00:00
<
hkaiser >
weilewei: yah, that's one way of doing it
00:19
kale[m] has quit [Ping timeout: 246 seconds]
00:19
kale[m] has joined #ste||ar
00:23
kale[m] has quit [Ping timeout: 246 seconds]
00:24
kale[m] has joined #ste||ar
00:55
Yorlik has quit [Ping timeout: 240 seconds]
01:24
<
weilewei >
hkaiser thanks
01:47
bita__ has joined #ste||ar
02:08
hkaiser has quit [Quit: bye]
02:22
RostamLog has joined #ste||ar
02:24
Yorlik has joined #ste||ar
04:00
akheir has quit [Quit: Leaving]
04:23
kale[m] has quit [Ping timeout: 246 seconds]
04:23
kale[m] has joined #ste||ar
05:26
Yorlik has quit [Ping timeout: 246 seconds]
05:43
bita__ has quit [Ping timeout: 260 seconds]
05:45
nikunj97 has joined #ste||ar
05:49
kale[m] has quit [Ping timeout: 260 seconds]
05:50
kale[m] has joined #ste||ar
07:00
nikunj97 has quit [Read error: Connection reset by peer]
07:00
Nikunj__ has joined #ste||ar
07:11
nikunj97 has joined #ste||ar
07:14
Nikunj__ has quit [Ping timeout: 260 seconds]
07:16
Nikunj__ has joined #ste||ar
07:17
nikunj97 has quit [Read error: Connection reset by peer]
07:20
nikunj97 has joined #ste||ar
07:22
Nikunj__ has quit [Ping timeout: 260 seconds]
09:49
kale[m] has quit [Ping timeout: 260 seconds]
09:53
kale[m] has joined #ste||ar
10:13
kale[m] has quit [Ping timeout: 240 seconds]
10:13
kale[m] has joined #ste||ar
11:00
nikunj97 has quit [Read error: Connection reset by peer]
11:11
nikunj97 has joined #ste||ar
11:49
Nikunj__ has joined #ste||ar
11:52
nikunj97 has quit [Ping timeout: 240 seconds]
12:12
Yorlik has joined #ste||ar
12:57
kale[m] has quit [Ping timeout: 260 seconds]
12:58
kale[m] has joined #ste||ar
14:00
Nikunj__ is now known as nikunj97
14:05
<
nikunj97 >
I'm getting bus errors while trying to allocate nsimd::pack<float> (a data structure) but it runs just fine using normal floats
14:09
<
ms[m] >
nikunj97: nope, sorry
14:09
<
ms[m] >
alignment issues with your SIMD types?
14:09
<
nikunj97 >
I use a library for my SIMD types, I do believe they should be aligned properly
14:10
<
nikunj97 >
how do I check if the simd types are aligned properly?
14:10
<
nikunj97 >
btw the code runs perfectly fine for smaller grid sizes
14:10
<
nikunj97 >
something like 8192x8192 runs smoothly. It is when I get to larger allocations that I face memory bus issues.
14:13
hkaiser has joined #ste||ar
14:24
<
ms[m] >
print the addresses ;) anyway, check if changing the allocator makes a difference
14:25
<
nikunj97 >
ms[m], tried both tcmalloc and jemalloc. It seems that the problem was with the processor itself. Running it on another node resolves the error.
14:25
<
nikunj97 >
weird stuff
14:27
<
nikunj97 >
btw it can't be a problem of alignment issues with nsimd as I got into the same error with large grid sizes of float
14:39
<
nikunj97 >
ms[m], it seems that the alignment issues are arising from boost::coroutines
14:40
<
nikunj97 >
or may be it is just the processor idk
14:48
K-ballo has quit [Quit: K-ballo]
14:49
K-ballo has joined #ste||ar
15:31
K-ballo has quit [Quit: K-ballo]
15:44
kale[m] has quit [Ping timeout: 246 seconds]
15:45
kale[m] has joined #ste||ar
15:45
hkaiser has quit [Quit: bye]
15:46
K-ballo has joined #ste||ar
15:49
kale[m] has quit [Ping timeout: 260 seconds]
15:50
kale[m] has joined #ste||ar
16:15
kale[m] has quit [Ping timeout: 256 seconds]
16:16
kale[m] has joined #ste||ar
17:20
kale[m] has quit [Ping timeout: 258 seconds]
17:21
kale[m] has joined #ste||ar
17:43
<
nikunj97 >
hpx::init: hpx::exception caught: failed to initialize machine affinity mask: HPX(kernel_error)
17:43
<
nikunj97 >
what is this error supposed to mean ^^
17:43
<
nikunj97 >
Is there a problem with my hpx installation?
17:50
bita__ has joined #ste||ar
18:09
<
zao >
Does hwloc reports a sane topology? Is this in a job or otherwise constrained.l by cgroups?
18:14
<
nikunj97 >
zao, lstopo shows the right topology
18:14
<
nikunj97 >
I'm running it on a single node
19:30
hkaiser has joined #ste||ar
19:57
kale[m] has quit [Ping timeout: 260 seconds]
19:57
kale[m] has joined #ste||ar
20:19
<
nikunj97 >
hkaiser, yt?
20:26
<
hkaiser >
nikunj97: here
20:26
<
nikunj97 >
hkaiser, I'm getting this error: hpx::init: hpx::exception caught: failed to initialize machine affinity mask: HPX(kernel_error)
20:26
<
hkaiser >
that means hwloc returned an error
20:27
<
nikunj97 >
the code runs perfectly fine on other processors. And ik hpx is setup properly coz examples and the other benchmark runs just fine
20:27
<
nikunj97 >
it is just this code that returns the erro
20:29
<
hkaiser >
can't think of what's happeneing, sorry
20:29
<
nikunj97 >
anyway to debug the code?
20:29
<
hkaiser >
use a debugger?
20:29
<
nikunj97 >
ugg.. sure
20:30
<
nikunj97 >
should I try rebuilding hwloc and build hpx again?
20:30
<
hkaiser >
shrug, not sure what's wrong
20:30
<
hkaiser >
I'd try to look at the arguments of the failing call
21:32
<
nikunj97 >
hkaiser, using pkg-config made it work for some reason
21:32
<
nikunj97 >
this is weird behavior ;-)
21:33
<
nikunj97 >
btw, looks like pkg-config does not add optimization flags
21:50
nikunj97 has quit [Read error: Connection reset by peer]
22:13
bita__ has quit [Ping timeout: 260 seconds]
22:34
kale[m] has quit [Ping timeout: 256 seconds]
22:34
kale[m] has joined #ste||ar
23:39
kale[m] has quit [Ping timeout: 264 seconds]
23:40
kale[m] has joined #ste||ar