暗号化の速度

次は暗号化の速度を計算．
計測マシンは，CPU: PentiumIII 933MHz x 2, Memory: 1GB, OS: Debian etch / Linux 2.6
測定するのは，AES 128bit CBCモード，AES 256bit CBCモード，sha1

% openssl speed aes-128-cbc aes-256-cbc sha1
Doing sha1 for 3s on 16 size blocks: 1030353 sha1's in 3.00s
Doing sha1 for 3s on 64 size blocks: 824617 sha1's in 3.01s
Doing sha1 for 3s on 256 size blocks: 500991 sha1's in 3.00s
Doing sha1 for 3s on 1024 size blocks: 194812 sha1's in 3.00s
Doing sha1 for 3s on 8192 size blocks: 29062 sha1's in 3.00s
Doing aes-128 cbc for 3s on 16 size blocks: 3020029 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 64 size blocks: 1348323 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 256 size blocks: 419475 aes-128 cbc's in 3.01s
Doing aes-128 cbc for 3s on 1024 size blocks: 111829 aes-128 cbc's in 3.00s
Doing aes-128 cbc for 3s on 8192 size blocks: 14270 aes-128 cbc's in 3.00s
Doing aes-256 cbc for 3s on 16 size blocks: 2647977 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 64 size blocks: 1078655 aes-256 cbc's in 3.01s
Doing aes-256 cbc for 3s on 256 size blocks: 319714 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 1024 size blocks: 83899 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 8192 size blocks: 10642 aes-256 cbc's in 3.00s
OpenSSL 0.9.8c 05 Sep 2006
built on: Mon Oct  2 16:47:57 UTC 2006
options:bn(64,32) md2(int) rc4(idx,int) des(ptr,risc1,16,long)
 aes(partial) blowfish(idx)
compiler: gcc -fPIC -DOPENSSL_PIC -DZLIB -DOPENSSL_THREADS -D_REENTRANT
 -DDSO_DLFCN -DHAVE_DLFCN_H -DL_ENDIAN -DTERMIO -O3 -march=i686 -Wa,--noexecstack
 -g -Wall -DOPENSSL_BN_ASM_PART_WORDS -DOPENSSL_IA32_SSE2 -DSHA1_ASM -DMD5_ASM
 -DRMD160_ASM -DAES_ASM
available timing options: TIMES TIMEB HZ=100 [sysconf value]
timing function used: times
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
sha1              5495.22k    17533.38k    42751.23k    66495.83k    79358.63k
aes-128 cbc      16106.82k    28764.22k    35676.28k    38170.97k    38966.61k
aes-256 cbc      14122.54k    22934.86k    27282.26k    28637.53k    29059.75k

という感じなんだけど，この値，k(キロ)じゃなくて，M(メガ)の間違いな気がする…

(8192bytes x 10642) / 3(s) x 1000(s) = 29056207333.33 = 29056.20M

というのはおいておいて，
AESの暗号化速度は，単純に相加平均して計算してみると，

AES 128: 252.3Mbps
AES 256: 192.3Mbps

という感じ，

sha1は，1024bytesのパケットを定常的に送信した場合

544.73Mbps

の速度が出る．
しかし，16bytesのパケットを定常的に送信したら，

44.0Mbps

しかでない．

PentiumIII 933MHzでAES 128使って暗号化している限り，限界値は250Mbpsあたり．
RC4使うと，この倍以上は出るっぽい．
しかし，1パケット毎に32バイト分だけ，AES 256で暗号化しているので，113Mbpsが限界っぽくて，netperfで測った速度はそれなりに良い結果なのがわかった．

つまり，どんなにプログラムを最適化しても，現状の2倍以上の性能は出せない．
通常の3倍には及ばないというわけです．
これならムーアの法則のほうが早いね．