hkaiser changed the topic of #ste||ar to: The topic is '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/
<wash[m]>
Buildbot was meh :p
hkaiser has joined #ste||ar
primef has quit [Ping timeout: 260 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 260 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
K-ballo has quit [Quit: K-ballo]
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
hkaiser has quit [Quit: bye]
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 260 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 248 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 260 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 248 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 260 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
nikunj has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
primef has quit [Ping timeout: 272 seconds]
nikunj has quit [Ping timeout: 260 seconds]
nikunj has joined #ste||ar
nikunj97 has joined #ste||ar
nikunj has quit [Ping timeout: 268 seconds]
primef has joined #ste||ar
nikunj97 has quit [Ping timeout: 240 seconds]
nikunj has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
nikunj97 has joined #ste||ar
nikunj has quit [Ping timeout: 260 seconds]
nikunj97 has quit [Read error: Connection reset by peer]
<mdiers_>
simbergm:
primef has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
<simbergm>
mdiers_: ?
<primef>
simbergm: Good morning, do you by chance know, where the tests for algorithms are located? I found them in a history commit but they seem to be gone. They were under tests/unit/parallel/algorithms
<simbergm>
libs/algorithms/tests/unit :)
<simbergm>
they were moved into a module
<simbergm>
primef: ^
<primef>
Thank you! Saved me from searching the whole repo :)
<primef>
Strange placement though. What's the difference between the ones in root/tests and the module ones?
<simbergm>
primef: they'll all eventually be moved to libs/*/tests
<simbergm>
the root tests directory just contains things that haven't moved into modules yet
<primef>
alright, thanks.
<zao>
You people need to start having flaky tests again, would be fun to get around writing that soak test bot.
<simbergm>
zao: don't tempt us... they're flaky enough as they are
<simbergm>
you can remind that someone *cough jbjnr* should look at the cross_pool_injection test
<simbergm>
jbjnr: I'm only half joking, I know you're busy with a million things...
<primef>
simbergm: sorry for bothering you again, any hint on speeding up build time? already disabled examples
<mdiers_>
simbergm: here
primef has quit [Ping timeout: 272 seconds]
primef has joined #ste||ar
<Yorlik>
simbergm: I think that survey is pretty well thought out.
K-ballo has joined #ste||ar
primef has quit [Ping timeout: 246 seconds]
<simbergm>
mdiers_: I just pinged you because you pinged me ;) but I guess that was a mistake...
<simbergm>
Yorlik: thanks for the feedback :)
<Yorlik>
:)
<simbergm>
Kor de Jong: seems like you figured out what you wanted, just check if you're aware of `--hpx:bind=balanced/numabalanced`, `--hpx:threads=cores`, and `--hpx:use-process-mask`? manually giving the binding specs is usually not necessary
<mdiers_>
simbergm: ;-) The survey looks good
hkaiser has joined #ste||ar
mdiers_ has quit [Remote host closed the connection]
<simbergm>
but discoverability of cdash features is horrible
<simbergm>
Kor de Jong: the default `balanced` should do what you want, i.e. start by filling one thread per core and only if you have too many threads will it start putting multiple threads per core (i.e. use hyperthreads)
<simbergm>
if you specify the number of cores and tasks to slurm `--hpx:use-process-mask` will just use the binding that slurm has set for the process
<simbergm>
i.e. behave roughly like openmp would in such situations
<simbergm>
it's not the default because it's kind of new, and there are cases when you don't have the mask set
<hkaiser>
simbergm: so part of the conversation is not mirrored here anymore - anything we can about that?
<simbergm>
hkaiser: gtg, but can you clarify? what is here?
<hkaiser>
here == IRC
<hkaiser>
there == matrix
<hkaiser>
gtg too
hkaiser has quit [Quit: bye]
nikunj has joined #ste||ar
<nikunj>
simbergm, yt?
primef has joined #ste||ar
hkaiser has joined #ste||ar
ct-clmsn has quit [Quit: Leaving]
hkaiser has quit [Ping timeout: 272 seconds]
hkaiser has joined #ste||ar
<heller2>
Looks like the messages from Kor de Jong are not forwarded to IRC
<hkaiser>
heller2: right, that's what I meant
primef has quit [Ping timeout: 246 seconds]
<heller2>
Strange
<heller2>
Kor de Jong: ping
<heller2>
Yeah, that's the idea indeed
<heller2>
For some reason, it's only you that's not being forwarded, and I have no idea why
<hkaiser>
heller2: I'm special ;-)
<heller2>
hkaiser: following up on our CI discussion... Have we looked into GitHub actions?
<hkaiser>
would be an option as well
<heller2>
It seems that it's almost exactly doing what we want
<hkaiser>
I think simbergm had a quick look at some point
<heller2>
They have osx, windows and Linux and the ability to run on custom hardware, together with a Travis/circle like configuration
<hkaiser>
cool
<heller2>
So it's what I want ;)
<heller2>
I'll look into it...
<zao>
We’ve started adopting GH actions for EasyBuild. A bit different but kind of works.
primef has joined #ste||ar
primef has quit [Ping timeout: 245 seconds]
primef has joined #ste||ar
<primef>
hkaiser: Hello, are you there?
hkaiser has quit [Ping timeout: 260 seconds]
<simbergm>
hkaiser, heller, very brief look
<simbergm>
My patience with these things is very short...
<simbergm>
Looks cool though
hkaiser has joined #ste||ar
<primef>
hkaiser: hello, just to let you know, yesterday I started looking into #3838. It might take me a bit to debug the application and hopefully discover something.
<hkaiser>
primef: cool, thanks!
<hkaiser>
primef: I sent your t-shirt on the way this morning, btw
<primef>
hkaiser: oh amazing! Thank you so much :) I'm really hoping it will arrive by thursday, so I can wear it at university during the presentation of my project :)
<zao>
This weekend is FOSDEM - anyone going by chance?
<diehlpk_work>
hkaiser, Build went from 25% to 56%, however there is an issue with Fedora's build system
<primef>
Does someone have an insight about the usage of "set(target_PARAMETERS THREADS_PER_LOCALITY 4)". Why do we set it in the CMake file? What is the default if not set? Why 4 threads per locality?
<K-ballo>
the default is 1
<K-ballo>
must be some test that requires concurrency
<maxwellr96>
I'm trying to run a particular performance test in Phylanx with Ctest, but it says No tests were found. Does anyone know why that might be happening? I'm using this command:
<maxwellr96>
ctest -R tests.performance
<diehlpk_work>
have you built the tests?
<maxwellr96>
YEs
<maxwellr96>
*Yes
<primef>
K-ballo: thanks. In a normal use case application (not a test case) does setting that parameter impact performance? So is it something I have to tweak or should I ignore it?
<primef>
Moreover, does is refer to os_threads? Or what kind of threads?
<primef>
Just to know, what the maximum is I can set it too
<primef>
*to
<K-ballo>
in a normal application the default is all threads, I think?
<K-ballo>
it corresponds to the -t command line argument
<K-ballo>
they are os threads, yes
<primef>
just setting -t is recognized by hpx as threads argument?
<primef>
as for this fact, I saw some examples using this:
<primef>
parsed_options opts =
<primef>
command_line_parser(argc, argv)
<primef>
.allow_unregistered()
<primef>
.options(desc_commandline)
<primef>
.style(command_line_style::unix_style)
<primef>
.run();
<primef>
what exactly is parsed here? As apparently it is not necessary to parse command line options defined using program_options -> options_description
hkaiser has quit [Quit: bye]
<K-ballo>
the command line? (argc, argv)
<primef>
I thought the command line is parsed by using
<primef>
options_description desc_commandline;
<primef>
desc_commandline.add_options()
<primef>
variables_map vm;
<primef>
what is the difference between the two approaches? Or are they complementary and not exclusive?
<K-ballo>
I don't see any parsing in those three lines
<K-ballo>
looks like pseudo code for describing the options, and a map for storing the option values, but no actual parsing
<K-ballo>
the command line facilities are based on Boost.ProgramOptions, you may want to look into their docs