 has joined #ste||ar
mdiers_1 has joined #ste||ar
jbjnr__ has joined #ste||ar
jbjnr_ has quit [Ping timeout: 268 seconds]
K-ballo has quit [Quit: K-ballo]
mdiers_ has quit [*.net *.split]
nikunj1997 has quit [*.net *.split]
mdiers_1 is now known as mdiers_
hkaiser has quit [Quit: bye]
jbjnr_ has joined #ste||ar
jbjnr__ has quit [Ping timeout: 268 seconds]
jbjnr__ has joined #ste||ar
adityaRakhecha has quit [Ping timeout: 256 seconds]
<zao>
Finally I will be able to understand what HPX does ;)
<K-ballo>
"the best library I ever used that's not bad at all
ste||ar-github has joined #ste||ar
<ste||ar-github>
[hpx] msimberg opened pull request #3621: Update copyright year and other small formatting fixes (master...update-docs-copyright-year) https://github.com/STEllAR-GROUP/hpx/pull/3621
ste||ar-github has left #ste||ar [#ste||ar]
<hkaiser>
K-ballo: John has changed his mind since his initial assessment ;)
<adityaRakhecha>
While building hpx, I am getting this error : CMake Error: The source directory "/home/aditya" does not appear to contain CMakeLists.txt. Specify --help for usage, or press the help button on the CMake GUI.
<zao>
You're probably not specifying the correct directory where it should look for the source.
<zao>
There's two kinds of CMake builds, either in-tree or separate build dirs.
<zao>
HPX strongly prefers the latter one.
<adityaRakhecha>
Got it. Thank you.
<zao>
You would have a separate build directory somewhere, and go something like "cmake /path/to/hpx/directory"
<zao>
A common way to do this is that if you've got your HPX checkout at /home/aditya/hpx, you would have a build directory in /home/aditya/hpx/build, and then run "cmake .."
<zao>
Which you may have found in an example somewhere :D
<adityaRakhecha>
yes, I was specifying the wrong directory.
<adityaRakhecha>
While I build hpx, can you please suggest me any concept or topic I should know other than basic c++ and openCV
<zao>
If you are not familiar with futures, it's probably a good thing to read up on as HPX uses them quite a bit.
<adityaRakhecha>
ok
<zao>
John's video presentation of HPX linked above might be a good watch too, I haven't looked at it yet.
<adityaRakhecha>
Yes, I checked it after reading your comment.
<adityaRakhecha>
Will watch it for sure.
<adityaRakhecha>
after executing make install I am getting this : make: *** No rule to make target 'install'. Stop.
<zao>
Could you show the commands you used to build in a snippet like a github gist or a paste site?
<zao>
Also, what OS are you on?
<zao>
Note that if you're intending to install, you probably want to specify "-DCMAKE_INSTALL_PREFIX=$HOME/opt/hpx" or something, to avoid installing into system locations.
<adityaRakhecha>
I am using linux (on Ubuntu 16.04)
<mdiers_>
Hi, I had already talked with hkaiser on friday about the problem that I do not get components from a shared lib when compiling with clang. I switched from 1.2 to the trunk, which has brought no change. I get the error "the component is disabled for this locality (component_invalid [-1]): HPX (bad_request)" and the component is not listed on --hpx:list-component-types. At which point can I start to debug?
<hkaiser>
please use --hpx:debug-hpx-log to verify whether your module gets loaded
<mdiers_>
i create 2 logs, ...
<aserio>
hkaiser: will you be joining us downstairs?
<hkaiser>
uhh, sorry still working on the proposal at home, will call in in a sec
<aserio>
hkaiser: ok, see you in a second
<aserio>
heller_: ^^
chinz07[m] has quit [Ping timeout: 250 seconds]
simbergm has quit [Ping timeout: 252 seconds]
<hkaiser>
mdiers_: see above, I didn't tag you, sorry
<diehlpk_work>
adityaRakhecha, Please put it to a github repo and add a CMakeFile to build it
<adityaRakhecha>
Fine :)
<zao>
Kind of, yes.
<zao>
The idea that a future represents the eventual result of a computation, allowing you to chain operations together without having to have the values ready at the point of decision.
<zao>
The nifty thing with HPX is that it expands this idea of tasks and futures to a distributed program, running on multiple nodes.
<zao>
HPX's futures look kind of like the standard library ones, but ours are much cooler :)
<zao>
(in ways I cannot describe, as I don't know how :D)
<adityaRakhecha>
cool. I will look into it :D
<zao>
So say that you're making a parallel matrix multiplication, the result should probably be a future of a matrix.
daissgr has joined #ste||ar
<zao>
diehlpk_work: Do correct me if I'm wrong :)
<diehlpk_work>
zao, Sound as a good first introduction to our future implementation
aserio has joined #ste||ar
parsa[[w]] has quit [Read error: Connection reset by peer]
parsa[w] has joined #ste||ar
hkaiser has joined #ste||ar
david_pfander has quit [Quit: david_pfander]
bibek has quit [Quit: Konversation terminated!]
bibek has joined #ste||ar
dmarce1 has quit [Quit: Leaving.]
jaafar has quit [Ping timeout: 268 seconds]
<adityaRakhecha>
Just ran my first hpx program B)
quaz0r has quit [Ping timeout: 268 seconds]
<zao>
Did it work? :)
maxwellr96 has joined #ste||ar
quaz0r has joined #ste||ar
<adityaRakhecha>
yes. It was just a basic one I made to check everything is working fiine.
<adityaRakhecha>
For the parallel multiplication task, Can I use OpenMP library or do I have to write my own implementation ?
<zao>
I would personally do a schoolbook implementation, but whatever works for you :)
<aserio>
We suggest that you don't mix the OpenMP runtime with the HPX runtime as they will fight each other for the cores
<aserio>
?what feature were you looking for
<zao>
In general, there's not too much gain in having SIMD or threadedness inside a task, as ...
<zao>
^ that
<zao>
A typical HPX application will have enough tasks to fill all the cores already, so unless you expect to have a very monolithic bottleneck task, there's not too much point in OpenMP use.
<adityaRakhecha>
Without OpenMP I have to make my own algorithm for threads with load balancing , race condition , etc. I am not denying to do it, it's just that it would take time as I have to learn more for this.
<adityaRakhecha>
Correct me if I am wrong
<zao>
You kind of are :)
<aserio>
I am not following what you are trying to do
<aserio>
If you need parallel algorithms we have those available
<zao>
aserio: Patrick suggested a newcomer task of making a HPX program that did a parallel matrix multiplication to get used to HPX.
<aserio>
so the parallel algorithms would be the way to go
<aserio>
adityaRakhecha: how much do you know about futures?
<zao>
A possible design could be to spawn off a task for each cell of the resulting matrix, joining them together into the result as they complete.
<zao>
(not necessarily a _good_ design :D)
<adityaRakhecha>
I want to make a program for matrix multiplication using parallel for loops.
<aserio>
adityaRakhecha: you will want to look at the parallel algorithms then
<aserio>
we have a parallel for
<adityaRakhecha>
I know about STL. I will learn about futures.
<aserio>
adityaRakhecha: Futures are key in HPX
jaafar has joined #ste||ar
<adityaRakhecha>
Yes, zao told me, just give me sometime I will learn it :)
<adityaRakhecha>
aserio: Should look for parallel algorithms in hpx then. Right!
<aserio>
adityaRakhecha: our parallel algorithms also conform to the standard ones
<adityaRakhecha>
This is good then.
<adityaRakhecha>
So first I should learn about futures and then look for parallel algo.
<aserio>
sounds good
<aserio>
adityaRakhecha: you may also be interested in looking at one of the Tutorials we have given
<aserio>
you can find them under the Tutorial header here:http://stellar-group.org/publications/
<adityaRakhecha>
This is good. Thank You.
aserio1 has joined #ste||ar
aserio has quit [Ping timeout: 264 seconds]
aserio1 is now known as aserio
daissgr has quit [Ping timeout: 250 seconds]
aserio has quit [Ping timeout: 250 seconds]
aserio has joined #ste||ar
aserio has quit [Ping timeout: 250 seconds]
aserio has joined #ste||ar
<heller_>
hkaiser: aserio: huge apologies... Had to run to a family emergency. Just returned... Will take care of bibeks issue now
daissgr has joined #ste||ar
hkaiser has quit [Quit: bye]
<heller_>
bibek: alright, responded, should be good to go now
<bibek>
heller_: Thank you very much.
<bibek>
:)
<heller_>
circleci seems to have a hickup at the moment though
<heller_>
ah yes
<heller_>
bibek: we really need to fix the phylanx build system ;)