Hi,
I'm doing some experiment on openssl-0.9.7-stable-SNAP-20020421. I replaced the AES code (the original AES code) with Brian Gladman's AES code (with some modification). (http://fp.gladman.plus.com/cryptography_technology/rijndael/) Then, I measure my new implementation's speed using 'openssl speed aes' command in openssl. Here's the original result: --- OpenSSL 0.9.7-dev XX xxx XXXX built on: Thu Jun 5 02:05:14 GMT-5 2002 options:bn(64,32) md2(int) rc4(idx,int) des(ptr,risc1,16,long) aes(partial) idea(int) blowfish(idx) compiler: gcc -fPIC -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -DOPENSSL_NO_KRB5 -DL_ENDIAN -DTERMIO -O3 -fomit-frame-pointer -mcpu=pentium -Wall -DSHA1_ASM -DMD5_ASM -DRMD160_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 aes-128 cbc 6258.41k 6621.88k 6719.09k 6749.09k 6777.02k aes-192 cbc 5425.44k 5594.44k 5557.54k 5569.25k 5624.07k aes-256 cbc 4800.59k 4929.72k 4979.67k 4992.43k 4931.75k --- And here's my implementation result: --- OpenSSL 0.9.7-dev XX xxx XXXX built on: Thu Jun 6 01:47:25 GMT-5 2002 options:bn(64,32) md2(int) rc4(idx,int) des(ptr,risc1,16,long) aes(partial) idea(int) blowfish(idx) compiler: gcc -fPIC -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -DOPENSSL_NO_KRB5 -DL_ENDIAN -DTERMIO -O3 -fomit-frame-pointer -mcpu=pentium -Wall -DSHA1_ASM -DMD5_ASM -DRMD160_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 aes-128 cbc 27900.53k 34734.04k 36680.45k 37256.87k 37552.13k aes-192 cbc 28848.54k 34869.08k 36367.02k 37445.97k 36814.85k aes-256 cbc 26579.55k 34030.28k 36479.91k 36972.89k 37538.47k --- I'm using PentiumIII-450Mhz-512KB cache, 126.69MB Physical Memory, Linux 2.4.7-10, and gcc-2.96. There are significant increases. I'm afraid that there is something wrong with my implementation. How could the speed increase so significant ? Any comments or explanations are welcome. -- Satria Bakti M. (13297096) Electrical Engineering Department Institut Teknologi Bandung -- ______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List [EMAIL PROTECTED] Automated List Manager [EMAIL PROTECTED]
