# SHA1 block procedure for Alpha.
# On 21264 performance is 33% better than code generated by vendor
-# compiler, and 75% better than GCC [3.4]. Implementation features
-# vectorized byte swap, but not Xupdate.
+# compiler, and 75% better than GCC [3.4], and in absolute terms is
+# 8.7 cycles per processed byte. Implementation features vectorized
+# byte swap, but not Xupdate.
@X=( "\$0", "\$1", "\$2", "\$3", "\$4", "\$5", "\$6", "\$7",
"\$8", "\$9", "\$10", "\$11", "\$12", "\$13", "\$14", "\$15");