hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar-group.org | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | This channel is logged: irclog.cct.lsu.edu
parsa[fn] has quit [Quit: Vazeyti]
K-ballo has quit [Ping timeout: 248 seconds]
K-ballo has joined #ste||ar
Yorlik_ has joined #ste||ar
Yorlik has quit [Ping timeout: 265 seconds]
hkaiser has quit [Quit: Bye!]
<gonidelis[m]> why is the range-expression deduced as a reference here (hence the error)?
<K-ballo> gonidelis[m]: that's not why the error
<K-ballo> you are trying to for-each a tuple
K-ballo1 has joined #ste||ar
K-ballo has quit [Ping timeout: 248 seconds]
K-ballo1 is now known as K-ballo
hkaiser has joined #ste||ar
<gonidelis[m]> Yeah figures. My bad
K-ballo has quit [Ping timeout: 260 seconds]
K-ballo has joined #ste||ar
K-ballo1 has joined #ste||ar
K-ballo has quit [Ping timeout: 268 seconds]
K-ballo1 is now known as K-ballo
weilewei has joined #ste||ar
<weilewei> hkaiser When I use more than 8 threads, the hpx version of lulesh will have much higher runtime than OpenMP version: https://snipboard.io/nSitsY.jpg
<weilewei> basically, I am changing hpx:threads and OMP_NUM_THREADS respectively
<weilewei> is it because there is too much parallelism needed when threads >= 8?
<hkaiser> most like because there is too little
<hkaiser> the amount of work per thread becomes too small
<weilewei> got it
<hkaiser> weilewei: I asked before if you are sure that all loops are better of with parallelization
<weilewei> but I wonder why the OpenMP version does not have such effects?
<weilewei> yeah, I recalled that
<hkaiser> omp doesn't create threads
<hkaiser> however, with HPX master thing should be better, I merged som eimprovements just the other day
<weilewei> so omp decides what will be the best number of threads?
<hkaiser> no, it just runs the functions (loop bodies), whereas HPX creates an HPX thread for those
<weilewei> I see
<hkaiser> for small number of cores things seem to be fine as you say, thereafter the amount of work becomes too small for HPX
<weilewei> Yes, for threads <=4, hpx version perform better than kokkos+openmp and openmp
<weilewei> I am thinking what will be a better way to show the graph lol
weilewei has quit [Quit: Ping timeout (120 seconds)]
weilewei has joined #ste||ar
weilewei has quit [Quit: Client closed]
<gonidelis[m]> hkaiser here
<hkaiser> gonidelis[m]: sec
<gonidelis[m]> no rush
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]
tufei_ has quit [Ping timeout: 255 seconds]
diehlpk has joined #ste||ar
diehlpk has quit [Quit: Leaving.]