sha1-sparcv9a.pl: fix bug in commentary section.

diff --git a/crypto/sha/asm/sha1-sparcv9a.pl b/crypto/sha/asm/sha1-sparcv9a.pl
index 6902b52b24fb6234e7f24c4cac87f59757036b83..be7da91e483c6d587bf203343c241c5c18dd1f44 100644 (file)
--- a/crypto/sha/asm/sha1-sparcv9a.pl
+++ b/crypto/sha/asm/sha1-sparcv9a.pl
@@ -112,8 +112,9 @@ ___
 # virtually uniform performance of ~9.3 cycles per SHA1 round. Timings
 # mentioned above are theoretical lower limits. Real-life performance
 # was measured to be 6.6 cycles per SHA1 round on USIIi and 8.3 on
 # virtually uniform performance of ~9.3 cycles per SHA1 round. Timings
 # mentioned above are theoretical lower limits. Real-life performance
 # was measured to be 6.6 cycles per SHA1 round on USIIi and 8.3 on

-# USIII. The latter means that processor manual must have an error in
-# instruction latency table or there is some unmentioned shortcut...
+# USIII. The latter is lower than half-round VIS timing, because there
+# are 16 Xupdate-free rounds, which "push down" average theoretical
+# timing to 8 cycles...

 }
 
 # The reference Xupdate procedure is then "strained" over *pairs* of
 }
 
 # The reference Xupdate procedure is then "strained" over *pairs* of