Dan,
You ran interoperability tests, e.g. current non-optimized
bits interoperating with your new optimized bits via IPsec- right?
pkgdefs/SUNWcslr/Makefile
-------------------------
mcp-0 I assume this package never delivered the objects and
that's why you removed LICENSEFILES_i386.
lib/pkcs11/pkcs11_softtoken/amd64/Makefile
------------------------------------------
mcp-1 lines 71,75 s/amd64/$(MACH64)/
aes/amd64/aes_amd64.s
---------------------
mcp-2 line 167 s/has //
aes/aes_cbc_crypt.c
-------------------
mcp-3 line 296,508,824 s/16/AES_BLOCK_LEN/
Mark
Dan Anderson wrote:
> Here's a review for AES optimization on OpenSolaris x86 64-bit:
> 5072963 Need an optimized AES implementation for amd64
>
> http://dan.drydog.com/reviews/5072963-aes/
>
> I optimized AES by replacing C code with Dr. Brian Gladman's (faster) 64-bit
> assembly code available from http://fp.gladman.plus.com/AES/ Dr. Gladman's
> code also performed better than 64-bit assembly available from OpenSSL.
>
> I translated Dr. Gladman's Intel-based yasm assembly source to
> OpenSolaris/ATT/UNIX/Linux assembly source format and made changes to fit
> OpenSolaris prototypes and structures and to remove some byte-swapping.
>
> Here's a summary of microbenchmark results:
>
> Algo. .Benchmark . . . . amd64 . em64t
> aes128 openssl speed . . 21-34% 11-20%
> aes256 {en,de}crypt CLI .24-33% 14-20%
> aes128 userland . . . . . . 37% 12%
> aes256 Kernel module . . 47-50% 16-18%
>
> Please send comments by COB June 2, 2008.
>
> - Dan
>
>
> This message posted from opensolaris.org
> _______________________________________________
> security-discuss mailing list
> security-discuss at opensolaris.org
>
