03:34
hkaiser has quit [Quit: bye]
03:59
nikunj has quit [Remote host closed the connection]
04:03
nikunj has joined #ste||ar
04:39
nikunj97 has joined #ste||ar
06:19
jaafar has joined #ste||ar
06:22
Nikunj__ has joined #ste||ar
06:25
nikunj97 has quit [Ping timeout: 258 seconds]
06:32
ronniegandhi has joined #ste||ar
07:28
mdiers_ has quit [Remote host closed the connection]
07:29
mdiers_ has joined #ste||ar
08:16
Nikunj__ has quit [Quit: Leaving]
10:13
nikunj has quit [Remote host closed the connection]
10:21
nikunj has joined #ste||ar
14:14
ronniegandhi has quit [Ping timeout: 258 seconds]
14:27
hkaiser has joined #ste||ar
14:35
mdiers_ has quit [Quit: mdiers_]
14:36
mdiers_ has joined #ste||ar
14:39
<
hkaiser >
mdiers_: very high idle-rate - so not enough parallelism in your code for the number of cores you're running on
14:40
<
hkaiser >
try to increase the task sizes or reduce the number of cores, or both
14:43
<
mdiers_ >
the percentage should be high?
14:47
<
mdiers_ >
hkaiser: my idea was to calculate one block of work per numa domain to reduce the number of cores
14:48
<
hkaiser >
mdiers_: idle-rate should be low
14:48
<
hkaiser >
I think thinking about numa domain-related optimizations is premature, get things efficient first
14:49
<
hkaiser >
if you don't have sufficient work even numa domain can't help you
14:51
<
mdiers_ >
hkaiser: is 0.01% low? :confused
14:52
<
hkaiser >
0.01% is the unit of measure, the actual value is between 4180 (41.8%) and 8064 (80.64%)
14:54
<
mdiers_ >
ahhhh, tkz ;-)
15:09
<
mdiers_ >
how big should a task be? is a loop over 150 items with 10 calculations each too little?
15:10
<
mdiers_ >
hkaiser: thank you very much for the idle counter ;-)
15:12
<
hkaiser >
about 300 microseconds at least
15:15
<
hkaiser >
mdiers_: there is another counter giving you the average task length measured
15:15
<
hkaiser >
I think /threads/time/average or somesuch
15:21
ronniegandhi has joined #ste||ar
15:33
<
mdiers_ >
ok, the dist example is at 4.5 us, that explains it.
15:35
<
mdiers_ >
thanks a lot for the 300 microseconds. i will see how i get it changed.
16:26
<
Yorlik >
Question is at line 109 ff
16:27
<
Yorlik >
I am not sure if it is possible to do it at all, probably not the way how I wrote it in the comment.
16:27
<
Yorlik >
However - maybe someone has a good idea.
16:27
<
Yorlik >
Basically I want to pull out a component typed pointer through the type erasure.
16:32
<
hkaiser >
why do you type erase if you still need the type?
16:32
<
hkaiser >
for an example
16:32
<
Yorlik >
Because it kills any existing indirection.
16:35
<
Yorlik >
Thanks for that lnk - reading up ..
16:53
<
Yorlik >
BTW - the type is erased to have one single type to deal with at the top level, whioch can be stored in a vector. This top level type holds a type erased pointer to the polymorphic data.
18:09
nikunj97 has joined #ste||ar
19:00
nikunj has quit [Remote host closed the connection]
19:01
nikunj has joined #ste||ar
19:02
ronniegandhi has quit [Remote host closed the connection]
19:37
nikunj97 has quit [Ping timeout: 260 seconds]
19:59
Abhishek09 has joined #ste||ar
20:04
Abhishek09 has quit [Remote host closed the connection]
23:08
Yorlik has quit [Ping timeout: 240 seconds]