aserio 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
EverYoung has quit [Ping timeout: 246 seconds]
vamatya has quit [Ping timeout: 260 seconds]
eschnett has joined #ste||ar
ajaivgeorge has quit [Ping timeout: 240 seconds]
ajaivgeorge has joined #ste||ar
ajaivgeorge_ has joined #ste||ar
ajaivgeorge has quit [Client Quit]
ajaivgeorge_ has quit [Quit: ajaivgeorge_]
ajaivgeorge has joined #ste||ar
ajaivgeorge is now known as ajaivgeorge_
ajaivgeorge_ has quit [Ping timeout: 240 seconds]
ajaivgeorge has joined #ste||ar
ajaivgeorge is now known as ajaivgeorge_
K-ballo has quit [Quit: K-ballo]
EverYoung has joined #ste||ar
hkaiser has quit [Quit: bye]
EverYoung has quit [Ping timeout: 260 seconds]
parsa[[w]] has joined #ste||ar
parsa[w] has quit [Ping timeout: 246 seconds]
ajaivgeorge_ has quit [Ping timeout: 240 seconds]
patg has joined #ste||ar
patg is now known as Guest75503
jaafar has quit [Ping timeout: 240 seconds]
Guest75503 has quit [Quit: See you later]
pree has joined #ste||ar
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 258 seconds]
taeguk has joined #ste||ar
bikineev_ has quit [Remote host closed the connection]
<david_pfander>
wash[m]: Hi, it looks like hpx is unable to find the Vc package if I use the STEllAR-GROUP/Vc (instead of Vir's repo). the VcConfig complains that Vc_LIBRARIES isn't set. Could you fix that?
<hkaiser>
david_pfander: Vc master?
<david_pfander>
hkaiser: yes
<hkaiser>
could you create a ticket on the HPX repo, pls?
<hkaiser>
I'll try to look into it
<david_pfander>
hkaiser: I can, if that's the place for. There unfortunately is no issue tracker for the STEllAR-GROUP/Vc repo
<hkaiser>
david_pfander: just to confirm, you're talking about Vc V1, right?
<david_pfander>
hkaiser: no(!!!!!) Vc V2!
<hkaiser>
david_pfander: well, octotiger picks up Vc through HPX
<david_pfander>
or rather, not it has become Vc 1.7 internally
<david_pfander>
*now
<david_pfander>
I know
<david_pfander>
if been using Vc1, but that doesn't support AVX512 afaik
<hkaiser>
david_pfander: ahh, you need to set HPX_WITH_DATAPAR_VC_NO_LIBRARY=On for this
<hkaiser>
pls create a ticket anyways, I think we can handle this automatically
<david_pfander>
hkaiser: ok, I'll try that later. Honestly Hartmut, how can I guess that? I spend have the afternoon trying to compile hpx...
<hkaiser>
david_pfander: ask wash, he added this
<david_pfander>
hkaiser: I'm creating an HPX ticket
<hkaiser>
yes, please - I'll try to fix it
<hkaiser>
the stellar/vc repo is probably outdated anywyas
<hkaiser>
david_pfander: have you tried Matthias repo, how is that different from ours?
<david_pfander>
hkaiser: that one works, but it doesn't support AVX512 afaik
<hkaiser>
k
<david_pfander>
hkaiser: I've now spend several days getting octotiger compiled on a KNL machine... and so far my experience was one horror after another...
<hkaiser>
david_pfander: what do you suggest?
<david_pfander>
hkaiser: we should at least settle for one of the Vc's, as their API is quite different
<hkaiser>
david_pfander: I think we should use the datapar API (i.e. V2?)
<david_pfander>
hkaiser: also, it would be nice if someone would fix the static compilation bug I filed. Because cross-compilation for KNL is not trivial on all computers
<hkaiser>
david_pfander: we all work as much on HPX as we can, everything is voluteered time
<david_pfander>
hkaiser: the dapar API/v2 is the one that we would like to have in the future, right? So, I'm for only supporting that one
<hkaiser>
ok, fine by me
<david_pfander>
hkaiser: I would also like to propose that we drop all other kinds of vectorization in octotiger, as it seems to be the only way for more people to actually use Vc2
<hkaiser>
fine by me as well
<david_pfander>
hkaiser: and sorry, just a bit frustrated right now, if've been a bit unlucky with the KNL machines
<david_pfander>
out of 3, one has too old software, one a configuration problem, and one is overloaded :(
<hkaiser>
heh
eschnett has quit [Ping timeout: 240 seconds]
ajaivgeorge has quit [Quit: ajaivgeorge]
ajaivgeorge has joined #ste||ar
denis_blank has quit [Quit: denis_blank]
<wash[m]>
I concur, would prefer to only support Vc2
<hkaiser>
wash[m]: do we need to update our repo from Vir's?
hkaiser has quit [Quit: bye]
<wash[m]>
Hkaiser: yes I suspect so
<wash[m]>
Yes I suspect so
pree has quit [Ping timeout: 240 seconds]
<david_pfander>
wash[m]: any idea why vc would generate avx instructions (v-prefixed assembly), but only use xmm (instead of the correct ymm) registers?
<david_pfander>
wash[m]: Possible reason: I'm compiling on a AMD Zen machine, I had to add an entry for zen in the optimize for architecture cmake file (based on the haswell entry)
<david_pfander>
(I checked that hpx and octotiger have all the avx2 and fma enabled)
EverYoung has joined #ste||ar
EverYoung has quit [Ping timeout: 240 seconds]
Smasher has quit [Ping timeout: 255 seconds]
Smasher has joined #ste||ar
hkaiser has joined #ste||ar
pree has joined #ste||ar
pree has quit [Client Quit]
pree has joined #ste||ar
mcopik has quit [Ping timeout: 240 seconds]
david_pfander has quit [Ping timeout: 260 seconds]