K-ballo 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 quit [Quit: bye]
bita has quit [Remote host closed the connection]
bita has joined #ste||ar
bita has quit [Ping timeout: 264 seconds]
gnikunj[m] has quit [*.net *.split]
heller1 has quit [*.net *.split]
parsa[m] has quit [*.net *.split]
klaus[m] has quit [*.net *.split]
gonidelis[m] has quit [*.net *.split]
jaafar has quit [*.net *.split]
tid_the_harveste has quit [*.net *.split]
ms[m] has quit [*.net *.split]
mdiers[m] has quit [*.net *.split]
rori has quit [*.net *.split]
tiagofg[m] has quit [*.net *.split]
srinivasyadav227 has quit [*.net *.split]
wash[m] has quit [*.net *.split]
Vir has quit [*.net *.split]
diehlpk_work has quit [*.net *.split]
beauty2 has quit [*.net *.split]
zao has quit [*.net *.split]
bobakk3r_ has quit [*.net *.split]
parsa has quit [*.net *.split]
k-ballo[m] has quit [*.net *.split]
ramses[m] has quit [*.net *.split]
tarzeau has quit [*.net *.split]
K-ballo has quit [*.net *.split]
peltonp1 has quit [*.net *.split]
pedro_barbosa[m] has quit [*.net *.split]
vroni[m] has quit [*.net *.split]
jaafar has joined #ste||ar
parsa[m] has joined #ste||ar
diehlpk_work has joined #ste||ar
heller1 has joined #ste||ar
Vir has joined #ste||ar
ramses[m] has joined #ste||ar
wash[m] has joined #ste||ar
K-ballo has joined #ste||ar
parsa has joined #ste||ar
k-ballo[m] has joined #ste||ar
tarzeau has joined #ste||ar
klaus[m] has joined #ste||ar
gnikunj[m] has joined #ste||ar
peltonp1 has joined #ste||ar
bobakk3r_ has joined #ste||ar
pedro_barbosa[m] has joined #ste||ar
vroni[m] has joined #ste||ar
beauty2 has joined #ste||ar
gonidelis[m] has joined #ste||ar
zao has joined #ste||ar
klaus[m] has quit [Ping timeout: 258 seconds]
k-ballo[m] has quit [Ping timeout: 265 seconds]
vroni[m] has quit [Ping timeout: 268 seconds]
gnikunj[m] has quit [Ping timeout: 258 seconds]
parsa[m] has quit [Ping timeout: 258 seconds]
heller1 has quit [Ping timeout: 258 seconds]
gonidelis[m] has quit [Ping timeout: 258 seconds]
pedro_barbosa[m] has quit [Ping timeout: 268 seconds]
ramses[m] has quit [Ping timeout: 265 seconds]
tiagofg[m] has joined #ste||ar
pedro_barbosa[m] has joined #ste||ar
mdiers[m] has joined #ste||ar
k-ballo[m] has joined #ste||ar
srinivasyadav227 has joined #ste||ar
heller1 has joined #ste||ar
gnikunj[m] has joined #ste||ar
gonidelis[m] has joined #ste||ar
parsa[m] has joined #ste||ar
tid_the_harveste has joined #ste||ar
ms[m] has joined #ste||ar
rori has joined #ste||ar
mdiers[m] has quit [Ping timeout: 240 seconds]
ms[m] has quit [Ping timeout: 246 seconds]
heller1 has quit [Ping timeout: 240 seconds]
k-ballo[m] has quit [Ping timeout: 240 seconds]
pedro_barbosa[m] has quit [Ping timeout: 240 seconds]
tiagofg[m] has quit [Ping timeout: 240 seconds]
rori has quit [Ping timeout: 244 seconds]
srinivasyadav227 has quit [Ping timeout: 244 seconds]
gonidelis[m] has quit [Ping timeout: 244 seconds]
gnikunj[m] has quit [Ping timeout: 246 seconds]
parsa[m] has quit [Ping timeout: 258 seconds]
tid_the_harveste has quit [Ping timeout: 240 seconds]
<gonidelis[m]>
hkaiser: i think you can guess who the OP is ;)
<hkaiser>
ahh
<gonidelis[m]>
(i was waiting to present it to you, after the poll was over)
<hkaiser>
nice
<gonidelis[m]>
the results are interesting
<gonidelis[m]>
not great
<gonidelis[m]>
but interesting for sure
<hkaiser>
nod
<gonidelis[m]>
hkaiser for sure I can't expect from all cpp developers to care about performance. writing single core apps is the textbook anyways
<hkaiser>
nod
<srinivasyadav227>
hey! hkaiser: I have been going through documentation, different examples and c++ conferences in YouTube to get better insight of HPX programming model, I have couple of questions ? may I ask here? :-)
<gonidelis[m]>
So if you observe it from that point of view, then the positive responses are quite encouraging
<hkaiser>
gonidelis[m]: most people using c++ claim to do it for perf, however
<hkaiser>
srinivasyadav227: absolutely
<hkaiser>
gonidelis[m]: the issue with the parallel algorithms is that there are almost no implementations available
<srinivasyadav227>
does HPX concentrate mostly on CPU parallelism or even massive parallelism using GPUs? like providing variable grain size of parallelism for various tasks
<hkaiser>
they're not uniformly available and there is no teaching of this stuff happening
<hkaiser>
srinivasyadav227: gpu programming (writing kernels) is out of scope for hpx, what it does for you however is to enable you coordinating execution on cpus and gpus
<gonidelis[m]>
hkaiser even if half of the reactions are positive, it's still good. I mean we take over half of the population sample ;)
<hkaiser>
you can launch gpu kernels and get an hpx future back allowing to integrate the gpu operations with what's going on on the cpus
<hkaiser>
gonidelis[m]: +1
<gonidelis[m]>
hkaiser so you suggest that if the implementations show up then the interest will get higher ??
<hkaiser>
I'd expect so, yes
<gonidelis[m]>
hkaiser ok we are in a good situation then
<hkaiser>
look, nvidia and sycl have their implementations for gpus now
<srinivasyadav227>
hkaiser: oh got it..
<hkaiser>
srinivasyadav227: this is becomming more and more important, especially on newer (amd) architectures, where cpus and gpus share the same memory space
<srinivasyadav227>
hkaiser: the amd ROCm ?
<hkaiser>
since there will be no cache coherence between cpus and gpus on such architectures you need a means of corrdinating execution transparent to users
<hkaiser>
yah, that as well
<srinivasyadav227>
hkaiser: okay..and one more question :) ..how are Golang light weight threads (go routines) are different in performance compared HPX lightweight threads (tasks)? is it more like using CPU all the time with futures and continuations in HPX for speed up?
<hkaiser>
hpx threads are similar to go-routines
<hkaiser>
both are light-weight
<hkaiser>
in go they are built into the language, in c++ we built hpx to make them usable
<hkaiser>
srinivasyadav227: futures and continuations are the abstractions that allow to reason about threads without having to worry about threads ;-)
<hkaiser>
srinivasyadav227: got watch my talk from a while back 'Threads are the GOTO of today's computing'
<k-ballo[m]>
if cpp developers don't care about performance, why are they doing cpp in the first place? .. legacy?
<hkaiser>
unlikely
<hkaiser>
perf is the only real reason to use cpp, at least for companies
<gonidelis[m]>
k-ballo hkaiser your do they always aim to parallelization ?