Adapted from md5block_amd64.s.

name                 old speed      new speed      delta
Hash8Bytes           14.0MB/s ± 1%  39.9MB/s ± 0%  +185.52%   (p=0.000 n=9+10)
Hash1K                176MB/s ± 1%   661MB/s ± 1%  +274.44%  (p=0.000 n=10+10)
Hash8K                196MB/s ± 0%   742MB/s ± 1%  +278.35%   (p=0.000 n=10+9)
Hash8BytesUnaligned  14.2MB/s ± 2%  39.8MB/s ± 0%  +180.06%  (p=0.000 n=10+10)
Hash1KUnaligned       177MB/s ± 1%   651MB/s ± 0%  +267.38%  (p=0.000 n=10+10)
Hash8KUnaligned       197MB/s ± 1%   731MB/s ± 1%  +271.73%  (p=0.000 n=10+10)

Change-Id: I45ece98ee10f30fcd192b9c3d743ba61c248f36a
Reviewed-on: https://go-review.googlesource.com/22505Reviewed-by: Bill O'Farrell <billotosyr@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>