00:43
EverYoung has quit [Ping timeout: 276 seconds]
01:01
Matombo has quit [Remote host closed the connection]
01:11
bikineev_ has quit [Remote host closed the connection]
01:13
<
github >
hpx/fixing_2847 d101822 Hartmut Kaiser: Fixing unrelated compilation errors
01:19
eschnett has joined #ste||ar
01:27
comtix has joined #ste||ar
01:33
comtix has left #ste||ar ["Leaving"]
02:22
hkaiser has quit [Quit: bye]
02:26
K-ballo has quit [Quit: K-ballo]
06:46
jaafar has quit [Ping timeout: 246 seconds]
07:33
david_pfander has joined #ste||ar
07:33
david_pfander has quit [Client Quit]
07:35
david_pfander has joined #ste||ar
07:42
Matombo has joined #ste||ar
07:49
mcopik has joined #ste||ar
07:54
eschnett has quit [Quit: eschnett]
07:55
eschnett has joined #ste||ar
08:06
bikineev has joined #ste||ar
08:37
Matombo has quit [Remote host closed the connection]
08:46
bikineev has quit [Remote host closed the connection]
09:02
<
github >
[hpx] atrantan opened pull request #2855: Add documentation for coarrays (master...coarray_docs)
https://git.io/v5TJ3
09:08
<
github >
hpx/gh-pages 117a115 StellarBot: Updating docs
10:13
hkaiser has joined #ste||ar
10:35
<
github >
hpx/fix_ubsan 1839bda Thomas Heller: Fixing component handling for lcos...
10:36
<
heller >
hkaiser: ^^
10:36
<
github >
[hpx] sithhell opened pull request #2856: Fixing component handling for lcos (master...fix_ubsan)
https://git.io/v5TOE
10:41
<
github >
hpx/fixing_2847 68a3177 Hartmut Kaiser: Fixing unrelated compilation errors
10:44
<
hkaiser >
heller: I don't like that - let's rather get rid of the simple_component_tag alltogether - that shouldn't break any code
10:45
<
heller >
hkaiser: that was something I thought of as well, wasn't sure you liked it so I went with the verbose solution
10:45
<
heller >
let me adapt
10:45
<
hkaiser >
heller: yah and I see now why adsan is complaining - this is benign, though
10:46
<
hkaiser >
heller: we've kept the simple_component_base as a compatibility solution, but the tag shouldn't matter
10:54
Smasher has quit [Ping timeout: 246 seconds]
11:01
hkaiser has quit [Read error: Connection reset by peer]
11:05
hkaiser has joined #ste||ar
11:08
<
heller >
hkaiser: yup
11:08
<
hkaiser >
nice catch, though
11:15
<
heller >
I am not sure if it is 100% benign though
11:15
<
heller >
is the layout between base_lco_with_value and simple_component garuanteed though?
11:21
Smasher has joined #ste||ar
11:29
<
hkaiser >
heller: I think the problem is that component_base and simple_component_base are not the same type, changing the derivation to a simple typedef should change things as well
11:29
K-ballo has joined #ste||ar
11:30
<
heller >
hkaiser: the problem is not with not deriving from _base
11:30
<
hkaiser >
heller: I don't think the problem is that base_lco_with_value and simple_component_base are differen ttypes
11:31
<
hkaiser >
it's that component_base and simple_component_base are not the same
11:31
<
heller >
the problem occurs when creating simple_component<Component>
11:32
<
heller >
I am not exactly sure how it ever worked...
11:32
<
heller >
does simple_component_base derive from simple_component?
11:32
<
heller >
or vice versa?
11:33
<
heller >
I am not sure how this create function ever worked
11:33
<
heller >
not for base_lco_with_value
11:33
<
hkaiser >
sure it will
11:33
<
hkaiser >
the tags need to be typedef'd as well, though
11:34
<
hkaiser >
heller: I think it's not about base_lco_with_value
11:34
<
heller >
first thing I did was to use simple_component_tag instead component_tag
11:35
<
heller >
the error then was a invalid downcast of lcos::server::channel<int> to simple_component<...>
11:36
<
heller >
using support is required nowadays, right?
11:37
<
heller >
I would rather switch it around, such that template <typename Component> using simple_component = component<Component>
11:37
<
hkaiser >
btw, the derivation chain is [simple_]component<> -> your component type -> [simple_]component_base<>
11:37
<
heller >
except for base lcos ;)
11:37
<
hkaiser >
heller: yah, fine by me
11:38
<
hkaiser >
base_lco is not a component
11:38
<
hkaiser >
it's an interface
11:39
<
heller >
ok, then I am really confused
11:39
<
hkaiser >
the component type derives from both, the base_lco and the component_base
11:39
<
heller >
base_lco_with_value shouldn't derive from the component tag then
11:40
<
hkaiser >
ok, good point - not sure why I added that
11:41
<
heller >
let me revisit the patch ...
11:41
<
heller >
makes more sense now
11:42
<
heller >
hkaiser: also, if base_lco_with_value is not a component, why does it have an entry in the component type database?
11:42
<
heller >
ah, because it is used as the base type for the actions...
11:44
<
heller >
given the inheritance chain above, this should have never worked in the first place
11:45
<
hkaiser >
yah, that looks suspicious ;)
11:45
<
hkaiser >
well, the derived type has no members, so it did work, but only coincidentally
11:54
<
heller >
hkaiser: the component_tag in base_loc_with_value is needed
11:54
<
heller >
so in some sense, base_lco_with_value is an abstract component
12:07
bikineev has joined #ste||ar
12:08
pree has joined #ste||ar
12:15
eschnett has quit [Quit: eschnett]
12:58
bikineev has quit [Ping timeout: 240 seconds]
13:04
bikineev has joined #ste||ar
13:04
eschnett has joined #ste||ar
13:09
pree has quit [Ping timeout: 240 seconds]
13:12
<
github >
hpx/fixing_2847 274c6e4 Hartmut Kaiser: Fixing unrelated compilation errors
13:17
<
hkaiser >
(at least with the way we have it now)
13:18
<
github >
hpx/fixing_2847 8563441 Hartmut Kaiser: Switching assert for an exception (review comment)
13:20
<
heller >
hkaiser: yes, I fixed that already
13:21
<
heller >
hkaiser: /home/inf3/heller/programming/hpx/hpx/traits/serialization_access_data.hpp:100:17: runtime error: store to misaligned address 0x7fd7948cbc11 for type 'std::uint64_t' (aka 'unsigned long'), which requires 8 byte alignment
13:21
<
heller >
0x7fd7948cbc11: note: pointer points here
13:21
<
heller >
00 00 00 00 00 00 00 00 00 00 00 00 ff ff ff ff ff ff ff 00 00 00 00 00 00 00 00 00 00 00 00 00
13:22
<
hkaiser >
not sure what you changed...
13:23
<
heller >
that's unrelated to my changes
13:24
<
hkaiser >
which test is causing this?
13:33
bikineev has quit [Ping timeout: 248 seconds]
13:35
<
github >
hpx/fix_ubsan 938127f Thomas Heller: Switching simple_component to be component...
13:35
<
heller >
investigating now
13:35
<
heller >
the component change should be complete now
13:35
eschnett has quit [Quit: eschnett]
13:35
<
hkaiser >
heller: great, thanks!
13:36
<
heller >
hkaiser: ahh, I think I know what's causing the unaligned stores
13:38
<
hkaiser >
heller: first I would like to go ahead with the resource_partitioner branch - any objections?
13:39
<
heller >
hkaiser: a few
13:41
hkaiser has quit [Quit: bye]
13:54
<
github >
hpx/fix_ubsan 1ba33f7 Thomas Heller: Switching simple_component to be component...
14:07
<
diehlpk_work >
heller, Do you know if all students passed GSoC 2015?
14:08
<
diehlpk_work >
Did we attend GSoC in 2014?
14:09
<
heller >
diehlpk_work: easy to find out
14:10
<
diehlpk_work >
I found it for 2015 up to 2017
14:11
<
diehlpk_work >
Where can I find the amount of passed students for 2014 and 2015?
14:11
<
diehlpk_work >
I start to make a statistics on the blog
14:12
<
diehlpk_work >
With nice plots ov amount of students per year, how many of them failed, PR per GSoC and so on
14:17
eschnett has joined #ste||ar
14:20
<
heller >
diehlpk_work: i'll check
14:21
hkaiser has joined #ste||ar
14:38
<
github >
hpx/resource_partitioner a051f7e Hartmut Kaiser: Addressing review comments
15:33
hkaiser has quit [Read error: Connection reset by peer]
15:35
hkaiser has joined #ste||ar
15:43
<
K-ballo >
hpx w/boost 1.65.0 works on vs17 u3.1
15:44
<
zao >
With Boost.Config mutations?
15:44
<
zao >
Or out-of-box?
15:44
<
K-ballo >
only mutation was silencing the annoying #message
15:44
<
K-ballo >
so you can consider it out-of-box
15:45
<
K-ballo >
it's also using the defaults, so /std:c++14 and so on
15:46
hkaiser has quit [Read error: Connection reset by peer]
15:47
hkaiser has joined #ste||ar
15:47
<
hkaiser >
K-ballo: yah, works with /std:c++latest as well
15:47
<
K-ballo >
with or without _HAS_AUTO_PTR ?
15:47
hkaiser has quit [Read error: Connection reset by peer]
15:48
hkaiser has joined #ste||ar
15:48
<
hkaiser >
K-ballo: sorry, horrible network here
15:48
<
hkaiser >
K-ballo: with _HAS_AUTO_PTR
15:50
<
K-ballo >
I'll give it a try later, plus /pedantic- as well
15:52
hkaiser_ has joined #ste||ar
15:52
hkaiser has quit [Read error: Connection reset by peer]
15:53
bikineev has joined #ste||ar
15:55
eschnett has quit [Quit: eschnett]
15:55
hkaiser_ has quit [Read error: Connection reset by peer]
15:57
hkaiser has joined #ste||ar
15:57
eschnett has joined #ste||ar
15:59
<
K-ballo >
wow, immediate error in the windows sdk
16:01
hkaiser has quit [Read error: Connection reset by peer]
16:02
david_pfander has quit [Ping timeout: 240 seconds]
16:04
hkaiser has joined #ste||ar
16:05
pree has joined #ste||ar
16:05
hkaiser has quit [Read error: Connection reset by peer]
16:21
hkaiser has joined #ste||ar
16:25
hkaiser has quit [Read error: Connection reset by peer]
16:35
hkaiser has joined #ste||ar
16:39
jaafar has joined #ste||ar
16:39
hkaiser has quit [Read error: Connection reset by peer]
16:43
hkaiser has joined #ste||ar
16:47
ajaivgeorge has joined #ste||ar
16:51
hkaiser has quit [Read error: Connection reset by peer]
16:54
hkaiser has joined #ste||ar
16:55
pree has quit [Remote host closed the connection]
16:55
pree has joined #ste||ar
16:57
EverYoung has joined #ste||ar
17:02
hkaiser has quit [Ping timeout: 255 seconds]
17:09
EverYoun_ has joined #ste||ar
17:12
EverYoung has quit [Ping timeout: 255 seconds]
17:13
EverYoun_ has quit [Ping timeout: 246 seconds]
17:20
EverYoung has joined #ste||ar
17:29
hkaiser has joined #ste||ar
17:36
hkaiser has quit [Read error: Connection reset by peer]
17:43
hkaiser has joined #ste||ar
17:47
hkaiser has quit [Read error: Connection reset by peer]
17:49
hkaiser has joined #ste||ar
17:54
EverYoung has quit [Remote host closed the connection]
17:54
hkaiser has quit [Read error: Connection reset by peer]
17:55
EverYoung has joined #ste||ar
17:59
EverYoung has quit [Ping timeout: 255 seconds]
18:01
EverYoung has joined #ste||ar
18:06
jakemp has joined #ste||ar
18:09
hkaiser has joined #ste||ar
18:11
hkaiser has quit [Read error: Connection reset by peer]
18:25
hkaiser has joined #ste||ar
18:31
hkaiser has quit [Ping timeout: 246 seconds]
18:39
bikineev has quit [Ping timeout: 240 seconds]
18:46
<
github >
hpx/pack-short-circuit dfc2e88 Agustin K-ballo Berge: Short-circuit all_of/any_of/none_of instantiations
19:05
bikineev has joined #ste||ar
19:14
<
K-ballo >
silly circle-ci is broken again
19:24
bikineev has quit [Remote host closed the connection]
19:25
bikineev has joined #ste||ar
19:54
bikineev has quit [Remote host closed the connection]
20:03
denis_blank has joined #ste||ar
20:05
bikineev has joined #ste||ar
20:13
hkaiser has joined #ste||ar
20:32
hkaiser has quit [Ping timeout: 248 seconds]
20:40
eschnett has quit [Quit: eschnett]
20:46
<
github >
hpx/resource_partitioner 78b4f27 Hartmut Kaiser: Addressing review comments
21:11
EverYoun_ has joined #ste||ar
21:14
EverYoung has quit [Ping timeout: 255 seconds]
21:25
<
github >
[hpx] hkaiser closed pull request #2853: Update Table 17. Non-modifying Parallel Algorithms in Documentation (master...patch-1)
https://git.io/v5UZL
21:25
<
github >
[hpx] hkaiser closed pull request #2855: Add documentation for coarrays (master...coarray_docs)
https://git.io/v5TJ3
21:26
hkaiser has joined #ste||ar
21:28
<
heller >
hkaiser: I am really a bit worried about this extra pointer
21:28
<
heller >
worried is too strong
21:28
<
heller >
I just don't like it ;)
21:35
hkaiser has quit [Ping timeout: 246 seconds]
21:43
<
github >
hpx/inspect_assert feed278 Hartmut Kaiser: Deprecate use of BOOST_ASSERT and ensure HPX_ASSERT has corresponding #include present
21:43
hkaiser has joined #ste||ar
21:43
<
hkaiser >
heller: I don't like it myself - just don't have a better solution
21:57
EverYoun_ has quit [Remote host closed the connection]
22:02
EverYoung has joined #ste||ar
22:05
<
heller >
hkaiser: not sure if it is really worth it
22:05
<
hkaiser >
that's what I'm saying - might not be worth it
22:06
jfbastien has quit [Read error: Connection reset by peer]
22:06
<
heller >
the loads are obviously in the respective read function
22:06
denis_blank has quit [Quit: denis_blank]
22:07
<
hkaiser >
what's the problem with unaligned reads/writes on x86?
22:07
<
heller >
do you remember what the gain of this optimization was?
22:07
<
hkaiser >
heller: this massively improves performance
22:07
<
K-ballo >
should make no difference in gcc/clang
22:08
<
heller >
yeah, memcpy is usually very cheap
22:08
<
K-ballo >
memcpy is a builtin for them
22:08
<
heller >
and intrinsically known by compilers
22:08
<
hkaiser >
K-ballo: it should not indeed, my measurements showed a significant effect
22:08
<
K-ballo >
msvc, on the other hand...
22:08
<
hkaiser >
K-ballo: it's an intrinsic there as well
22:08
<
K-ballo >
since when?
22:09
<
heller >
might not be a big deal on x86, other platforms might suffer though
22:09
<
heller >
and well, it is flagged by ubsan
22:09
<
K-ballo >
on x86 it's no longer a performance issue, but it is still undefined behavior
22:10
<
hkaiser >
is it now?
22:10
<
K-ballo >
"when you specify /Oi" ok, I'll have to recheck
22:11
<
heller >
hkaiser: I didn't run into a problem with it other than the ubsan diagnostic
22:11
<
K-ballo >
what's /Oi? is that /O<i> ?
22:11
<
hkaiser >
heller: let's wrap it into a pp constant and disable it for ubsan and other problematic platforms
22:13
<
hkaiser >
that could have been why I did see the benefit, it's not an intrinsic without /Oi
22:13
<
heller >
hkaiser: ok, which benchmark did you run that showed the performance increase?
22:13
<
K-ballo >
so it's an intrinsic in release builds.. I wonder now if calls to memcpy with hardcoded constants would have the same effect
22:14
<
hkaiser >
I think serialization_overhead
22:14
<
heller >
I'll try and see
22:14
<
hkaiser >
K-ballo: in our release builds /Oi is off, at least it's not explicitly on
22:14
<
hkaiser >
could be implicit with /O2
22:15
<
hkaiser >
yes, /O2 is the same as /Og /Oi /Ot /Oy /Ob2 /Gs /GF /Gy
22:15
pree has quit [Ping timeout: 248 seconds]
22:15
<
heller >
another option would be to have seperate unrolled stores/loads with char instead of the "natural" ones
22:15
<
hkaiser >
heller: try it
22:20
<
K-ballo >
on a regular VS project I see both /O2 and /Oi in Release configuration, but not so on HPX's project
22:28
<
hkaiser >
K-ballo: I think I removed /O2 at some point as one of the versions was giving an ICE
22:28
<
hkaiser >
I think VS2015
22:28
<
K-ballo >
RelWithDebInfo has /O2, no /Oi; Release has /Ox, no /Oi
22:28
<
hkaiser >
yah, /O2implies /Oi
22:29
<
hkaiser >
same for /Ox
22:29
<
K-ballo >
yet the project configuration says No
22:29
<
K-ballo >
whereas a regular VS project gets both /O2 and /Oi for Release, and project configuration says Yes
22:31
<
K-ballo >
I suppose No means no /Oi whereas
*No* means /Oi- ?
22:32
<
hkaiser >
who knows
22:33
<
K-ballo >
in that case, the entire effect of the optimization likely comes from the hard coded constants and not the intrinsics
22:50
eschnett has joined #ste||ar
22:55
EverYoung has quit [Remote host closed the connection]
23:09
EverYoung has joined #ste||ar
23:09
EverYoung has quit [Remote host closed the connection]
23:10
EverYoung has joined #ste||ar
23:23
hkaiser has quit [Read error: Connection reset by peer]
23:31
eschnett has quit [Quit: eschnett]
23:41
EverYoun_ has joined #ste||ar
23:44
EverYoung has quit [Ping timeout: 246 seconds]
23:45
EverYoun_ has quit [Ping timeout: 246 seconds]
23:51
hkaiser has joined #ste||ar