diff options
Diffstat (limited to 'src/lib/libcrypto/sha/asm/sha1-thumb.pl')
| -rw-r--r-- | src/lib/libcrypto/sha/asm/sha1-thumb.pl | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/src/lib/libcrypto/sha/asm/sha1-thumb.pl b/src/lib/libcrypto/sha/asm/sha1-thumb.pl index 7c9ea9b029..553e9cedb5 100644 --- a/src/lib/libcrypto/sha/asm/sha1-thumb.pl +++ b/src/lib/libcrypto/sha/asm/sha1-thumb.pl | |||
| @@ -14,7 +14,7 @@ | |||
| 14 | # The code does not present direct interest to OpenSSL, because of low | 14 | # The code does not present direct interest to OpenSSL, because of low |
| 15 | # performance. Its purpose is to establish _size_ benchmark. Pretty | 15 | # performance. Its purpose is to establish _size_ benchmark. Pretty |
| 16 | # useless one I must say, because 30% or 88 bytes larger ARMv4 code | 16 | # useless one I must say, because 30% or 88 bytes larger ARMv4 code |
| 17 | # [avialable on demand] is almost _twice_ as fast. It should also be | 17 | # [available on demand] is almost _twice_ as fast. It should also be |
| 18 | # noted that in-lining of .Lcommon and .Lrotate improves performance | 18 | # noted that in-lining of .Lcommon and .Lrotate improves performance |
| 19 | # by over 40%, while code increases by only 10% or 32 bytes. But once | 19 | # by over 40%, while code increases by only 10% or 32 bytes. But once |
| 20 | # again, the goal was to establish _size_ benchmark, not performance. | 20 | # again, the goal was to establish _size_ benchmark, not performance. |