Re: [openpgp] OpenPGP encryption block modes

[Werner Koch <[e-mail]>]

On Wed, 10 Aug 2022 21:50, brian m. carlson said:

> at 866 MB/s.  This is a laptop with a Core i7-1280P and 32 GB of RAM
> using OpenSSL 3.0.5[0], which tends to have excellent performance.

I can throw in some other numbers from Libgcrypt 1.10.1 [0]

On a i5-2410M CPU @ 2.30GHz T220:

 AES256         |  nanosecs/byte   mebibytes/sec   cycles/byte  auto Mhz
        CFB enc |      2.42 ns/B     393.8 MiB/s      7.00 c/B      2890
        CFB dec |     0.367 ns/B      2596 MiB/s      1.06 c/B      2890
        GCM enc |      1.31 ns/B     726.5 MiB/s      3.79 c/B      2890
        GCM dec |     0.980 ns/B     973.6 MiB/s      2.83 c/B      2890
       GCM auth |     0.610 ns/B      1564 MiB/s      1.76 c/B      2890
        OCB enc |     0.399 ns/B      2390 MiB/s      1.15 c/B      2890
        OCB dec |     0.391 ns/B      2439 MiB/s      1.13 c/B      2890
       OCB auth |     0.379 ns/B      2516 MiB/s      1.10 c/B      2890
 SHA1           |      1.88 ns/B     507.8 MiB/s      5.43 c/B      2890
 SHA256         |      4.41 ns/B     216.0 MiB/s     12.76 c/B      2890
 SHA512         |      3.05 ns/B     312.5 MiB/s      8.82 c/B      2890

On a i5-10210U CPU @ 1.60GHz NUC:

 AES256         |  nanosecs/byte   mebibytes/sec   cycles/byte  auto Mhz
        CFB enc |     0.900 ns/B      1059 MiB/s      3.59 c/B      3989
        CFB dec |     0.219 ns/B      4350 MiB/s     0.875 c/B      3989
        GCM enc |     0.315 ns/B      3029 MiB/s      1.32 c/B      4188
        GCM dec |     0.315 ns/B      3025 MiB/s      1.32 c/B      4188
       GCM auth |     0.098 ns/B      9745 MiB/s     0.410 c/B      4189
        OCB enc |     0.209 ns/B      4565 MiB/s     0.875 c/B      4188
        OCB dec |     0.212 ns/B      4495 MiB/s     0.889 c/B      4188
       OCB auth |     0.209 ns/B      4562 MiB/s     0.876 c/B      4188
 SHA1           |     0.877 ns/B      1087 MiB/s      3.50 c/B      3988
 SHA256         |      2.01 ns/B     475.4 MiB/s      8.00 c/B      3988
 SHA512         |      1.40 ns/B     683.6 MiB/s      5.56 c/B      3988

On a AMD Ryzen 7 4800U NUC:

 AES256         |  nanosecs/byte   mebibytes/sec   cycles/byte  auto Mhz
        CFB enc |     0.819 ns/B      1164 MiB/s      3.50 c/B      4269
        CFB dec |     0.113 ns/B      8431 MiB/s     0.481 c/B      4256
        GCM enc |     0.242 ns/B      3938 MiB/s      1.03 c/B      4260±1
        GCM dec |     0.243 ns/B      3931 MiB/s      1.04 c/B      4268±3
       GCM auth |     0.120 ns/B      7931 MiB/s     0.513 c/B      4264±1
        OCB enc |     0.134 ns/B      7142 MiB/s     0.569 c/B      4259
        OCB dec |     0.131 ns/B      7297 MiB/s     0.556 c/B      4258±1
       OCB auth |     0.132 ns/B      7210 MiB/s     0.563 c/B      4259
 SHA1           |     0.443 ns/B      2151 MiB/s      1.89 c/B      4262±1
 SHA256         |     0.475 ns/B      2010 MiB/s      2.02 c/B      4265
 SHA512         |      1.22 ns/B     783.6 MiB/s      5.17 c/B      4245±4


I can't find our GnuPG benchmarks comparing CFB and OCB right now, but
the above numbers should give some hints: For the Ryzen OCB encryption
is roughly 15 times faster than CFB+SHA1.  Decryption is 4 times faster.



Salam-Shalom,

   Werner



[0] Libgcrypt 1.10:
    ./bench-slope --cpu-mhz auto cipher aes256
    ./bench-slope --cpu-mhz auto hash sha1 sha256
-- 
The pioneers of a warless world are the youth that
refuse military service.             - A. Einstein