Add Memcpy32 bench.

Add Memcpy32 bench.

This compares 32-bit copies using memcpy, autovectorization, and when SSE2 is
available, aligned and unaligned SSE2.

Running this on my desktop (Intel(R) Xeon(R) CPU E5-2690 0 @ 2.90GHz), I see 
all four perform essentially the same, except Clang's autovectorization looks
 a little better than GCC's.  memcpy is calling libc 2.19's __memcpy_sse2_unaligned.


BUG=skia:

Committed: http://code.google.com/p/skia/source/detail?r=14799

[mtklein, 2014-05-15 15:35:46]

If we're going to be looking at adding sk_memcpy32, I want to make sure we've
got a microbench just for that.

[qiankun, 2014-05-20 09:12:33]

lgtm besides a nit.
These four functions perform almost same on my machine.

https://codereview.chromium.org/290533002/diff/40001/gyp/bench.gypi
File gyp/bench.gypi (right):

https://codereview.chromium.org/290533002/diff/40001/gyp/bench.gypi#newcode57
gyp/bench.gypi:57: '../bench/MemcpyBench.cpp',
File order is broken. Keep in alphabetical order.

[mtklein, 2014-05-20 14:28:01]

lgtm

https://codereview.chromium.org/290533002/diff/40001/gyp/bench.gypi
File gyp/bench.gypi (right):

https://codereview.chromium.org/290533002/diff/40001/gyp/bench.gypi#newcode57
gyp/bench.gypi:57: '../bench/MemcpyBench.cpp',
On 2014/05/20 09:12:33, qiankun wrote:
> File order is broken. Keep in alphabetical order.

Whoops.  Done.

[commit-bot: I haz the power, 2014-05-20 14:28:14]

CQ is trying da patch. Follow status at
 https://skia-tree-status.appspot.com/cq/mtklein@chromium.org/290533002/60001

[commit-bot: I haz the power, 2014-05-20 14:54:06]

Change committed as 14799

[qiankun, 2014-05-22 08:07:12]

On 2014/05/20 14:54:06, I haz the power (commit-bot) wrote:
> Change committed as 14799
memcpy32_sse2_align perform best for large data size on ASUS T100 which equipped
with Intel(R) Atom(TM) CPU  Z3740  @ 1.33GHz.

   memcpy32_sse2_unalign_100000   NONRENDERING:  cmsecs =    105.42
    memcpy32_sse2_unalign_10000   NONRENDERING:  cmsecs =      6.29
     memcpy32_sse2_unalign_1000   NONRENDERING:  cmsecs =      0.52
      memcpy32_sse2_unalign_100   NONRENDERING:  cmsecs =      0.07
       memcpy32_sse2_unalign_10   NONRENDERING:  cmsecs =      0.03
     memcpy32_sse2_align_100000   NONRENDERING:  cmsecs =     85.95
      memcpy32_sse2_align_10000   NONRENDERING:  cmsecs =      6.24
       memcpy32_sse2_align_1000   NONRENDERING:  cmsecs =      0.32
        memcpy32_sse2_align_100   NONRENDERING:  cmsecs =      0.06
         memcpy32_sse2_align_10   NONRENDERING:  cmsecs =      0.03
  memcpy32_autovectorize_100000   NONRENDERING:  cmsecs =    210.41
   memcpy32_autovectorize_10000   NONRENDERING:  cmsecs =     15.44
    memcpy32_autovectorize_1000   NONRENDERING:  cmsecs =      1.45
     memcpy32_autovectorize_100   NONRENDERING:  cmsecs =      0.18
      memcpy32_autovectorize_10   NONRENDERING:  cmsecs =      0.03
         memcpy32_memcpy_100000   NONRENDERING:  cmsecs =    103.59
          memcpy32_memcpy_10000   NONRENDERING:  cmsecs =      4.34
           memcpy32_memcpy_1000   NONRENDERING:  cmsecs =      0.32
            memcpy32_memcpy_100   NONRENDERING:  cmsecs =      0.06
             memcpy32_memcpy_10   NONRENDERING:  cmsecs =      0.03