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/ | GSoC: https://github.com/STEllAR-GROUP/hpx/wiki/Google-Summer-of-Code-%28GSoC%29-2020
hkaiser has quit [Quit: bye]
k-ballo[m] has joined #ste||ar
kale[m] has joined #ste||ar
linus2 has quit [Ping timeout: 260 seconds]
linus2 has joined #ste||ar
kale[m] has quit [Ping timeout: 258 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 258 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 264 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Read error: Connection reset by peer]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 258 seconds]
bita_ has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Read error: Connection reset by peer]
kale[m] has joined #ste||ar
jbjnr has joined #ste||ar
kale[m] has quit [Read error: Connection reset by peer]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 246 seconds]
kale[m] has joined #ste||ar
nanmiao11 has quit [Remote host closed the connection]
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
weilewei has quit [Remote host closed the connection]
kale[m] has quit [Ping timeout: 240 seconds]
kale[m] has joined #ste||ar
hkaiser has joined #ste||ar
<gonidelis[m]> `sized_sentinel_for` is for random access iterators, right ?
<K-ballo> not necessarily, why?
<K-ballo> random access iterators are the only iterators that are sized sentinels for themselves, was that the question?
<diehlpk_work> ms[m], The JOSS paper will be accepted in the next two days
nanmiao11 has joined #ste||ar
<jbjnr> is the JOSS paper just a way to harvest citations?
<gonidelis[m]> K-ballo: why do we need a specific `sized_sentinel_for` trait and couldn't use instead `is_sentinel_for<>`...?
<K-ballo> sized sentinel is a refinement of sentinel
<hkaiser> jbjnr: essentially yes
<K-ballo> "sentinel_­for<S, I> && ..." that means refinement
<hkaiser> jbjnr: did the scheduler benchmarks run this time?
<jbjnr> yes, they look ok. Sorry forgot to post an update
<gonidelis[m]> K-ballo: but we use it to find out whether we could jump from the iterator to the sentinel in constant time... right?
<K-ballo> not... really
<K-ballo> sized sentinel can tell you how far you are from it
<gonidelis[m]> ok same thing
<K-ballo> not really
<K-ballo> it's possible you can tell how far you are from the sentinel, but not jump there in constant time
<gonidelis[m]> K-ballo: yeah you are right
<gonidelis[m]> so where do we used sized_sentinel_for ? I can't recall
<K-ballo> we? as in hpx? no idea
<hkaiser> jbjnr: thanks
<gonidelis[m]> yeah eyah
<hkaiser> gonidelis[m]: in distance()
<gonidelis[m]> duh.... it was just in front of me.
weilewei has joined #ste||ar
<hkaiser> rori, ms[m]: thanks for the release candidate! Good job!
<gonidelis[m]> hkaiser: you have an email ;)
jbjnr_ has joined #ste||ar
jbjnr has quit [Read error: Connection reset by peer]
<ms[m]> thank you rori for all the work on the release!
<rori> you're welcome :)
akheir has joined #ste||ar
<gonidelis[m]> rori: bravo!
kale[m] has quit [Ping timeout: 244 seconds]
<hkaiser> gonidelis[m]: very nice writeup, thanks
<gonidelis[m]> thank you very very much for your help
<gonidelis[m]> Do you think that Google may reject it as informal maybe?
<hkaiser> nah
<hkaiser> use it as your report
<gonidelis[m]> hkaiser: great. Thanks a lot! As I already told you... I would love it if someone who wanted to start working with us could understand the process of what to study and how ot prepare from a studen't prespective
bita_ has joined #ste||ar
<hkaiser> gonidelis[m]: when is the evaluation deadline?
<gonidelis[m]> monday
<gonidelis[m]> hkaiser: august 31
jbjnr__ has joined #ste||ar
jbjnr_ has quit [Read error: Connection reset by peer]
<hkaiser> gonidelis[m]: thanks
<weilewei> hkaiser can you take a look into my runtime config option for libcds max concurrent threads? I pass different values, but libcds threads did not change. I added here: https://github.com/STEllAR-GROUP/hpx/pull/4909/files#diff-c0a2f60ecf644e04af989f798db74d3dR52 and here:
<weilewei> I tried the following: ./bin/libcds_map_hazard_pointer_hpxthread_test hpx:ini=hpx.cds.num_concurrent_hazard_pointer_threads=256 but here still prints 128: https://github.com/STEllAR-GROUP/hpx/pull/4909/files#diff-49373334e165709ea12188f97ff57491R150-R153
hkaiser_ has joined #ste||ar
hkaiser has quit [Ping timeout: 240 seconds]
<hkaiser_> weilewei: this atomic is a global one, i.e. it will be initialized before HPX is initialized, so you will always get the default value, as your settings have not been handled yet
<hkaiser_> also, why is it an atomic?
<hkaiser_> do you anticipate this value changing concurrently?
<weilewei> Oh, I see, no, it is not necessary an atomic value.
<weilewei> Let me change it later hkaiser_
<diehlpk_work> ms[m], The JOSS paper got accepted
<diehlpk_work> We need some release of HPX to attach. Should we use HPX 1.5.0 or do a new release HPX 1.4.1-joss for the paper?
<diehlpk_work> We have to upload the release to Zenodo and the author list there has to match the JOSS paper.
<diehlpk_work> Let me know what you would prefer?
<diehlpk_work> 1) Have the 1.4.1-joss release and modify the Zenodo list there
<diehlpk_work> 2) Have 1.5.0 and have two Zenodo releases
<hkaiser_> diehlpk_work: let's wait for V1.5
<hkaiser_> why 2 zenodo releases?
<hkaiser_> there is only one zenodo doi that automatically refers to the latest release
<diehlpk_work> hkaiser_, The HPX 1.5.0 Zenodo release will have all authors contributing to HPX 1.5.0
<diehlpk_work> However, we need for the JOSS paper the same authors as on the JOSS paper
<diehlpk_work> So, we need two releases
<diehlpk_work> or not releases, better two Zenodo DOIs
<diehlpk_work> So it might would better do do the HPX 1.4.1-joss release with a Zenodo DOI matching the JOSS paper
<ms[m]> diehlpk_work: that seems strange... do we then have to make a x-joss release for every new release? or is the joss paper forever going to refer to 1.4.1-joss (or 1.5.0 if we wait for that?)
weilewei has quit [Remote host closed the connection]
<diehlpk_work> hkaiser_, we only need one release and the JOSS paper points to this release since the reviwers reviewed this version of HPX
<ms[m]> and is there going to be a chicken and egg problem with having the joss paper citation in the docs? or do we already have a doi for it now that it's accepted?
<hkaiser_> diehlpk_work: sure, let the paper refer to V1.5, it will be released next week or so
<hkaiser_> we can wait with the paper until its out
<hkaiser_> (if needed)
<hkaiser_> we should also merge the paper into the release
<diehlpk_work> ms[m], We need a release and Zenodo DOI to get the paper published
<hkaiser_> the zenodo doi will be generated automatically once we release it
<diehlpk_work> hkaiser_, Yes, and we can not use this Zendo doi
<hkaiser_> I think we should refer to the generic zenodo doi instead
<hkaiser_> yah, it's a circular dependency
<hkaiser_> we could merge the paper after the release, that would allow us to have it refer to the zeonodo doi generated by the release
<diehlpk_work> hkaiser_, we can not use the automaticaly generated Zendo DOI
<hkaiser_> why not?
<diehlpk_work> this will have any commiter to the gitgub repo
<diehlpk_work> which does not match the JOSS authors
<hkaiser_> ok, so what?
<diehlpk_work> The Zenodo DOI *have to match* the authors on the JOSS paper
<hkaiser_> all authors are listed by zenodo
<hkaiser_> no idea how this can be done
<diehlpk_work> Yes, but not all are on the JOSS paper
<hkaiser_> correct
<diehlpk_work> I need to use the tar ball of the HPX 1.5.0 release and do a manually zendo release
<diehlpk_work> And add the authors by hand
<diehlpk_work> So we would have two Zendo DOIs for HPX 1.5.0
<hkaiser_> uhh
<hkaiser_> well, whatever
<diehlpk_work> HPX 1.5.0 and HOX 1.5.0-joss or whatever
<hkaiser_> ok
<diehlpk_work> Ok, let us use HPX 1.5.0 and I will generate the package
weilewei has joined #ste||ar
<hkaiser_> yah, Dan said, the title of the zenodo release should match the paper title
<diehlpk_work> Yes, another point
<hkaiser_> diehlpk_work: alternatively we could use figshare for the archived version, that would avoid the zenodo mess
<diehlpk_work> hkaiser_, Sounds good
<weilewei> hkaiser_ I removed the atomic but still kept the static for that variable, and it is still unchanged after I pass different runtime value
<hkaiser_> weilewei: sure it stays unchanged
<hkaiser_> why should change now?
<weilewei> I thought we can pass different runtime value, and update the max_concurrent_attach_thread_hazard_pointer_
<hkaiser_> sure you can
<hkaiser_> the problem is that you initialize the global value directly, which happens before HPX is initialized
<hkaiser_> so you can't ask HPX for the value as it has not even processed the command linen at that point
<weilewei> oh I see, so I should not make the variable global
<hkaiser_> you can make it global if you want (which I don't like, btw), you need to initialize it later, however, for instance whenever the first libcds operation is invoked
<hkaiser_> you might not need a variable in the first place
<hkaiser_> just call get_config_value() whenever you need the number of threads
<weilewei> I see... let me try it
nanmiao11 has quit [Remote host closed the connection]
<weilewei> hkaiser_ I removed that variable, and use get_cofig_value() but it also gives me 10, no matters if I give any command line value: https://github.com/STEllAR-GROUP/hpx/pull/4909/files#diff-49373334e165709ea12188f97ff57491R68-R72
kale[m] has joined #ste||ar
<hkaiser_> weilewei: difficult to tell from just looking at the code
<hkaiser_> when is that called? is HPX up and running at that point already?
<weilewei> Yes, it is called when HPX is up
kale[m] has quit [Ping timeout: 258 seconds]
kale[m] has joined #ste||ar
<hkaiser_> weilewei: then I don't know, how do you set the value?
<weilewei> hkaiser_ I think I found the issue, I accidentally exported HPX_NUM_CONCURRENT_HAZARD_POINTER_THREADS=10 a very small number, and the executable picks it up; Now I export it to a large value 100
<weilewei> now it works, and I can update the value as well. also I missed -- before --hpx:ini=hpx.cds.num_concurrent_hazard_pointer_threads=256 so it does not recognize the value
<hkaiser_> ok
weilewei has quit [Remote host closed the connection]
weilewei has joined #ste||ar
jaafar has joined #ste||ar
diehlpk_work has quit [Remote host closed the connection]
diehlpk_work has joined #ste||ar
nanmiao11 has joined #ste||ar
jaafar has quit [Ping timeout: 260 seconds]
nanmiao11 has quit [Remote host closed the connection]
nanmiao11 has joined #ste||ar
jbjnr__ has quit [Ping timeout: 272 seconds]
bita_ has quit [Ping timeout: 260 seconds]
jaafar has joined #ste||ar
jaafar has quit [Ping timeout: 264 seconds]
weilewei has quit [Remote host closed the connection]