# achieves respectful 432MBps on 2.8GHz processor now. For reference.
# If executed on Xeon, current RC4_CHAR code-path is 2.7x faster than
# RC4_INT code-path. While if executed on Opteron, it's only 25%
-# slower than the RC4_INT one [meaning that if CPU µ-arch detection
+# slower than the RC4_INT one [meaning that if CPU µ-arch detection
# is not implemented, then this final RC4_CHAR code-path should be
# preferred, as it provides better *all-round* performance].