2018-02-17 12:35 GMT+03:00 Niels Möller <[email protected]>:
> Dmitry Eremin-Solenikov <[email protected]> writes:
>
>> For benchmarking purposes provide wrappers around OpenSSL AES GCM
>> implementation. Note, digest callback will work only for encryption due
>> to OpenSSL internals.
>
> This patch and the next now merged to master-updates.
Thank you!
>
>> @@ -80,7 +80,7 @@ openssl_evp_set_encrypt_key(void *p, const uint8_t *key,
>> {
>> struct openssl_cipher_ctx *ctx = p;
>> ctx->evp = EVP_CIPHER_CTX_new();
>> - assert(EVP_EncryptInit_ex(ctx->evp, cipher, NULL, key, NULL) == 1);
>> + assert(EVP_CipherInit_ex(ctx->evp, cipher, NULL, key, NULL, 1) == 1);
>> EVP_CIPHER_CTX_set_padding(ctx->evp, 0);
>> }
>
> It's not right to use assert on expressions with side-effects. Since will
> break builds with ./configure CFLAGS='-DNDEBUG'.
>
> One would need to either assign return value to a variable and assert on
> that, or define some alternative assert-like makro which always
> evaluates its argument.
I've sent a followup patch with an additional variable. It is easier
to maintain.
Regarding the numbers, here is what I see on my laptop (i3-4005U):
libnettle: fat library initialization.
libnettle: cpu features: vendor:intel,aesni
libnettle: using aes instructions.
libnettle: not using sha_ni instructions.
libnettle: intel SSE2 will be used for memxor.
sha1_compress: 463.20 cycles
salsa20_core: 403.70 cycles
sha3_permute: 1838.10 cycles (76.59 / round)
benchmark call overhead: 0.003764 us 6.02 cycles
Algorithm mode Mbyte/s cycles/byte cycles/block
poly1305-aes update 951.68 1.60 1641.84
aes128 ECB encrypt 2328.39 0.66 10.49
aes128 ECB decrypt 1850.82 0.82 13.19
aes128 CBC encrypt 267.16 5.71 91.38
aes128 CBC decrypt 2015.34 0.76 12.11
aes128 (in-place) 1598.20 0.95 15.28
aes128 CTR 1683.51 0.91 14.50
aes128 (in-place) 1343.33 1.14 18.17
aes192 ECB encrypt 1695.07 0.90 14.40
aes192 ECB decrypt 1737.14 0.88 14.05
aes192 CBC encrypt 231.50 6.59 105.46
aes192 CBC decrypt 1548.88 0.99 15.76
aes192 (in-place) 1290.06 1.18 18.92
aes192 CTR 1349.77 1.13 18.09
aes192 (in-place) 1135.87 1.34 21.49
aes256 ECB encrypt 1462.52 1.04 16.69
aes256 ECB decrypt 1371.13 1.11 17.81
aes256 CBC encrypt 204.46 7.46 119.41
aes256 CBC decrypt 1296.66 1.18 18.83
aes256 (in-place) 1056.29 1.44 23.11
aes256 CTR 1123.06 1.36 21.74
aes256 (in-place) 981.12 1.56 24.88
openssl aes128 ECB encrypt 2308.04 0.66 10.58
openssl aes128 ECB decrypt 2294.12 0.67 10.64
openssl aes192 ECB encrypt 1929.11 0.79 12.66
openssl aes192 ECB decrypt 1926.87 0.79 12.67
openssl aes256 ECB encrypt 1653.41 0.92 14.77
openssl aes256 ECB decrypt 1666.08 0.92 14.65
gcm_aes128 encrypt 200.26 7.62 121.91
gcm_aes128 decrypt 200.26 7.62 121.91
gcm_aes128 update 240.93 6.33 101.33
gcm_aes192 encrypt 193.38 7.89 126.25
gcm_aes192 decrypt 194.50 7.85 125.52
gcm_aes192 update 241.02 6.33 101.29
gcm_aes256 encrypt 189.66 8.05 128.73
gcm_aes256 decrypt 189.74 8.04 128.67
gcm_aes256 update 240.14 6.35 101.67
openssl gcm_aes128 encrypt 1492.12 1.02 16.36
openssl gcm_aes128 decrypt 1496.77 1.02 16.31
openssl gcm_aes128 update 3683.07 0.41 6.63
openssl gcm_aes192 encrypt 1308.33 1.17 18.66
openssl gcm_aes192 decrypt 1315.36 1.16 18.56
openssl gcm_aes192 update 3662.13 0.42 6.67
openssl gcm_aes256 encrypt 1170.64 1.30 20.86
openssl gcm_aes256 decrypt 1173.83 1.30 20.80
openssl gcm_aes256 update 3489.30 0.44 7.00
--
With best wishes
Dmitry
_______________________________________________
nettle-bugs mailing list
[email protected]
http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
