Dakota St. Laurent
c8179b88a7
added C version of example02
2015-07-16 00:05:12 -04:00
Dakota St. Laurent
15d0ffaac2
use system header files instead of local ones
2015-07-07 17:14:07 -04:00
Dakota St. Laurent
d954c321a0
added example 02
2015-06-24 13:28:54 -04:00
Dakota St. Laurent
eaef887f8e
major update to example01 - old tests weren't really useful
2015-06-23 16:52:01 -04:00
Dakota St. Laurent
5c5a02b499
added fifth version, which gives very strange results...
2015-06-19 09:50:49 -04:00
Dakota St. Laurent
de0533d168
lots of changes to example01:
...
- added a fourth version, which includes additional overhead also in version 01 (indexing, determining ratio, etc)
- added a 'verbose' command line option, where if verbose is true then ratios of timings will be outputted, otherwise just durations will be outputted
- made certain variables to be constants
- cleaned up comments and what not
- removed all instances of kernelfunctor
2015-06-18 14:15:37 -04:00
Dakota St. Laurent
835c48d476
reorganize directory
2015-06-17 17:05:28 -04:00
Dakota St. Laurent
bd11706eac
update example01 with different number of global and local threads -- now first trial is slower... why
2015-06-16 18:49:02 -04:00
Dakota St. Laurent
fd6b886c67
add configuration and compilation instructions; properly indent a function in example01
2015-06-12 19:00:45 -04:00
Dakota St. Laurent
1ea82c8b76
added implementation of one thread per element for example01 - it's the fastest thus far
2015-06-11 18:00:51 -04:00
Dakota St. Laurent
ef9df7f67c
add code to 01 that repeatedly writes buffers to show overhead; change vector names in kernel code to v1,v2,v3 to show that they're not the same as A,B,C in the C++ code
2015-06-09 16:07:18 -04:00
Dakota St. Laurent
ea3055fbc1
example 01 is compiling and running, and it's freaking fast ... I'm skeptical.
2015-06-05 19:36:32 -04:00
Dakota St. Laurent
07f78f6b3f
init commit; example 00 is working, example 01 is not
2015-06-04 13:48:36 -04:00
