U++ forum: Welcome to the forum

Re: String::Cat optimization [message #34591 is a reply to message #34590]

Thu, 01 December 2011 11:37

mirek
Messages: 13984
Registered: November 2005

Ultimate Member

dolik.rce wrote on Thu, 01 December 2011 04:21

koldo wrote on Thu, 01 December 2011 08:39

Please could somebody try it in Linux?

Here it is:

GCC-4.6.2 Optimal with svo_memcpy
TIMING Cat 40         :  2.65 s  - 265.34 ns (11.06 s  / 10000000 ), min:  0.00 ns, max:  5.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  0.00 ns -  0.00 ns ( 7.88 s  / 10000000 ), min:  0.00 ns, max:  3.00 ms, nesting: 1 - 10000000

GCC-4.6.2 Optimal with memcpy
TIMING Cat 40         :  2.94 s  - 293.57 ns (11.13 s  / 10000000 ), min:  0.00 ns, max:  4.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  0.00 ns -  0.00 ns ( 8.11 s  / 10000000 ), min:  0.00 ns, max:  1.00 ms, nesting: 1 - 10000000


GCC-4.6.2 Speed with svo_memcpy
TIMING Cat 40         : 246.75 ms - 24.68 ns (11.14 s  / 10000000 ), min:  0.00 ns, max:  4.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  0.00 ns -  0.00 ns ( 7.93 s  / 10000000 ), min:  0.00 ns, max:  4.00 ms, nesting: 1 - 10000000

GCC-4.6.2 Speed with memcpy
TIMING Cat 40         :  2.79 s  - 279.30 ns (11.45 s  / 10000000 ), min:  0.00 ns, max:  6.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  2.03 s  - 202.80 ns (10.68 s  / 10000000 ), min:  0.00 ns, max:  4.00 ms, nesting: 1 - 10000000



CLANG-2.9 Optimal with svo_memcpy
TIMING Cat 40         :  1.65 s  - 165.35 ns (11.07 s  / 10000000 ), min:  0.00 ns, max:  1.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  0.00 ns -  0.00 ns ( 8.09 s  / 10000000 ), min:  0.00 ns, max:  1.00 ms, nesting: 1 - 10000000

CLANG-2.9 Optimal with memcpy
TIMING Cat 40         :  4.20 s  - 420.01 ns (11.33 s  / 10000000 ), min:  0.00 ns, max:  2.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  1.20 s  - 119.81 ns ( 8.32 s  / 10000000 ), min:  0.00 ns, max:  1.00 ms, nesting: 1 - 10000000


CLANG-2.9 Speed with svo_memcpy
TIMING Cat 40         :  4.28 s  - 428.33 ns (11.21 s  / 10000000 ), min:  0.00 ns, max:  2.00 ms, nesting: 1 - 10000000
TIMING Cat 18         : 797.30 ms - 79.73 ns ( 7.72 s  / 10000000 ), min:  0.00 ns, max:  5.00 ms, nesting: 1 - 10000000

CLANG-2.9 Speed with memcpy
TIMING Cat 40         :  5.87 ms -  0.59 ns (11.09 s  / 10000000 ), min:  0.00 ns, max:  2.00 ms, nesting: 1 - 10000000
TIMING Cat 18         :  0.00 ns -  0.00 ns ( 8.36 s  / 10000000 ), min:  0.00 ns, max:  5.00 ms, nesting: 1 - 10000000

Honza

Looking at it, it seems like there is something fishy about times... maybe something with TIMING is now broken? Or perhaps it does bot behave well now in linux?

Report message to a moderator