进阶 www.wolfssl.com 2026-11-15 08:26:50 · 13 阅读
第27章 wolfSSL 性能基准测试详解
第27章 基准测试
wolfSSL 与 wolfCrypt 基准测试
参考基准测试目录
wolfJSSE 与 SunJSSE 对比
SiFive
HiFive Unleashed
Xilinx ZCU102 Petalinux
Renesas RX65N(搭载 TSIP,Trusted Secure IP)
Renesas AP-RX71M-0A
Renesas Synergy DK-S7G2
Espressif ESP32 WROOM
Intel x86_64
NXP i.MX6
Apple iPhone X
Microchip PIC32MZ
STM32(STM32F439、STM32F756)
STM32F777NI
Apple iPhone 8/8 Plus/X(Apple A11 Bionic)
Intel SGX
Intel QuickAssist
ARMv8
AVX1、AVX2
Atmel SAMD21
STM32F221G-EVAL
Texas Instruments – Tiva C 系列 TM4C1294XL – Connected Launchpad
Freescale TWR-K70F120M
Apple TV 2
mbed
相对密码算法性能
后量子 Kyber 基准测试(Linux)
后量子 Kyber 基准测试(ARM Cortex-M4)
后量子 Kyber 基准测试(MacOS)
Raspberry Pi Pico-W
Altera Agilex 5
wolfSSL 嵌入式 SSL/TLS 库从底层设计之初就充分考虑了可移植性、性能和内存占用。本章汇集了 wolfSSL 和 wolfCrypt 加密库的现有基准测试数据,并介绍如何在你自己的平台上对 wolfSSL 进行基准测试。如需更多基准测试数据,或对特定平台有疑问,请发送邮件至 facts@wolfssl.com 联系我们。
wolfCrypt 基准测试应用
很多用户都想知道 wolfSSL 嵌入式 SSL/TLS 库在特定硬件设备或环境下的表现。由于如今嵌入式、企业级和云环境所使用的平台和编译器种类繁多,很难给出通用的性能数据。
为了帮助用户和客户评估 wolfSSL 和 wolfCrypt 的性能,wolfSSL 附带了一个基准测试应用。由于底层加密是 SSL/TLS 中对性能影响最大的部分,这个应用主要测试 wolfCrypt 各算法的性能。
运行基准测试
基准测试工具位于 wolfSSL 包的 “./wolfcrypt/benchmark” 目录。构建好 wolfSSL 及相关示例和应用后,在包根目录下执行以下命令即可运行:
./wolfcrypt/benchmark/benchmark
典型输出如下(展示 MB/s 吞吐量和每字节的时钟周期数):
wolfCrypt Benchmark (block bytes 1048576, min 1.0 sec each)
RNG 100 MB took 1.047 seconds, 95.466 MB/s Cycles per byte = 22.92
AES-128-CBC-enc 200 MB took 1.020 seconds, 196.027 MB/s Cycles per byte = 11.16
AES-128-CBC-dec 215 MB took 1.008 seconds, 213.318 MB/s Cycles per byte = 10.26
AES-192-CBC-enc 175 MB took 1.016 seconds, 172.265 MB/s Cycles per byte = 12.70
AES-192-CBC-dec 180 MB took 1.009 seconds, 178.405 MB/s Cycles per byte = 12.27
AES-256-CBC-enc 150 MB took 1.007 seconds, 148.932 MB/s Cycles per byte = 14.69
AES-256-CBC-dec 160 MB took 1.026 seconds, 155.994 MB/s Cycles per byte = 14.03
AES-128-GCM-enc 60 MB took 1.010 seconds, 59.427 MB/s Cycles per byte = 36.82
AES-128-GCM-dec 65 MB took 1.070 seconds, 60.750 MB/s Cycles per byte = 36.02
AES-192-GCM-enc 60 MB took 1.050 seconds, 57.138 MB/s Cycles per byte = 38.30
AES-192-GCM-dec 60 MB took 1.024 seconds, 58.590 MB/s Cycles per byte = 37.35
AES-256-GCM-enc 55 MB took 1.029 seconds, 53.438 MB/s Cycles per byte = 40.95
AES-256-GCM-dec 60 MB took 1.090 seconds, 55.069 MB/s Cycles per byte = 39.74
CHACHA 360 MB took 1.001 seconds, 359.628 MB/s Cycles per byte = 6.09
CHA-POLY 285 MB took 1.014 seconds, 280.943 MB/s Cycles per byte = 7.79
MD5 450 MB took 1.010 seconds, 445.573 MB/s Cycles per byte = 4.91
POLY1305 1265 MB took 1.000 seconds, 1264.402 MB/s Cycles per byte = 1.73
SHA 475 MB took 1.000 seconds, 474.914 MB/s Cycles per byte = 4.61
SHA-224 210 MB took 1.018 seconds, 206.308 MB/s Cycles per byte = 10.61
SHA-256 210 MB took 1.018 seconds, 206.200 MB/s Cycles per byte = 10.61
SHA-384 280 MB took 1.016 seconds, 275.520 MB/s Cycles per byte = 7.94
SHA-512 275 MB took 1.000 seconds, 274.868 MB/s Cycles per byte = 7.96
SHA3-224 240 MB took 1.006 seconds, 238.506 MB/s Cycles per byte = 9.18
SHA3-256 225 MB took 1.007 seconds, 223.454 MB/s Cycles per byte = 9.79
SHA3-384 175 MB took 1.002 seconds, 174.610 MB/s Cycles per byte = 12.53
SHA3-512 125 MB took 1.031 seconds, 121.254 MB/s Cycles per byte = 18.05
HMAC-MD5 445 MB took 1.001 seconds, 444.651 MB/s Cycles per byte = 4.92
HMAC-SHA 470 MB took 1.009 seconds, 465.749 MB/s Cycles per byte = 4.70
HMAC-SHA224 200 MB took 1.001 seconds, 199.874 MB/s Cycles per byte = 10.95
HMAC-SHA256 205 MB took 1.004 seconds, 204.228 MB/s Cycles per byte = 10.72
HMAC-SHA384 290 MB took 1.009 seconds, 287.401 MB/s Cycles per byte = 7.61
HMAC-SHA512 290 MB took 1.013 seconds, 286.214 MB/s Cycles per byte = 7.65
RSA 2048 public 2800 ops took 1.014 sec, avg 0.362 ms, 2761.995 ops/sec
RSA 2048 private 300 ops took 1.308 sec, avg 4.359 ms, 229.402 ops/sec
DH 2048 key gen 735 ops took 1.001 sec, avg 1.361 ms, 734.608 ops/sec
DH 2048 key agree 800 ops took 1.123 sec, avg 1.404 ms, 712.131 ops/sec
ECC 256 key gen 1108 ops took 1.001 sec, avg 0.903 ms, 1107.306 ops/sec
ECDHE 256 agree 1200 ops took 1.043 sec, avg 0.869 ms, 1150.329 ops/sec
ECDSA 256 sign 1200 ops took 1.078 sec, avg 0.898 ms, 1113.279 ops/sec
ECDSA 256 verify 1700 ops took 1.045 sec, avg 0.615 ms, 1627.064 ops/sec
这个应用对于比较更换数学库前后的公钥运算速度特别有用。你可以分别测试默认数学库(./configure)、fastmath 库(./configure –enable-fastmath)、fasthugemath 库(./configure –enable-fasthugemath)和 sp-math-all 库(./configure –enable-sp-math-all)的结果。
注意:默认情况下,报告单位会根据每项基准的数值自动缩放。如果想让所有报告值使用统一的固定单位,可在构建时定义 WOLFSSL_BENCHMARK_FIXED_UNITS_XX,其中 XX 可以是 GB、MB、KB 或 B(字节)。例如,./configure CFLAGS=”-DWOLFSSL_BENCHMARK_FIXED_UNITS_MB” 会以 MB 显示所有数值。
内存占用
wolfSSL 的编译后二进制体积(footprint)在 20-100kB 之间,具体取决于构建选项和所用编译器。在嵌入式系统上使用优化的嵌入式编译器时,构建体积通常在 60kB 左右,其中包含功能完整的 TLS 1.2 客户端和服务端。关于构建选项和进一步定制 wolfSSL 的方法,请参阅 CyaSSL 手册第 2 章或 wolfSSL Tuning Guide。
运行时内存方面,wolfSSL 每个 SSL/TLS 会话通常占用 1-36 kB。每连接的 RAM 占用取决于输入/输出缓冲区大小、公钥算法和密钥长度。wolfSSL 的 I/O 缓冲区默认为 128 字节,由 ./wolfssl/internal.h 中的 RECORD_SIZE 宏控制,最大为每缓冲区 16 kB(符合 SSL/TLS RFC 规定)。举例来说,使用标准 16kB 缓冲区时,单连接下 wolfSSL 的总运行时内存为 3kB(库本身)+ 16kB(输入缓冲区)+ 16kB(输出缓冲区)≈ 35kB。
TLS 上下文(WOLFSSL_CTX)由客户端或服务端的所有 TLS 连接共享,其运行时内存占用取决于加载的证书数量和证书文件大小,也与会话缓存及是否开启会话证书存储(–enable-session-certs)有关。如果想缩小会话缓存,可以定义 SMALL_SESSION_CACHE(把默认的 33 个会话缓存缩减为 6 个),可节省约 2.5 kB;也可以定义 NO_SESSION_CACHE 完全禁用会话缓存,节省近 3 kB 内存。
参考基准测试
在把 wolfSSL 移植到各种平台的过程中,我们经常会在这些平台上进行基准测试。下面列出部分测试数据供参考。如果你在特定平台上测试过 wolfSSL,欢迎把测试结果(附上具体平台和库配置)发给我们,我们会将其加入列表!
平台:
wolfSSL JSSE Provider(Mac OSX 上与默认 SunJSSE 对比)
基准测试:
以下数据展示了使用 wolfSSL Java JSSE Provider 相比默认 SunJSSE provider 的性能提升。
客户端及配置 平均连接时间
wolfSSL 纯 C(无 Java,软件实现) 9.694 ms
wolfSSL 纯 C(无 Java,intelasm + sp + sp-asm) 7.302 ms
wolfJSSE 客户端(仅软件实现) 10.92 ms
wolfJSSE 客户端(sp + intelasm) 8.42 ms
wolfJSSE 客户端(TLS 1.3 sp + intelasm) 8.04 ms
SunJSSE provider 客户端(Mac 默认) 13.34 ms
关于使用 wolfSSL JSSE Provider 的更多信息,请参阅用户手册。
返回页首
平台:
SiFive:SiFive HiFive1 Rev B
基准测试:
测试数据采集于 320MHz 时钟频率,配置包含 4608 字节栈空间和 2048 字节堆空间。
wolfCrypt Benchmark (block bytes 1024, min 1.0 sec each)
RNG 250 KB took 1.098 seconds, 227.714 KB/s
AES-128-CBC-enc 50 KB took 1.132 seconds, 44.175 KB/s
AES-128-CBC-dec 50 KB took 1.142 seconds, 43.778 KB/s
AES-192-CBC-enc 50 KB took 1.250 seconds, 40.007 KB/s
AES-192-CBC-dec 50 KB took 1.260 seconds, 39.677 KB/s
AES-256-CBC-enc 50 KB took 1.368 seconds, 36.552 KB/s
AES-256-CBC-dec 50 KB took 1.378 seconds, 36.279 KB/s
AES-128-GCM-enc 25 KB took 1.225 seconds, 20.412 KB/s
AES-128-GCM-dec 25 KB took 1.225 seconds, 20.402 KB/s
AES-192-GCM-enc 25 KB took 1.290 seconds, 19.373 KB/s
AES-192-GCM-dec 25 KB took 1.291 seconds, 19.366 KB/s
AES-256-GCM-enc 25 KB took 1.352 seconds, 18.487 KB/s
AES-256-GCM-dec 25 KB took 1.353 seconds, 18.478 KB/s
CHACHA 1 MB took 1.006 seconds, 1.020 MB/s
CHA-POLY 700 KB took 1.032 seconds, 678.045 KB/s
POLY1305 2 MB took 1.007 seconds, 2.255 MB/s
SHA 2 MB took 1.002 seconds, 1.511 MB/s
SHA-256 525 KB took 1.011 seconds, 519.279 KB/s
SHA-512 275 KB took 1.017 seconds, 270.477 KB/s
HMAC-SHA 1 MB took 1.013 seconds, 1.399 MB/s
HMAC-SHA256 525 KB took 1.019 seconds, 515.020 KB/s
HMAC-SHA512 275 KB took 1.032 seconds, 266.351 KB/s
ECC 256 key gen 2 ops took 1.104 sec, avg 551.834 ms, 1.812 ops/sec
ECDHE 256 agree 2 ops took 1.101 sec, avg 550.400 ms, 1.817 ops/sec
ECDSA 256 sign 2 ops took 1.173 sec, avg 586.502 ms, 1.705 ops/sec
ECDSA 256 verify 2 ops took 2.153 sec, avg 1076.294 ms, 0.929 ops/sec
CURVE 25519 key gen 2 ops took 1.629 sec, avg 814.423 ms, 1.228 ops/sec
CURVE 25519 agree 2 ops took 1.626 sec, avg 813.156 ms, 1.230 ops/sec
ED 25519 key gen 1 ops took 1.436 sec, avg 1436.096 ms, 0.696 ops/sec
ED 25519 sign 2 ops took 2.913 sec, avg 1456.421 ms, 0.687 ops/sec
ED 25519 verify 2 ops took 5.012 sec, avg 2506.012 ms, 0.399 ops/sec
返回页首
平台:
HiFive Unleashed @ 1.4GHz
基准测试:
新的汇编实现速度约提升了 50 倍
./configure --enable-riscv-asm && make
root@HiFiveU:~/wolfssl-riscv# ./wolfcrypt/benchmark/benchmark -aes-cbc -aes-gcm------------------------------------------------------------------------------
wolfSSL version 5.7.0
------------------------------------------------------------------------------
Math: Multi-Precision: Wolf(SP) word-size=64 bits=3072 sp_int.c
wolfCrypt Benchmark (block bytes 1048576, min 1.0 sec each)
AES-128-CBC-enc 20 MiB took 1.076 seconds, 18.588 MiB/s
AES-128-CBC-dec 20 MiB took 1.083 seconds, 18.473 MiB/s
AES-192-CBC-enc 20 MiB took 1.245 seconds, 16.062 MiB/s
AES-192-CBC-dec 20 MiB took 1.246 seconds, 16.047 MiB/s
AES-256-CBC-enc 15 MiB took 1.057 seconds, 14.189 MiB/s
AES-256-CBC-dec 15 MiB took 1.055 seconds, 14.212 MiB/s
AES-128-GCM-enc 15 MiB took 1.300 seconds, 11.543 MiB/s
AES-128-GCM-dec 15 MiB took 1.300 seconds, 11.535 MiB/s
AES-192-GCM-enc 15 MiB took 1.425 seconds, 10.526 MiB/s
AES-192-GCM-dec 15 MiB took 1.425 seconds, 10.523 MiB/s
AES-256-GCM-enc 10 MiB took 1.032 seconds, 9.687 MiB/s
AES-256-GCM-dec 10 MiB took 1.032 seconds, 9.691 MiB/s
GMAC Table 4-bit 31 MiB took 1.025 seconds, 30.251 MiB/s
Benchmark complete
master 分支
./configure —enable-all && make
root@HiFiveU:~/wolfssl# ./wolfcrypt/benchmark/benchmark -aes-cbc -aes-gcm
------------------------------------------------------------------------------
wolfSSL version 5.7.0
------------------------------------------------------------------------------
Math: Multi-Precision: Wolf(SP) word-size=64 bits=4096 sp_int.c
wolfCrypt Benchmark (block bytes 1048576, min 1.0 sec each)
AES-128-CBC-enc 5 MiB took 12.798 seconds, 0.391 MiB/s
AES-128-CBC-dec 5 MiB took 12.672 seconds, 0.395 MiB/s
AES-192-CBC-enc 5 MiB took 15.301 seconds, 0.327 MiB/s
AES-192-CBC-dec 5 MiB took 15.181 seconds, 0.329 MiB/s
AES-256-CBC-enc 5 MiB took 17.820 seconds, 0.281 MiB/s
AES-256-CBC-dec 5 MiB took 17.669 seconds, 0.283 MiB/s
AES-128-GCM-enc 5 MiB took 12.870 seconds, 0.388 MiB/s
AES-128-GCM-dec 5 MiB took 12.870 seconds, 0.388 MiB/s
AES-192-GCM-enc 5 MiB took 15.375 seconds, 0.325 MiB/s
AES-192-GCM-dec 5 MiB took 15.376 seconds, 0.325 MiB/s
AES-256-GCM-enc 5 MiB took 17.878 seconds, 0.280 MiB/s
AES-256-GCM-dec 5 MiB took 17.896 seconds, 0.279 MiB/s
AES-128-GCM-STREAM-enc 5 MiB took 12.878 seconds, 0.388 MiB/s
AES-128-GCM-STREAM-dec 5 MiB took 12.878 seconds, 0.388 MiB/s
AES-192-GCM-STREAM-enc 5 MiB took 15.379 seconds, 0.325 MiB/s
AES-192-GCM-STREAM-dec 5 MiB took 15.385 seconds, 0.325 MiB/s
AES-256-GCM-STREAM-enc 5 MiB took 17.881 seconds, 0.280 MiB/s
AES-256-GCM-STREAM-dec 5 MiB took 17.888 seconds, 0.280 MiB/s
GMAC Table 4-bit 30 MiB took 1.006 seconds, 29.831 MiB/s
Benchmark complete
返回页首
平台:
Renesas:Renesas RX65N(R5F565NEHDFB)
基准测试:
以下数据展示了通过 Renesas TSIP 驱动使用硬件加密后,Renesas RX65N 上的性能提升。
使用硬件加速器(含 TSIP)
wolfCrypt Benchmark (block bytes 1024, min 1.0 sec each)
RNG 2 MB took 1.001 seconds, 1.928 MB/s
AES-128-CBC-enc 3 MB took 1.005 seconds, 2.818 MB/s
AES-128-CBC-dec 3 MB took 1.002 seconds, 2.631 MB/s
AES-256-CBC-enc 3 MB took 1.001 seconds, 2.779 MB/s
AES-256-CBC-dec 3 MB took 1.002 seconds, 2.680 MB/s
AES-128-GCM-enc 3 MB took 1.007 seconds, 2.740 MB/s
AES-128-GCM-dec 3 MB took 1.005 seconds, 2.647 MB/s
AES-256-GCM-enc 3 MB took 1.002 seconds, 2.679 MB/s
AES-256-GCM-dec 3 MB took 1.005 seconds, 2.649 MB/s
ARC4 3 MB took 1.001 seconds, 3.440 MB/s
RABBIT 4 MB took 1.004 seconds, 3.987 MB/s
3DES 200 KB took 1.053 seconds, 189.988 KB/s
MD5 6 MB took 1.004 seconds, 5.790 MB/s
SHA 23 MB took 1.001 seconds, 23.104 MB/s
SHA-256 26 MB took 1.000 seconds, 25.974 MB/s
SHA-512 275 KB took 1.063 seconds, 258.702 KB/s
HMAC-MD5 6 MB took 1.003 seconds, 5.722 MB/s
HMAC-SHA 22 MB took 1.001 seconds, 22.321 MB/s
HMAC-SHA256 25 MB took 1.000 seconds, 24.995 MB/s
HMAC-SHA512 275 KB took 1.079 seconds, 254.795 KB/s
PBKDF2 672 bytes took 1.005 seconds, 668.790 bytes/s
RSA 2048 public 12 ops took 1.015 sec, avg 84.608 ms, 11.819 ops/sec
RSA 2048 private 2 ops took 2.388 sec, avg 1194.200 ms, 0.837 ops/sec
DH 2048 key gen 4 ops took 1.268 sec, avg 316.900 ms, 3.156 ops/sec
DH 2048 agree 4 ops took 1.652 sec, avg 412.950 ms, 2.422 ops/sec
ECC 256 key gen 3 ops took 1.061 sec, avg 353.667 ms, 2.828 ops/sec
ECDHE 256 agree 4 ops took 1.407 sec, avg 351.725 ms, 2.843 ops/sec
ECDSA 256 sign 4 ops took 1.476 sec, avg 369.050 ms, 2.710 ops/sec
ECDSA 256 verify 2 ops took 1.451 sec, avg 725.450 ms, 1.378 ops/sec
CURVE 25519 key gen 3 ops took 1.412 sec, avg 470.500 ms, 2.125 ops/sec
CURVE 25519 agree 4 ops took 1.882 sec, avg 470.475 ms, 2.126 ops/sec
ED 25519 key gen 92 ops took 1.001 sec, avg 10.878 ms, 91.926 ops/sec
ED 25519 sign 64 ops took 1.007 sec, avg 15.733 ms, 63.561 ops/sec
ED 25519 verify 30 ops took 1.027 sec, avg 34.220 ms, 29.223 ops/sec
仅软件实现(不含 TSIP)
wolfCrypt Benchmark (block bytes 1024, min 1.0 sec each)
RNG 325 KB took 1.003 seconds, 323.899 KB/s
AES-128-CBC-enc 675 KB took 1.003 seconds, 672.780 KB/s
AES-128-CBC-dec 650 KB took 1.011 seconds, 643.182 KB/s
AES-192-CBC-enc 625 KB took 1.022 seconds, 611.426 KB/s
AES-192-CBC-dec 600 KB took 1.027 seconds, 584.511 KB/s
AES-256-CBC-enc 575 KB took 1.026 seconds, 560.320 KB/s
AES-256-CBC-dec 550 KB took 1.027 seconds, 535.749 KB/s
AES-128-GCM-enc 275 KB took 1.053 seconds, 261.258 KB/s
AES-128-GCM-dec 275 KB took 1.053 seconds, 261.258 KB/s
AES-192-GCM-enc 275 KB took 1.094 seconds, 251.348 KB/s
AES-192-GCM-dec 275 KB took 1.094 seconds, 251.279 KB/s
AES-256-GCM-enc 250 KB took 1.035 seconds, 241.569 KB/s
AES-256-GCM-dec 250 KB took 1.033 seconds, 242.084 KB/s
ARC4 3 MB took 1.001 seconds, 3.439 MB/s
RABBIT 4 MB took 1.000 seconds, 3.906 MB/s
3DES 200 KB took 1.086 seconds, 184.179 KB/s
MD5 6 MB took 1.000 seconds, 5.784 MB/s
SHA 2 MB took 1.006 seconds, 1.893 MB/s
SHA-256 750 KB took 1.023 seconds, 733.138 KB/s
SHA-512 275 KB took 1.029 seconds, 267.198 KB/s
HMAC-MD5 6 MB took 1.004 seconds, 5.715 MB/s
HMAC-SHA 2 MB took 1.002 seconds, 1.877 MB/s
HMAC-SHA256 750 KB took 1.032 seconds, 726.956 KB/s
HMAC-SHA512 275 KB took 1.045 seconds, 263.133 KB/s
PBKDF2 96 bytes took 1.054 seconds, 91.056 bytes/s
RSA 2048 public 12 ops took 1.024 sec, avg 85.292 ms, 11.724 ops/sec
RSA 2048 private 2 ops took 2.388 sec, avg 1193.750 ms, 0.838 ops/sec
DH 2048 key gen 4 ops took 1.265 sec, avg 316.275 ms, 3.162 ops/sec
DH 2048 agree 4 ops took 1.644 sec, avg 410.950 ms, 2.433 ops/sec
ECC 256 key gen 3 ops took 1.064 sec, avg 354.667 ms, 2.820 ops/sec
ECDHE 256 agree 4 ops took 1.435 sec, avg 358.725 ms, 2.788 ops/sec
ECDSA 256 sign 4 ops took 1.498 sec, avg 374.425 ms, 2.671 ops/sec
ECDSA 256 verify 2 ops took 1.437 sec, avg 718.300 ms, 1.392 ops/sec
CURVE 25519 key gen 2 ops took 1.005 sec, avg 502.350 ms, 1.991 ops/sec
CURVE 25519 agree 2 ops took 1.004 sec, avg 502.100 ms, 1.992 ops/sec
ED 25519 key gen 91 ops took 1.008 sec, avg 11.076 ms, 90.287 ops/sec
ED 25519 sign 66 ops took 1.026 sec, avg 15.538 ms, 64.359 ops/sec
ED 25519 verify 30 ops took 1.023 sec, avg 34.090 ms, 29.334 ops/sec
算法 软件加密 TSIP 硬件加速加密
RNG 231.160 KB/s 1.423 MB/s
SHA 1.239 MB/s 22.254 MB/s
SHA-256 515.565 KB/s 25.217 MB/s
密码套件 软件加密(秒) TSIP 硬件加速加密(秒)
TLS_RSA_WITH_AES_128_CBC_SHA 0.381 0.028
TLS_RSA_WITH_AES_128_CBC_SHA256 0.383 0.028
TLS_RSA_WITH_AES_256_CBC_SHA 0.382 0.030
TLS_RSA_WITH_AES_256_CBC_SHA256 0.385 0.029
关于在 Renesas 平台使用 wolfSSL 以及 wolfSSL 对 Renesas 的支持的更多信息,请参阅:https://www.wolfssl.com/docs/renesas/
返回页首
平台:
Renesas:Renesas RX72N(RTK5RX72N0C00000BJ)
基准测试:
使用硬件加速器(含 TSIP)
wolfCrypt Benchmark (block bytes 1024, min 1.0 sec each)
RNG 4 MB took 1.004 seconds, 3.818 MB/s
AES-128-CBC-enc 5 MB took 1.001 seconds, 4.852 MB/s
AES-128-CBC-dec 5 MB took 1.004 seconds, 4.692 MB/s
AES-256-CBC-enc 5 MB took 1.000 seconds, 4.688 MB/s
AES-256-CBC-dec 5 MB took 1.002 seconds, 4.534 MB/s
AES-128-GCM-enc 4 MB took 1.005 seconds, 4.348 MB/s
AES-128-GCM-dec 4 MB took 1.001 seconds, 4.292 MB/s
AES-256-GCM-enc 4 MB took 1.004 seconds, 4.257 MB/s
AES-256-GCM-dec 4 MB took 1.004 seconds, 4.206 MB/s
ARC4 7 MB took 1.001 seconds, 7.120 MB/s
RABBIT 9 MB took 1.000 seconds, 9.155 MB/s
3DES 525 KB took 1.035 seconds, 507.050 KB/s
MD5 15 MB took 1.001 seconds, 14.998 MB/s
SHA 30 MB took 1.001 seconds, 30.084 MB/s
SHA-256 35 MB took 1.000 seconds, 34.588 MB/s
SHA-512 750 KB took 1.001 seconds, 749.475 KB/s
HMAC-MD5 15 MB took 1.000 seconds, 14.844 MB/s
HMAC-SHA 29 MB took 1.000 seconds, 29.440 MB/s
HMAC-SHA256 34 MB took 1.000 seconds, 33.709 MB/s
HMAC-SHA512 750 KB took 1.016 seconds, 738.334 KB/s
PBKDF2 1 KB took 1.025 seconds, 1.219 KB/s
RSA 2048 public 26 ops took 1.020 sec, avg 39.212 ms, 25.503 ops/sec
RSA 2048 private 2 ops took 1.101 sec, avg 550.250 ms, 1.817 ops/sec
DH 2048 key gen 7 ops took 1.037 sec, avg 148.114 ms, 6.752 ops/sec
DH 2048 agree 6 ops took 1.184 sec, avg 197.250 ms, 5.070 ops/sec
ECC 256 key gen 7 ops took 1.069 sec, avg 152.743 ms, 6.547 ops/sec
ECDHE 256 agree 8 ops took 1.231 sec, avg 153.825 ms, 6.501 ops/sec
ECDSA 256 sign 8 ops took 1.271 sec, avg 158.825 ms, 6.296 ops/sec
ECDSA 256 verify 4 ops took 1.242 sec, avg 310.475 ms, 3.221 ops/sec
CURVE 25519 key gen 5 ops took 1.055 sec, avg 211.080 ms, 4.738 ops/sec
CURVE 25519 agree 6 ops took 1.266 sec, avg 211.050 ms, 4.738 ops/sec
ED 25519 key gen 208
开发板上启用硬件加速和 SP 数学库后的 TLS 客户端平均连接时间:
未使用 SP 数学库的纯软件速度: AES-128-CBC-enc 2 MB took 1.010 seconds, 1.910 MB/s SHA-256 4 MB took 1.000 seconds, 3.638 MB/s HMAC-SHA256 4 MB took 1.010 seconds, 3.626 MB/s RSA 2048 public 24 ops took 1.080 sec, avg 45.000 ms, 22.222 ops/sec RSA 2048 private 2 ops took 1.840 sec, avg 920.000 ms, 1.087 ops/sec ECC 256 key gen 3 ops took 1.170 sec, avg 390.000 ms, 2.564 ops/sec ECDHE 256 agree 4 ops took 1.560 sec, avg 390.000 ms, 2.564 ops/sec ECDSA 256 sign 4 ops took 1.570 sec, avg 392.500 ms, 2.548 ops/sec ECDSA 256 verify 2 ops took 1.560 sec, avg 780.000 ms, 1.282 ops/sec 启用硬件加速和 SP 数学库后的速度: AES-128-CBC-enc 3 MB took 1.010 seconds, 2.901 MB/s SHA-256 11 MB took 1.000 seconds, 11.353 MB/s HMAC-SHA256 11 MB took 1.000 seconds, 11.206 MB/s RSA 2048 public 94 ops took 1.010 sec, avg 10.745 ms, 93.069 ops/sec RSA 2048 private 4 ops took 1.530 sec, avg 382.500 ms, 2.614 ops/sec ECC 256 key gen 129 ops took 1.000 sec, avg 7.752 ms, 129.000 ops/sec ECDHE 256 agree 32 ops took 1.040 sec, avg 32.500 ms, 30.769 ops/sec ECDSA 256 sign 90 ops took 1.000 sec, avg 11.111 ms, 90.000 ops/sec ECDSA 256 verify 26 ops took 1.030 sec, avg 39.615 ms, 25.243 ops/sec Ed25519 和 Curve25519 测试结果: CURVE 25519 key gen 36 ops took 1.020 sec, avg 28.333 ms, 35.294 ops/sec CURVE 25519 agree 36 ops took 1.010 sec, avg 28.056 ms, 35.644 ops/sec ED 25519 key gen 99 ops took 1.000 sec, avg 10.101 ms, 99.000 ops/sec ED 25519 sign 82 ops took 1.000 sec, avg 12.195 ms, 82.000 ops/sec ED 25519 verify 32 ops took 1.010 sec, avg 31.562 ms, 31.683 ops/sec 关于在 Renesas 平台上使用 wolfSSL 以及 wolfSSL 对 Renesas 的支持详情,请参考:https://www.wolfssl.com/docs/renesas/ 返回页面顶部 平台: Xilinx ZCU102 Petalinux: 基准测试: RSA 评估:
仅使用 wolfSSL:
wolfSSL 加 ARMv8:
wolfSSL 硬件加速:
返回页面顶部 平台: Espressif ESP32 WROOM: Benchmark: 以下基准测试数据在 esp32-wroom-32 上使用 fastmath 库采集,CPU 频率 240MHz,ESP-IDF 版本为 v3.3-beta1-39-g6cb37ecc5(commit hash:6cb37ecc5),操作系统为 Ubuntu 18.04.1 LTS(Bionic Beaver)。软件加密: AES-128-CBC-enc 1 MB took 1.001 seconds, 1.146 MB/s AES-128-CBC-dec 1 MB took 1.017 seconds, 1.104 MB/s AES-192-CBC-enc 1 MB took 1.018 seconds, 1.055 MB/s AES-192-CBC-dec 1 MB took 1.006 seconds, 1.019 MB/s AES-256-CBC-enc 1000 KB took 1.000 seconds, 1000.000 KB/s AES-256-CBC-dec 975 KB took 1.007 seconds, 968.222 KB/s AES-128-GCM-enc 350 KB took 1.055 seconds, 331.754 KB/s AES-128-GCM-dec 350 KB took 1.054 seconds, 332.068 KB/s AES-192-GCM-enc 325 KB took 1.013 seconds, 320.829 KB/s AES-192-GCM-dec 325 KB took 1.013 seconds, 320.829 KB/s AES-256-GCM-enc 325 KB took 1.041 seconds, 312.200 KB/s AES-256-GCM-dec 325 KB took 1.041 seconds, 312.200 KB/s SHA 6 MB took 1.004 seconds, 5.714 MB/s SHA-256 2 MB took 1.006 seconds, 1.747 MB/s SHA-384 1 MB took 1.011 seconds, 1.159 MB/s SHA-512 1 MB took 1.009 seconds, 1.161 MB/s HMAC-SHA 6 MB took 1.001 seconds, 5.634 MB/s HMAC-SHA256 2 MB took 1.000 seconds, 1.733 MB/s HMAC-SHA384 1 MB took 1.004 seconds, 1.046 MB/s HMAC-SHA512 1 MB took 1.002 seconds, 1.048 MB/s RSA 2048 public 16 ops took 1.056 sec, avg 66.000 ms, 15.152 ops/sec RSA 2048 private 2 ops took 2.488 sec, avg 1244.000 ms, 0.804 ops/sec ECC 256 key gen 4 ops took 1.101 sec, avg 275.250 ms, 3.633 ops/sec ECDHE 256 agree 4 ops took 1.098 sec, avg 274.500 ms, 3.643 ops/sec ECDSA 256 sign 4 ops took 1.111 sec, avg 277.750 ms, 3.600 ops/sec ECDSA 256 verify 2 ops took 1.099 sec, avg 549.500 ms, 1.820 ops/sec 硬件加密: AES-128-CBC-enc 6 MB took 1.004 seconds, 5.958 MB/s AES-128-CBC-dec 5 MB took 1.002 seconds, 5.287 MB/s AES-192-CBC-enc 6 MB took 1.004 seconds, 5.958 MB/s AES-192-CBC-dec 5 MB took 1.002 seconds, 5.287 MB/s AES-256-CBC-enc 6 MB took 1.001 seconds, 5.951 MB/s AES-256-CBC-dec 5 MB took 1.004 seconds, 5.277 MB/s AES-128-GCM-enc 375 KB took 1.067 seconds, 351.453 KB/s AES-128-GCM-dec 375 KB took 1.067 seconds, 351.453 KB/s AES-192-GCM-enc 350 KB took 1.010 seconds, 346.535 KB/s AES-192-GCM-dec 350 KB took 1.009 seconds, 346.878 KB/s AES-256-GCM-enc 350 KB took 1.016 seconds, 344.488 KB/s AES-256-GCM-dec 350 KB took 1.016 seconds, 344.488 KB/s SHA 14 MB took 1.000 seconds, 14.062 MB/s SHA-256 15 MB took 1.000 seconds, 15.234 MB/s SHA-384 17 MB took 1.000 seconds, 17.383 MB/s SHA-512 18 MB took 1.001 seconds, 17.512 MB/s HMAC-SHA 14 MB took 1.000 seconds, 13.818 MB/s HMAC-SHA256 15 MB took 1.001 seconds, 14.951 MB/s HMAC-SHA384 17 MB took 1.001 seconds, 16.683 MB/s HMAC-SHA512 17 MB took 1.000 seconds, 16.943 MB/s RSA 2048 public 20 ops took 1.017 sec, avg 50.850 ms, 19.666 ops/sec RSA 2048 private 4 ops took 1.059 sec, avg 264.750 ms, 3.777 ops/sec ECC 256 key gen 4 ops took 1.092 sec, avg 273.000 ms, 3.663 ops/sec ECDHE 256 agree 4 ops took 1.089 sec, avg 272.250 ms, 3.673 ops/sec ECDSA 256 sign 4 ops took 1.101 sec, avg 275.250 ms, 3.633 ops/sec ECDSA 256 verify 2 ops took 1.092 sec, avg 546.000 ms, 1.832 ops/sec 返回页面顶部 平台: Intel x86_64: Benchmark: 近期版本的 wolfSSL 加入了针对 Intel x86_64 平台的新汇编代码,性能大幅提升。前两张图展示的是 AES-GCM 的性能提升。在这些基准测试图中,Y 轴表示数据块大小,X 轴表示相对于上一版 wolfSSL 的速度倍数。 AES-GCM 的汇编代码已重写,以充分发挥 AVX1 和 AVX2 指令的优势。目前 AES-GCM 的性能已不逊于甚至超越 OpenSSL。 ChaCha20-Poly1305 是一种较新的认证加密算法,设计初衷是作为 AES-GCM 的替代方案。该算法简单,在没有 AES 和 GCM 硬件加速的 CPU 上也能快速运行。 最常用的摘要算法是 SHA-256 和 SHA-384。随着 AES-GCM 在 TLS 中的引入,SHA-256 和 SHA-384 在应用数据认证中的使用有所减少,但它们仍用于握手消息认证、单向函数(伪随机数生成器所需)以及数字签名。 Curve25519 是 Montgomery 椭圆曲线的一组参数,安全性约为 128 位,用于密钥交换,凭借其速度优势及被纳入多项标准而广受欢迎。该算法已作为 TLS v1.3 的一部分,NIST 也在考虑将其纳入 SP 800-186。Ed25519 是 Twisted Edwards 曲线的一组参数,与 Curve25519 在数学上相关,具有相同的安全特性。基于 Twisted Edwards 曲线设计的新签名方案速度很快,也已纳入 TLS v1.3。此外还起草了一份规范草案,描述如何使用基于 Ed25519 的 EdDSA 数字证书。 RSA 是证书中最常用的公钥算法。执行 TLS 握手时,服务器会对迄今为止的消息哈希进行签名,客户端则验证证书链中各证书的签名,并用证书中的公钥验证消息哈希。签名和验证是握手中最耗时的操作。 DH 一直是握手中的首选密钥交换算法,但由于椭圆曲线变体在相同安全级别下速度要快得多,DH 正逐渐被淘汰。密钥交换是 TLS 握手中第二耗时的操作。 椭圆曲线密码学(ECC)是有限域(FF)密码学的替代方案,后者包括 RSA、DSA 和 DH 等算法。ECDSA 是 RSA 和 DSA 的椭圆曲线变体,ECDH 是 DH 的椭圆曲线变体。ECDSA 和 ECDH 可用于任何其有限域对应算法适用的场景。ECC 需要一条预定义的曲线来执行运算。最常用的曲线是 P-256,它具有 128 位强度,并被纳入多项标准,包括 TLS、IETF 证书标准以及 NIST 的 FIPS 186-4。浏览器和 Web 服务器越来越倾向于使用 ECDH 而非 DH,因为前者速度快得多。 如果您对在您的平台上使用 wolfSSL 嵌入式 TLS 库或其性能优化有任何疑问,请联系 support@wolfssl.com。 返回页面顶部 平台: NXP i.MX6 Benchmark: 以下基准测试数据在 NXP i.MX6 上运行 wolfSSL 基准测试应用、搭载 Green Hills INTEGRITY OS 采集而来。 返回页面顶部 平台: iPhone X: Benchmark: 基准测试在 Apple iPhone X 上进行,该设备搭载 Apple 新一代 A11 处理器。以下测试使用 ARMv8 加密扩展和单精度数学运算来展示性能。 使用我们的 ARMv8 加密汇编加速后,对称加密 AES 和 SHA 的性能如下: Algorithm Performance AES-128 CBC Encrypt 912.347 MB/s(36.58 倍) AES-128 CBC Decrypt 6,084.83 MB/s(256.15 倍) AES-128 GCM Encrypt 1,242.28 MB/s(193.65 倍) AES-128 GCM Decrypt 575.83 MB/s(90.26 倍) SHA-256 1,717.28 MB/s(56.11 倍) 通过 `./configure –enable-armasm` 或 `WOLFSSL_ARMASM` 宏可启用此特性。 使用我们的单精度数学加速后,非对称算法 RSA、DH 和 ECC 的性能如下: Algorithm Performance RSA 2048 public 1,211.27 ops/sec(1.50 倍) RSA 2048 private 32.59 ops/sec(1.18 倍) DH 2048 key gen 77.44 ops/sec(1.24 倍) DH 2048 key agree 77.45 ops/sec(1.29 倍) ECC 256 key gen 1670.65 ops/sec(8.67 倍) ECDHE 256 agree 396.88 ops/sec(2.05 倍) ECDSA 256 sign 1,212.33 ops/sec(6.42 倍) ECDSA 256 verify 331.02 ops/sec(2.38 倍) 通过 `./configure –enable-sp` 或 `WOLFSSL_HAVE_SP_RSA`、`WOLFSSL_HAVE_SP_DH` 和 `WOLFSSL_HAVE_SP_ECC` 宏可启用此特性。 TLS v1.2 按密码套件划分的性能基准测试结果如下: Algorithm Performance DHE-RSA-AES128-SHA256 CPS 22.5,读取 388 MB/s,写入 106 MB/s ECDHE-RSA-AES128-GCM-SHA256 CPS 26.2,读取 598 MB/s(RX),写入 125 MB/s ECDHE-ECDSA-AES128-GCM-SHA256 CPS 83.4,读取 504.8 MB/s,写入 92.2 MB/s 以上基准测试在 iPhone X 上使用单线程运行 wolfCrypt 和 wolfSSL 基准测试工具完成。
`X` = 相比默认软件实现的性能提升倍数。
`CPS` = 每秒连接数 参考:wolfCrypt/wolfSSL 在 iPhone 8/8 Plus/X (A11) 上的基准测试博客文章 返回页面顶部 平台: Microchip PIC32MZ: PIC32MZ,主频 200MHz 基准测试: 测试数据基于 PIC32MZ Ethernet Starter Kit,使用 wolfCrypt benchmark 应用,通过 MPLAB X 编译。 软件加密 硬件加密 AES-CBC 0.26 Mb/s 5.78 Mb/s AES-CTR 0.69 Mb/s 5.67 Mb/s 3DES 6.19 Mb/s 6.19 Mb/s MD5 6.22 Mb/s 16.84 Mb/s SHA-1 3.46 Mb/s 16.65 Mb/s SHA-256 1.678 Mb/s 15.84 Mb/s 参考: wolfSSL PIC32 支持 返回页面顶部 平台: STM32: STM32F439,主频 180MHz
STM32F756,主频 216MHz 基准测试: 测试数据基于 STMicroelectronics 评估板,使用 wolfCrypt benchmark 应用,通过 IAR EWARM 编译(优化等级:High/Size) 参考: wolfSSL STM32 支持 返回页面顶部 平台: STM32F777NI: ARM Cortex M7
216 MHz 使用 GCC ARM (arm-none-eabi-gcc),基于 OpenSTM32 (System Workbench) 裸机环境 基准测试: 加密基准测试: wolfSSL 软件加密,普通大整数数学库 RNG 3 MB 用时 1.000 秒 3.149 MB/s AES-Enc 6 MB 用时 1.000 秒 6.494 MB/s AES-Dec 7 MB 用时 1.000 秒 6.519 MB/s AES-GCM-Enc 3 MB 用时 1.004 秒 553 MB/s AES-GCM-Dec 3 MB 用时 1.004 秒 2.553 MB/s AES-CTR 7 MB 用时 1.000 秒 6.543 MB/s CHACHA 16 MB 用时 1.000 秒 15.723 MB/s CHA-POLY 10 MB 用时 1.000 秒 10.474 MB/s 3DES 1 MB 用时 1.008 秒 1.405 MB/s MD5 24 MB 用时 1.000 秒 24.243 MB/s POLY1305 42 MB 用时 1.000 秒 41.821 MB/s SHA 14 MB 用时 1.000 秒 14.380 MB/s SHA-224 8 MB 用时 1.000 秒 8.423 MB/s SHA-256 8 MB 用时 1.000 秒 8.423 MB/s SHA-384 2 MB 用时 1.000 秒 2.319 MB/s SHA-512 2 MB 用时 1.000 秒 2.319 MB/s STM32F7 硬件加密,普通大整数数学库 RNG 6 MB 用时 1.000 秒 6.030 MB/s AES-Enc 30 MB 用时 1.000 秒 30.396 MB/s AES-Dec 30 MB 用时 1.000 秒 30.371 MB/s AES-GCM-Enc 42 MB 用时 1.000 秒 42.261 MB/s AES-GCM-Dec 33 MB 用时 1.000 秒 32.861 MB/s AES-CTR 48 MB 用时 1.000 秒 47.827 MB/s CHACHA 16 MB 用时 1.000 秒 15.747 MB/s CHA-POLY 11 MB 用时 1.000 秒 10.522 MB/s 3DES 13 MB 用时 1.000 秒 12.988 MB/s MD5 41 MB 用时 1.000 秒 40.894 MB/s POLY1305 42 MB 用时 1.000 秒 41.846 MB/s SHA 38 MB 用时 1.004 秒 38.202 MB/s SHA-224 41 MB 用时 1.000 秒 41.309 MB/s SHA-256 39 MB 用时 1.000 秒 39.111 MB/s SHA-384 2 MB 用时 1.004 秒 2.310 MB/s SHA-512 2 MB 用时 1.004 秒 2.310 MB/s 参考: wolfSSL 与 STM32F7
STM32F77NI 返回页首 平台: Apple iPhone 8/8 Plus/X Apple A11 Bionic
2.39 GHz 六核 64 位 Benchmark: 在 Apple 最新 iPhone 8 / 8 Plus 和 iPhone X 所搭载的全新 A11 处理器上运行的基准测试。该处理器包含六个 ARMv8 CPU 核心(2 个 2.53GHz 高性能核心和 4 个 1.42GHz 高能效核心)。测试使用了 wolfSSL 针对 ARMv8 加密扩展和单精度数学运算的最新优化,充分展示了其卓越性能。使用 ARMv8 加密汇编优化后的对称算法 AES 和 SHA 表现如下: AES-128 CBC 加密:912.347 MB/s(36.58 倍) AES-128 CBC 解密:6,084.83 MB/s(256.15 倍) AES-128 GCM 加密:1,242.28 MB/s(193.65 倍) AES-128 GCM 解密:575.83 MB/s(90.26 倍) SHA-256:1,717.28 MB/s(56.11 倍) 该功能可通过 `./configure –enable-armasm` 或定义 `WOLFSSL_ARMASM` 启用。 使用单精度数学优化后的非对称算法 RSA、DH 和 ECC 表现如下: RSA 2048 公钥:1,211.27 ops/sec(1.50 倍) RSA 2048 私钥:32.59 ops/sec(1.18 倍) DH 2048 密钥生成:77.44 ops/sec(1.24 倍) DH 2048 密钥协商:77.45 ops/sec(1.29 倍) ECC 256 密钥生成:1670.65 ops/sec(8.67 倍) ECDHE 256 协商:396.88 ops/sec(2.05 倍) ECDSA 256 签名:1,212.33 ops/sec(6.42 倍) ECDSA 256 验证:331.02 ops/sec(2.38 倍) 该功能可通过 `./configure –enable-sp` 或定义 `WOLFSSL_HAVE_SP_RSA`、`WOLFSSL_HAVE_SP_DH` 和 `WOLFSSL_HAVE_SP_ECC` 启用。 以下是按密码套件统计的 TLS v1.2 性能测试结果: DHE-RSA-AES128-SHA256:CPS 22.5,读取 388 MB/s,写入 106 MB/s ECDHE-RSA-AES128-GCM-SHA256:CPS 26.2,读取 598 MB/s,写入 125 MB/s ECDHE-ECDSA-AES128-GCM-SHA256:CPS 83.4,读取 504.8 MB/s,写入 92.2 MB/s 测试在 iPhone X 上以单线程运行,使用 wolfCrypt 和 wolfSSL 自带的基准测试工具。
`X` = 相较于默认纯软件实现的性能提升倍数。
`CPS` = 每秒连接数 原始测试数据可在此下载:
https://www.wolfssl.com/files/benchmarks/iOSiPhoneXA11Benchmarks.pdf 参考
wolfSSL 博客文章 返回页首 平台: Intel SGX Intel(R) Core(TM) i5-6300U @ 2.4GHz Intel Compiler Benchmark: 以下是在 Intel SGX enclave 中运行 wolfCrypt 的基准测试结果。数据显示,在 SGX 内运行 wolfCrypt 加密操作对性能的影响极小。 非 SGX 环境测试结果 SHA-256 0.264s 189.078 MB/s
AES-GCM 3.142s 15.911 MB/s
RSA (2048) 加密 0.238 ms
RSA (2048) 解密 6.239 ms SGX 基准测试,enclave 内部 SHA-256 0.263s 190.436 MB/s -0.38% 差异
AES-GCM 3.128s 15.985 MB/s -0.45% 差异
RSA (2048) 加密 0.245 ms 2.94% 差异
RSA (2048) 解密 6.242 ms 0.05% 差异
参考资料 Intel Software Guard Extensions (SGX) SDK
返回页首 平台: Intel QuickAssist
DH895xCC Intel(R) Core(TM) i7-4790 CPU 3.60GHz 12GB 内存 基准测试: 加密基准测试: 使用 wolfSSL 的异步基准测试程序,在用户空间运行多线程并绑定 CPU 亲和性: RSA 2048 公钥:209,909 次/秒
RSA 2048 私钥:41,999 次/秒
DH 2048 密钥生成:112,491 次/秒
DH 2048 密钥协商:95,129 次/秒
ECDHE 256 协商:55,117 次/秒
ECDSA 256 签名:46,798 次/秒
ECDSA 256 验证:28,917 次/秒
AES-CBC 加密:2,932 MB/s
AES-CBC 解密:2,882 MB/s
AES-GCM:2,903 MB/s
3DES:1,511 MB/s
MD5:2,309 MB/s
SHA:5,068 MB/s
SHA-224:2,392 MB/s
SHA-256:1,275 MB/s
SHA-384:2,020 MB/s
SHA-512:1,908 MB/s 参考资料 wolfSSL 博客文章
Intel QuickAssist 返回页首 平台: HiKey LeMaker (ARMv8)
Kirin 620 SoC
ARM® Cortex™-A53 八核 64 位,主频最高 1.2GHz(ARM v8 指令集) 8GB eMMC 存储
1GB 内存 基准测试: 下面是硬件加速与普通软件加密对比的完整基准测试数据:
算法 软件加密 硬件加密 RNG 16.761 MB/s 82.599 MB/s AES-128-CBC 加密 26.491 MB/s 649.179 MB/s AES-128-CBC 解密 26.915 MB/s 607.407 MB/s AES-192-CBC 加密 22.796 MB/s 566.717 MB/s AES-192-CBC 解密 23.130 MB/s 553.092 MB/s AES-256-CBC 加密 20.004 MB/s 504.143 MB/s AES-256-CBC 解密 20.207 MB/s 491.374 MB/s AES-128-GCM 加密 6.224 MB/s 393.407 MB/s AES-128-GCM 解密 6.226 MB/s 182.279 MB/s AES-192-GCM 加密 5.895 MB/s 361.801 MB/s AES-192-GCM 解密 5.895 MB/s 175.676 MB/s AES-256-GCM 加密 5.609 MB/s 333.911 MB/s AES-256-GCM 解密 5.610 MB/s 169.085 MB/s CHACHA 60.510 MB/s 60.017 MB/s CHA-POLY 41.805 MB/s 41.410 MB/s MD5 156.310 MB/s 154.421 MB/s POLY1305 144.464 MB/s 143.058 MB/s SHA 89.874 MB/s 89.154 MB/s SHA-256 38.805 MB/s 533.139 MB/s HMAC-MD5 156.301 MB/s 154.083 MB/s HMAC-SHA 89.859 MB/s 89.045 MB/s HMAC-SHA256 38.814 MB/s 532.316 MB/s RSA,2048 位,公钥 171.995 次/秒 171.355 次/秒 RSA,2048 位,私钥 13.716 次/秒 13.686 次/秒 DH,2048 位,密钥生成 50.831 次/秒 50.575 次/秒 DH,2048 位,协商 41.826 次/秒 41.596 次/秒 返回页首 平台: AVX1:1.8GHz,Intel Core i5
AVX2:Intel Broadwell 基准测试: 加密基准测试: AVX2: SHA-256 处理 50 MB 用时 0.320 秒,156.118 MB/s
每字节周期数 = 9.75 = 47% AVX1: SHA-256 处理 50 MB 用时 0.272 秒,184.068 MB/s
每字节周期数 = 11.89 = 39% 普通模式:SHA-256 处理 50 MB 用时 0.376 秒,132.985 MB/s
每字节周期数 = 16.46 AVX2: SHA-384 处理 50 MB 用时 0.226 秒,221.318 MB/s
每字节周期数 = 6.88 = 42% AVX1: SHA-384 处理 50 MB 用时 0.192 秒,260.975 MB/s
每字节周期数 = 8.39 = 9% 普通模式:SHA-384 处理 50 MB 用时 0.209 秒,239.743 MB/s
每字节周期数 = 9.13 AVX2: SHA-512 处理 50 MB 用时 0.224 秒,223.120 MB/s
每字节周期数 = 6.82 = 75% AVX1: SHA-512 处理 50 MB 用时 0.188 秒,266.126 MB/s
每字节周期数 = 8.22 = 50% 普通模式:SHA-512 处理 50 MB 用时 0.281 秒,177.997 MB/s
每字节周期数 = 12.29 参考资料 wolfSSL 博客文章 返回页首 平台: Atmel SAMD21
ARM Cortex M0
48 MHz 基准测试: TLS 建立时间: 硬件加速 ATECC508A:平均 2.342 秒
纯软件实现:平均 13.422 秒 使用 ATECC508A 后,TLS 连接建立速度快了 5.73 倍。 纯软件实现(SAMD21 48MHz Cortex-M0,Fast Math TFM-ASM): ECC 256 密钥生成 3123.000 毫秒,5 次迭代平均值
EC-DHE 密钥协商 3117.000 毫秒,5 次迭代平均值
EC-DSA 签名耗时 1997.000 毫秒,5 次迭代平均值
EC-DSA 验签耗时 5057.000 毫秒,5 次迭代平均值 ATECC508A 硬件加速实现: ECC 256 密钥生成 144.400 毫秒,5 次迭代平均值
EC-DHE 密钥协商 134.200 毫秒,5 次迭代平均值
EC-DSA 签名耗时 293.400 毫秒,5 次迭代平均值
EC-DSA 验签耗时 208.400 毫秒,5 次迭代平均值 供参考,RNG、AES、MD5、SHA 和 SHA256 的基准测试结果如下: RNG 25 kB 耗时 0.784 秒,0.031 MB/s(来自 ATECC508A)
AES 25 kB 耗时 0.177 秒,0.138 MB/s
MD5 25 kB 耗时 0.050 秒,0.488 MB/s
SHA 25 kB 耗时 0.141 秒,0.173 MB/s
SHA-256 25 kB 耗时 0.352 秒,0.069 MB/s 参考资料 Microchip 硬件 TLS 平台
wolfSSL Microchip ATECC508A 概览与示例 返回页首 平台: STM32F221G-EVAL
ARM Cortex M3
120MHz
1 MB FLASH
128 KB SRAM 基准测试: 加密基准测试: 软件加密:wolfCrypt Benchmark,普通大整数运算库 AES 1024 kB 耗时 0.822 秒,1.22 MB/s
ARC4 1024 KB 耗时 0.219 秒,4.57 MB/s
DES 1024 KB 耗时 1.513 秒,0.66 MB/s
3DES 1024 KB 耗时 3.986 秒,0.25 MB/s MD5 1024 KB 耗时 0.119 秒,8.40 MB/s
SHA 1024 KB 耗时 0.279 秒,3.58 MB/s
SHA-256 1024 KB 耗时 0.690 秒,1.45 MB/s RSA 2048 加密耗时 111.17 毫秒,100 次迭代平均值
RSA 2048 解密耗时 1204.77 毫秒,100 次迭代平均值
DH 2048 密钥生成耗时 467.90 毫秒,100 次迭代平均值
DH 2048 密钥协商耗时 538.94 毫秒,100 次迭代平均值 STM32F2 硬件加密:wolfCrypt Benchmark,普通大整数运算库 AES 1024 kB 耗时 0.105 秒,9.52 MB/s
ARC4 1024 KB 耗时 0.219 秒,4.57 MB/s
DES 1024 KB 耗时 0.125 秒,8.00 MB/s
3DES 1024 KB 耗时 0.141 秒,7.09 MB/s MD5 1024 KB 耗时 0.045 秒,22.22 MB/s
SHA 1024 KB 耗时 0.047 秒,21.28 MB/s
SHA-256 1024 KB 耗时 0.690 秒,1.45 MB/s RSA 2048 加密耗时 111.09 毫秒(100 次迭代平均值)
RSA 2048 解密耗时 1204.88 毫秒(100 次迭代平均值)
DH 2048 密钥生成 467.56 毫秒(100 次迭代平均值)
DH 2048 密钥协商 542.11 毫秒(100 次迭代平均值) 参考 wolfSSL 与 STM32 返回页首 平台: Texas Instruments
Tiva C Series TM4C1294XL Connected Launchpad (www.ti.com)
ARM Cortex-M4
120 MHz
1 MB FLASH
256 KB SRAM
6 KB EEPROM 基准测试: 加密算法基准测试: AES 25 kB 耗时 0.038 秒,0.642 MB/s
Camellia 25 kB 耗时 0.032 秒,0.763 MB/s
ARC4 25 kB 耗时 0.006 秒,4.069 MB/s
RABBIT 25 kB 耗时 0.005 秒,4.883 MB/s
CHACHA 25 kB 耗时 0.007 秒,3.488 MB/s
3DES 25 kB 耗时 0.164 秒,0.149 MB/s MD5 25 kB 耗时 0.003 秒,8.138 MB/s
POLY1305 25 kB 耗时 0.004 秒,6.104 MB/s
SHA 25 kB 耗时 0.006 秒,4.069 MB/s
SHA-256 25 kB 耗时 0.014 秒,1.744 MB/s
SHA-512 25 kB 耗时 0.042 秒,0.581 MB/s RSA 2048 加密耗时 88.000 毫秒(1 次迭代平均值)
RSA 2048 解密耗时 1456.000 毫秒(1 次迭代平均值)
DH 2048 密钥生成 661.000 毫秒(1 次迭代平均值)
DH 2048 密钥协商 665.000 毫秒(1 次迭代平均值) ECC 256 密钥生成 130.400 毫秒(5 次迭代平均值)
EC-DHE 密钥协商 118.000 毫秒(5 次迭代平均值)
EC-DSA 签名耗时 136.800 毫秒(5 次迭代平均值)
EC-DSA 验证耗时 253.800 毫秒(5 次迭代平均值) 参考 CyaSSL 与 TI-RTOS 返回页首 平台: Freescale TWR-K70F120M (www.freescale.com)
Freescale Kinetis K70
120 MHz
2 GB FLASH
1 GB RAM 基准测试: 加密算法基准测试: AES 5120 kB 耗时 9.059 秒,0.55 MB/s
ARC4 5120 kB 耗时 2.190 秒,2.28 MB/s
DES 5120 kB 耗时 18.453 秒,0.27 MB/s MD5 5120 kB 耗时 1.396 秒,3.58 MB/s
SHA 5120 kB 耗时 3.635 秒,1.38 MB/s
SHA-256 5120 kB 耗时 9.145 秒,0.55 MB/s RSA 2048 加密耗时 73.99 毫秒(100 次迭代平均值)
RSA 2048 解密耗时 1359.09 毫秒(100 次迭代平均值)
DH 2048 密钥生成耗时 536.75 毫秒(100 次迭代平均值)
DH 2048 密钥协商耗时 540.99 毫秒(100 次迭代平均值) 构建细节 MQX RTOS,使用 fastmath 库并启用 TFM_TIMING_RESISTANT 在/cyassl/ctaocrypt/settings.h 中定义 FREESCALE_MQX
CodeWarrior 10.2 IDE 及编译器,针对速度优化
参考
Freescale TWR-K70F120M:http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=TWR-K70F120M
返回页面顶部
平台:
Apple TV 2
(www.apple.com)
Apple A4 (ARM Cortex-A8)
1 GHz
8 GB FLASH
256 MB RAM 基准测试: 加密基准测试: AES 5 MB 耗时 0.500 秒,9.99 MB/s
ARC4 5 MB 耗时 0.174 秒,28.66 MB/s
RABBIT 5 MB 耗时 0.126 秒,39.56 MB/s
3DES 5 MB 耗时 2.196 秒,2.28 MB/s MD5 5 MB 耗时 0.163 秒,30.73 MB/s
SHA 5 MB 耗时 0.137 秒,36.61 MB/s
SHA-256 5 MB 耗时 0.309 秒,16.20 MB/s RSA 1024 加密耗时 1.12 毫秒(100 次迭代平均值)
RSA 1024 解密耗时 17.81 毫秒(100 次迭代平均值)
DH 1024 密钥生成耗时 11.90 毫秒(100 次迭代平均值)
DH 1024 密钥协商耗时 11.22 毫秒(100 次迭代平均值) 构建细节 完整构建,使用 fastmath 编译(–enable-fastmath) 参考 博客文章:Running CyaSSL on the Apple TV 2 返回页面顶部 平台:
(www.mbed.org)
ARM Cortex-M3
96 MHz
512 kB FLASH
32 kB RAM 基准测试: 内存占用: RAM 占用:2.0 kB
Flash 占用*:64 kB
* 包含约 3 kB 的测试驱动代码。 加密基准测试: RSA 公钥运算:10 毫秒
RSA 私钥运算:165 毫秒 构建细节 – 完整构建,除 SHA-512、DH、DSA 和 HC-128 外全部包含
– 使用 mbed 云编译器编译 参考 http://mbed.org/users/toddouska/libraries/CyaSSL/lm43pv
http://mbed.org/users/toddouska/programs/cyassl-client/lm394s 返回页首 各密码算法性能对比 尽管具体密码算法的性能取决于主机平台,下图展示了 wolfCrypt 部分算法之间的相对性能。测试在 Macbook Pro(OS X 10.6.8,2.2 GHz Intel Core i7)上进行。 返回页首 如果你只想使用部分密码算法,可以自定义 wolfSSL 在建立 SSL/TLS 连接时使用的密码套件和/或算法。例如,要强制使用 128 位 AES,在调用 wolfSSL_CTX_new(SSL_CTX_new)之后添加如下代码: wolfSSL_CTX_set_cipher_list(ctx, “AES128-SHA”); 返回页首 后量子 Kyber 性能测试(Linux) 平台: 11th Gen Intel® Core™ i7-1185G7 @ 3.00GHz × 8 测试结果: 说明: – 仅使用 1 个核心 – 列出传统算法是为了便于对比 – 所用的 wolfSSL 配置为:
./configure --disable-psk --disable-shared --enable-intelasm --enable-aesni --enable-sp --enable-sp-math --enable-sp-asm --enable-kyber=wolfssl,all 'CFLAGS=-Os -DECC_USER_CURVES -DHAVE_ECC256 -DHAVE_ECC384' wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP384R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 953.534 ms Tx Total : 6.757 ms Rx : 2.753 MB/s Tx : 388.471 MB/s Connect : 55.076 ms Connect Avg : 2.503 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP384R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 956.304 ms Tx Total : 5.627 ms Rx : 2.745 MB/s Tx : 466.488 MB/s Connect : 54.827 ms Connect Avg : 2.492 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP521R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 948.864 ms Tx Total : 6.542 ms Rx : 2.766 MB/s Tx : 401.226 MB/s Connect : 59.490 ms Connect Avg : 2.704 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP521R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 951.101 ms Tx Total : 5.791 ms Rx : 2.760 MB/s Tx : 453.275 MB/s Connect : 59.598 ms Connect Avg : 2.709 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL1): Total : 5505024 bytes Num Conns : 22 Rx Total : 954.452 ms Tx Total : 7.089 ms Rx : 2.750 MB/s Tx : 370.297 MB/s Connect : 47.349 ms Connect Avg : 2.152 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL1): Total : 5505024 bytes Num Conns : 22 Rx Total : 956.931 ms Tx Total : 6.189 ms Rx : 2.743 MB/s Tx : 424.149 MB/s Connect : 46.129 ms Connect Avg : 2.097 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL3): Total : 5767168 bytes Num Conns : 23 Rx Total : 958.702 ms Tx Total : 7.431 ms Rx : 2.868 MB/s Tx : 370.070 MB/s Connect : 48.699 ms Connect Avg : 2.117 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL3): Total : 5767168 bytes Num Conns : 23 Rx Total : 961.281 ms Tx Total : 6.032 ms Rx : 2.861 MB/s Tx : 455.903 MB/s Connect : 47.497 ms Connect Avg : 2.065 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL5): Total : 5767168 bytes Num Conns : 23 Rx Total : 967.680 ms Tx Total : 6.709 ms Rx : 2.842 MB/s Tx : 409.906 MB/s Connect : 51.945 ms Connect Avg : 2.258 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL5): Total : 5767168 bytes Num Conns : 23 Rx Total : 969.719 ms Tx Total : 6.183 ms Rx : 2.836 MB/s Tx : 444.774 MB/s Connect : 49.456 ms Connect Avg : 2.150 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P256_KYBER_LEVEL1): Total : 2883584 bytes Num Conns : 12 Rx Total : 494.597 ms Tx Total : 2.225 ms Rx : 2.780 MB/s Tx : 618.065 MB/s Connect : 540.056 ms Connect Avg : 45.005 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P256_KYBER_LEVEL1): Total : 2883584 bytes Num Conns : 12 Rx Total : 495.337 ms Tx Total : 2.282 ms Rx : 2.776 MB/s Tx : 602.567 MB/s Connect : 538.472 ms Connect Avg : 44.873 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P384_KYBER_LEVEL3): Total : 2883584 bytes Num Conns : 12 Rx Total : 503.707 ms Tx Total : 3.427 ms Rx : 2.730 MB/s Tx : 401.222 MB/s Connect : 526.276 ms Connect Avg : 43.856 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P384_KYBER_LEVEL3): Total : 2883584 bytes Num Conns : 12 Rx Total : 505.002 ms Tx Total : 3.302 ms Rx : 2.723 MB/s Tx : 416.462 MB/s Connect : 523.324 ms Connect Avg : 43.610 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P521_KYBER_LEVEL5): Total : 2883584 bytes Num Conns : 12 Rx Total : 503.258 ms Tx Total : 3.171 ms Rx : 2.732 MB/s Tx : 433.556 MB/s Connect : 537.391 ms Connect Avg : 44.783 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P521_KYBER_LEVEL5): Total : 2883584 bytes Num Conns : 12 Rx Total : 504.367 ms Tx Total : 3.316 ms Rx : 2.726 MB/s Tx : 414.636 MB/s Connect : 532.820 ms Connect Avg : 44.402 ms 返回页首 后量子 Kyber 基准测试(ARM Cortex-M4) 平台: STM NUCLEO-F446ZE 基准测试: 说明: – 项目中的 HCLK 设置为 168MHz – 仅使用 1 个核心
– wolfSSL 数学配置设为 "Single Precision ASM Cortex-M3+ Math"
– 优化选项:-Ofast
– 为便于对比,保留了传统算法 RSA 2048 public 82 ops took 1.020 sec, avg 12.439 ms, 80.392 ops/sec RSA 2048 private 4 ops took 1.827 sec, avg 456.750 ms, 2.189 ops/sec DH 2048 key gen 5 ops took 1.181 sec, avg 236.200 ms, 4.234 ops/sec DH 2048 agree 6 ops took 1.419 sec, avg 236.500 ms, 4.228 ops/sec ECC SECP256R1 key gen 118 ops took 1.012 sec, avg 8.576 ms, 116.601 ops/sec ECDHE SECP256R1 agree 56 ops took 1.016 sec, avg 18.143 ms, 55.118 ops/sec KYBER512 128 key gen 232 ops took 1.004 sec, avg 4.328 ms, 231.076 ops/sec KYBER512 128 encap 192 ops took 1.008 sec, avg 5.250 ms, 190.476 ops/sec KYBER512 128 decap 178 ops took 1.004 sec, avg 5.640 ms, 177.291 ops/sec KYBER768 192 key gen 146 ops took 1.008 sec, avg 6.904 ms, 144.841 ops/sec KYBER768 192 encap 118 ops took 1.008 sec, avg 8.542 ms, 117.063 ops/sec KYBER768 192 decap 110 ops took 1.000 sec, avg 9.091 ms, 110.000 ops/sec KYBER1024 256 key gen 92 ops took 1.011 sec, avg 10.989 ms, 90.999 ops/sec KYBER1024 256 encap 76 ops took 1.000 sec, avg 13.158 ms, 76.000 ops/sec KYBER1024 256 decap 72 ops took 1.000 sec, avg 13.889 ms, 72.000 ops/sec 返回页首 后量子 Kyber 基准测试(MacOS) 平台: Apple MacBook Pro 18,3,搭载 Apple M1 Pro 处理器,主频 3.09 GHz 基准测试: 说明: – 仅使用 1 个核心 数学库: 多精度:Wolf(SP) no-dyn-stack word-size=64 bits=4096 sp_int.c 单精度:ecc 256 384 521 rsa/dh 2048 3072 4096 asm sp_arm64.c wolfCrypt 基准测试(block bytes 1048576,每项至少运行 1.0 秒) DH 2048 key gen 3997 ops took 1.000 sec, avg 0.250 ms, 3996.812 ops/sec DH 2048 agree 4100 ops took 1.001 sec, avg 0.244 ms, 4097.522 ops/sec KYBER512 128 key gen 96100 ops took 1.001 sec, avg 0.010 ms, 96037.765 ops/sec KYBER512 128 encap 78000 ops took 1.000 sec, avg 0.013 ms, 77970.220 ops/sec KYBER512 128 decap 58900 ops took 1.001 sec, avg 0.017 ms, 58867.158 ops/sec KYBER768 192 key gen 58200 ops took 1.000 sec, avg 0.017 ms, 58192.314 ops/sec KYBER768 192 encap 48700 ops took 1.001 sec, avg 0.021 ms, 48664.334 ops/sec KYBER768 192 decap 38100 ops took 1.001 sec, avg 0.026 ms, 38059.656 ops/sec KYBER1024 256 key gen 37800 ops took 1.003 sec, avg 0.027 ms, 37704.299 ops/sec KYBER1024 256 encap 32600 ops took 1.001 sec, avg 0.031 ms, 32566.427 ops/sec KYBER1024 256 decap 26000 ops took 1.001 sec, avg 0.039 ms, 25967.020 ops/sec ECC [ SECP256R1] 256 key gen 84100 ops took 1.001 sec, avg 0.012 ms, 84013.469 ops/sec ECDHE [ SECP256R1] 256 agree 24400 ops took 1.004 sec, avg 0.041 ms, 24300.995 ops/sec 返回页首 平台: Raspberry Pi Pico-W
ARM Cortext M0+,125MHz 基准测试: wolfSSL/wolfCrypt v5.6.3 配置选项 – SP Math: 启用:WOLFSSL_HAVE_SP_RSA、WOLFSSL_HAVE_SP_DH、WOLFSSL_HAVE_SP_ECC、WOLFSSL_SP_ARM_THUMB_ASM 禁用:WOLFSSL_SP_SMALL、WOLFSSL_SP_ASM – AES-GCM: 启用:GCM_TABLE_4BIT 禁用:GCM_SMALL ``` (基准测试结果保持原样) ``` 返回页首 平台: Altera Agilex 5: 硬件:Terasic DE-25 Nano(Agilex 5 A5EB013BB23BE4SR1)。 主机系统:Linux 6.12.11-altera 分支版本。 CPU:Cortex-A55 核心(2 个 @1.25 GHz)和 Cortex-A76 核心(2 个 @1.40 GHz)。 基准测试软件:wolfSSL 5.9.1 benchmark.c,使用 GCC + O2 编译。 数据块大小:1048576 每个算法耗时:约 1.0 秒 基准测试结果:
返回页首 基准测试说明 处理器的原生寄存器位宽(32 位 vs 64 位)在执行 1000+ 位公钥运算时会产生显著影响。 fastmath(–enable-fastmath)可以降低动态内存占用,并加快公钥运算速度。如果在 32 位平台上开启 fastmath 编译遇到问题,可以禁用共享库,避免 PIC 占用寄存器(另见 README 中的说明): ./configure –enable-fastmath –disable-shared
make clean
make *注意:切换 configure 选项后,建议对 wolfSSL 执行一次 "make clean" 默认情况下,fastmath 会尽可能使用汇编优化。如果汇编优化不可用,也可以在编译 wolfSSL 时向 CFLAGS 添加 TFM_NO_ASM,在没有汇编优化的情况下使用 fastmath: ./configure –enable-fastmath CFLAGS=-DTFM_NO_ASM 对于非嵌入式平台的用户,可以尝试使用 fasthugemath 进一步压榨 fastmath 的性能: ./configure –enable-fasthugemath wolfSSL 默认构建在内存占用与性能之间做了较好的平衡。如果你更关注其中某一方面,请参阅 wolfSSL 手册第 2 章了解更多配置选项。 大块数据传输:wolfSSL 默认使用 128 字节的 I/O 缓冲区,因为约 80% 的 SSL 流量都在这个范围内,同时也能限制动态内存使用。如果需要大块数据传输,可以配置使用 16K 缓冲区(SSL 的最大值)。 出版物与宣传资料 与我们的 SSL/TLS 及加密库基准测试相关的出版物:
wolfSSL+NTRU:高性能 SSL 这份宣传资料介绍了 wolfSSL 嵌入式 SSL 库搭配 Security Innovation 的 NTRU 密码算法所能获得的性能提升。NTRU 与 RSA 公钥算法类似,但速度可以快 20 到 200 倍。 下载宣传资料
wolfSSL 安全 memcached 基准测试 wolfSSL 加密速度快、内存占用低,可以轻松部署在支持数千连接的大规模服务器上。这份宣传资料展示了使用 wolfSSL 的 memcached 基准测试结果。 下载宣传资料
开发板上启用硬件加速和 SP 数学库后的 TLS 客户端平均连接时间:
未使用 SP 数学库的纯软件速度: AES-128-CBC-enc 2 MB took 1.010 seconds, 1.910 MB/s SHA-256 4 MB took 1.000 seconds, 3.638 MB/s HMAC-SHA256 4 MB took 1.010 seconds, 3.626 MB/s RSA 2048 public 24 ops took 1.080 sec, avg 45.000 ms, 22.222 ops/sec RSA 2048 private 2 ops took 1.840 sec, avg 920.000 ms, 1.087 ops/sec ECC 256 key gen 3 ops took 1.170 sec, avg 390.000 ms, 2.564 ops/sec ECDHE 256 agree 4 ops took 1.560 sec, avg 390.000 ms, 2.564 ops/sec ECDSA 256 sign 4 ops took 1.570 sec, avg 392.500 ms, 2.548 ops/sec ECDSA 256 verify 2 ops took 1.560 sec, avg 780.000 ms, 1.282 ops/sec 启用硬件加速和 SP 数学库后的速度: AES-128-CBC-enc 3 MB took 1.010 seconds, 2.901 MB/s SHA-256 11 MB took 1.000 seconds, 11.353 MB/s HMAC-SHA256 11 MB took 1.000 seconds, 11.206 MB/s RSA 2048 public 94 ops took 1.010 sec, avg 10.745 ms, 93.069 ops/sec RSA 2048 private 4 ops took 1.530 sec, avg 382.500 ms, 2.614 ops/sec ECC 256 key gen 129 ops took 1.000 sec, avg 7.752 ms, 129.000 ops/sec ECDHE 256 agree 32 ops took 1.040 sec, avg 32.500 ms, 30.769 ops/sec ECDSA 256 sign 90 ops took 1.000 sec, avg 11.111 ms, 90.000 ops/sec ECDSA 256 verify 26 ops took 1.030 sec, avg 39.615 ms, 25.243 ops/sec Ed25519 和 Curve25519 测试结果: CURVE 25519 key gen 36 ops took 1.020 sec, avg 28.333 ms, 35.294 ops/sec CURVE 25519 agree 36 ops took 1.010 sec, avg 28.056 ms, 35.644 ops/sec ED 25519 key gen 99 ops took 1.000 sec, avg 10.101 ms, 99.000 ops/sec ED 25519 sign 82 ops took 1.000 sec, avg 12.195 ms, 82.000 ops/sec ED 25519 verify 32 ops took 1.010 sec, avg 31.562 ms, 31.683 ops/sec 关于在 Renesas 平台上使用 wolfSSL 以及 wolfSSL 对 Renesas 的支持详情,请参考:https://www.wolfssl.com/docs/renesas/ 返回页面顶部 平台: Xilinx ZCU102 Petalinux: 基准测试: RSA 评估:
仅使用 wolfSSL:
wolfSSL 加 ARMv8:
wolfSSL 硬件加速:
返回页面顶部 平台: Espressif ESP32 WROOM: Benchmark: 以下基准测试数据在 esp32-wroom-32 上使用 fastmath 库采集,CPU 频率 240MHz,ESP-IDF 版本为 v3.3-beta1-39-g6cb37ecc5(commit hash:6cb37ecc5),操作系统为 Ubuntu 18.04.1 LTS(Bionic Beaver)。软件加密: AES-128-CBC-enc 1 MB took 1.001 seconds, 1.146 MB/s AES-128-CBC-dec 1 MB took 1.017 seconds, 1.104 MB/s AES-192-CBC-enc 1 MB took 1.018 seconds, 1.055 MB/s AES-192-CBC-dec 1 MB took 1.006 seconds, 1.019 MB/s AES-256-CBC-enc 1000 KB took 1.000 seconds, 1000.000 KB/s AES-256-CBC-dec 975 KB took 1.007 seconds, 968.222 KB/s AES-128-GCM-enc 350 KB took 1.055 seconds, 331.754 KB/s AES-128-GCM-dec 350 KB took 1.054 seconds, 332.068 KB/s AES-192-GCM-enc 325 KB took 1.013 seconds, 320.829 KB/s AES-192-GCM-dec 325 KB took 1.013 seconds, 320.829 KB/s AES-256-GCM-enc 325 KB took 1.041 seconds, 312.200 KB/s AES-256-GCM-dec 325 KB took 1.041 seconds, 312.200 KB/s SHA 6 MB took 1.004 seconds, 5.714 MB/s SHA-256 2 MB took 1.006 seconds, 1.747 MB/s SHA-384 1 MB took 1.011 seconds, 1.159 MB/s SHA-512 1 MB took 1.009 seconds, 1.161 MB/s HMAC-SHA 6 MB took 1.001 seconds, 5.634 MB/s HMAC-SHA256 2 MB took 1.000 seconds, 1.733 MB/s HMAC-SHA384 1 MB took 1.004 seconds, 1.046 MB/s HMAC-SHA512 1 MB took 1.002 seconds, 1.048 MB/s RSA 2048 public 16 ops took 1.056 sec, avg 66.000 ms, 15.152 ops/sec RSA 2048 private 2 ops took 2.488 sec, avg 1244.000 ms, 0.804 ops/sec ECC 256 key gen 4 ops took 1.101 sec, avg 275.250 ms, 3.633 ops/sec ECDHE 256 agree 4 ops took 1.098 sec, avg 274.500 ms, 3.643 ops/sec ECDSA 256 sign 4 ops took 1.111 sec, avg 277.750 ms, 3.600 ops/sec ECDSA 256 verify 2 ops took 1.099 sec, avg 549.500 ms, 1.820 ops/sec 硬件加密: AES-128-CBC-enc 6 MB took 1.004 seconds, 5.958 MB/s AES-128-CBC-dec 5 MB took 1.002 seconds, 5.287 MB/s AES-192-CBC-enc 6 MB took 1.004 seconds, 5.958 MB/s AES-192-CBC-dec 5 MB took 1.002 seconds, 5.287 MB/s AES-256-CBC-enc 6 MB took 1.001 seconds, 5.951 MB/s AES-256-CBC-dec 5 MB took 1.004 seconds, 5.277 MB/s AES-128-GCM-enc 375 KB took 1.067 seconds, 351.453 KB/s AES-128-GCM-dec 375 KB took 1.067 seconds, 351.453 KB/s AES-192-GCM-enc 350 KB took 1.010 seconds, 346.535 KB/s AES-192-GCM-dec 350 KB took 1.009 seconds, 346.878 KB/s AES-256-GCM-enc 350 KB took 1.016 seconds, 344.488 KB/s AES-256-GCM-dec 350 KB took 1.016 seconds, 344.488 KB/s SHA 14 MB took 1.000 seconds, 14.062 MB/s SHA-256 15 MB took 1.000 seconds, 15.234 MB/s SHA-384 17 MB took 1.000 seconds, 17.383 MB/s SHA-512 18 MB took 1.001 seconds, 17.512 MB/s HMAC-SHA 14 MB took 1.000 seconds, 13.818 MB/s HMAC-SHA256 15 MB took 1.001 seconds, 14.951 MB/s HMAC-SHA384 17 MB took 1.001 seconds, 16.683 MB/s HMAC-SHA512 17 MB took 1.000 seconds, 16.943 MB/s RSA 2048 public 20 ops took 1.017 sec, avg 50.850 ms, 19.666 ops/sec RSA 2048 private 4 ops took 1.059 sec, avg 264.750 ms, 3.777 ops/sec ECC 256 key gen 4 ops took 1.092 sec, avg 273.000 ms, 3.663 ops/sec ECDHE 256 agree 4 ops took 1.089 sec, avg 272.250 ms, 3.673 ops/sec ECDSA 256 sign 4 ops took 1.101 sec, avg 275.250 ms, 3.633 ops/sec ECDSA 256 verify 2 ops took 1.092 sec, avg 546.000 ms, 1.832 ops/sec 返回页面顶部 平台: Intel x86_64: Benchmark: 近期版本的 wolfSSL 加入了针对 Intel x86_64 平台的新汇编代码,性能大幅提升。前两张图展示的是 AES-GCM 的性能提升。在这些基准测试图中,Y 轴表示数据块大小,X 轴表示相对于上一版 wolfSSL 的速度倍数。 AES-GCM 的汇编代码已重写,以充分发挥 AVX1 和 AVX2 指令的优势。目前 AES-GCM 的性能已不逊于甚至超越 OpenSSL。 ChaCha20-Poly1305 是一种较新的认证加密算法,设计初衷是作为 AES-GCM 的替代方案。该算法简单,在没有 AES 和 GCM 硬件加速的 CPU 上也能快速运行。 最常用的摘要算法是 SHA-256 和 SHA-384。随着 AES-GCM 在 TLS 中的引入,SHA-256 和 SHA-384 在应用数据认证中的使用有所减少,但它们仍用于握手消息认证、单向函数(伪随机数生成器所需)以及数字签名。 Curve25519 是 Montgomery 椭圆曲线的一组参数,安全性约为 128 位,用于密钥交换,凭借其速度优势及被纳入多项标准而广受欢迎。该算法已作为 TLS v1.3 的一部分,NIST 也在考虑将其纳入 SP 800-186。Ed25519 是 Twisted Edwards 曲线的一组参数,与 Curve25519 在数学上相关,具有相同的安全特性。基于 Twisted Edwards 曲线设计的新签名方案速度很快,也已纳入 TLS v1.3。此外还起草了一份规范草案,描述如何使用基于 Ed25519 的 EdDSA 数字证书。 RSA 是证书中最常用的公钥算法。执行 TLS 握手时,服务器会对迄今为止的消息哈希进行签名,客户端则验证证书链中各证书的签名,并用证书中的公钥验证消息哈希。签名和验证是握手中最耗时的操作。 DH 一直是握手中的首选密钥交换算法,但由于椭圆曲线变体在相同安全级别下速度要快得多,DH 正逐渐被淘汰。密钥交换是 TLS 握手中第二耗时的操作。 椭圆曲线密码学(ECC)是有限域(FF)密码学的替代方案,后者包括 RSA、DSA 和 DH 等算法。ECDSA 是 RSA 和 DSA 的椭圆曲线变体,ECDH 是 DH 的椭圆曲线变体。ECDSA 和 ECDH 可用于任何其有限域对应算法适用的场景。ECC 需要一条预定义的曲线来执行运算。最常用的曲线是 P-256,它具有 128 位强度,并被纳入多项标准,包括 TLS、IETF 证书标准以及 NIST 的 FIPS 186-4。浏览器和 Web 服务器越来越倾向于使用 ECDH 而非 DH,因为前者速度快得多。 如果您对在您的平台上使用 wolfSSL 嵌入式 TLS 库或其性能优化有任何疑问,请联系 support@wolfssl.com。 返回页面顶部 平台: NXP i.MX6 Benchmark: 以下基准测试数据在 NXP i.MX6 上运行 wolfSSL 基准测试应用、搭载 Green Hills INTEGRITY OS 采集而来。 返回页面顶部 平台: iPhone X: Benchmark: 基准测试在 Apple iPhone X 上进行,该设备搭载 Apple 新一代 A11 处理器。以下测试使用 ARMv8 加密扩展和单精度数学运算来展示性能。 使用我们的 ARMv8 加密汇编加速后,对称加密 AES 和 SHA 的性能如下: Algorithm Performance AES-128 CBC Encrypt 912.347 MB/s(36.58 倍) AES-128 CBC Decrypt 6,084.83 MB/s(256.15 倍) AES-128 GCM Encrypt 1,242.28 MB/s(193.65 倍) AES-128 GCM Decrypt 575.83 MB/s(90.26 倍) SHA-256 1,717.28 MB/s(56.11 倍) 通过 `./configure –enable-armasm` 或 `WOLFSSL_ARMASM` 宏可启用此特性。 使用我们的单精度数学加速后,非对称算法 RSA、DH 和 ECC 的性能如下: Algorithm Performance RSA 2048 public 1,211.27 ops/sec(1.50 倍) RSA 2048 private 32.59 ops/sec(1.18 倍) DH 2048 key gen 77.44 ops/sec(1.24 倍) DH 2048 key agree 77.45 ops/sec(1.29 倍) ECC 256 key gen 1670.65 ops/sec(8.67 倍) ECDHE 256 agree 396.88 ops/sec(2.05 倍) ECDSA 256 sign 1,212.33 ops/sec(6.42 倍) ECDSA 256 verify 331.02 ops/sec(2.38 倍) 通过 `./configure –enable-sp` 或 `WOLFSSL_HAVE_SP_RSA`、`WOLFSSL_HAVE_SP_DH` 和 `WOLFSSL_HAVE_SP_ECC` 宏可启用此特性。 TLS v1.2 按密码套件划分的性能基准测试结果如下: Algorithm Performance DHE-RSA-AES128-SHA256 CPS 22.5,读取 388 MB/s,写入 106 MB/s ECDHE-RSA-AES128-GCM-SHA256 CPS 26.2,读取 598 MB/s(RX),写入 125 MB/s ECDHE-ECDSA-AES128-GCM-SHA256 CPS 83.4,读取 504.8 MB/s,写入 92.2 MB/s 以上基准测试在 iPhone X 上使用单线程运行 wolfCrypt 和 wolfSSL 基准测试工具完成。
`X` = 相比默认软件实现的性能提升倍数。
`CPS` = 每秒连接数 参考:wolfCrypt/wolfSSL 在 iPhone 8/8 Plus/X (A11) 上的基准测试博客文章 返回页面顶部 平台: Microchip PIC32MZ: PIC32MZ,主频 200MHz 基准测试: 测试数据基于 PIC32MZ Ethernet Starter Kit,使用 wolfCrypt benchmark 应用,通过 MPLAB X 编译。 软件加密 硬件加密 AES-CBC 0.26 Mb/s 5.78 Mb/s AES-CTR 0.69 Mb/s 5.67 Mb/s 3DES 6.19 Mb/s 6.19 Mb/s MD5 6.22 Mb/s 16.84 Mb/s SHA-1 3.46 Mb/s 16.65 Mb/s SHA-256 1.678 Mb/s 15.84 Mb/s 参考: wolfSSL PIC32 支持 返回页面顶部 平台: STM32: STM32F439,主频 180MHz
STM32F756,主频 216MHz 基准测试: 测试数据基于 STMicroelectronics 评估板,使用 wolfCrypt benchmark 应用,通过 IAR EWARM 编译(优化等级:High/Size) 参考: wolfSSL STM32 支持 返回页面顶部 平台: STM32F777NI: ARM Cortex M7
216 MHz 使用 GCC ARM (arm-none-eabi-gcc),基于 OpenSTM32 (System Workbench) 裸机环境 基准测试: 加密基准测试: wolfSSL 软件加密,普通大整数数学库 RNG 3 MB 用时 1.000 秒 3.149 MB/s AES-Enc 6 MB 用时 1.000 秒 6.494 MB/s AES-Dec 7 MB 用时 1.000 秒 6.519 MB/s AES-GCM-Enc 3 MB 用时 1.004 秒 553 MB/s AES-GCM-Dec 3 MB 用时 1.004 秒 2.553 MB/s AES-CTR 7 MB 用时 1.000 秒 6.543 MB/s CHACHA 16 MB 用时 1.000 秒 15.723 MB/s CHA-POLY 10 MB 用时 1.000 秒 10.474 MB/s 3DES 1 MB 用时 1.008 秒 1.405 MB/s MD5 24 MB 用时 1.000 秒 24.243 MB/s POLY1305 42 MB 用时 1.000 秒 41.821 MB/s SHA 14 MB 用时 1.000 秒 14.380 MB/s SHA-224 8 MB 用时 1.000 秒 8.423 MB/s SHA-256 8 MB 用时 1.000 秒 8.423 MB/s SHA-384 2 MB 用时 1.000 秒 2.319 MB/s SHA-512 2 MB 用时 1.000 秒 2.319 MB/s STM32F7 硬件加密,普通大整数数学库 RNG 6 MB 用时 1.000 秒 6.030 MB/s AES-Enc 30 MB 用时 1.000 秒 30.396 MB/s AES-Dec 30 MB 用时 1.000 秒 30.371 MB/s AES-GCM-Enc 42 MB 用时 1.000 秒 42.261 MB/s AES-GCM-Dec 33 MB 用时 1.000 秒 32.861 MB/s AES-CTR 48 MB 用时 1.000 秒 47.827 MB/s CHACHA 16 MB 用时 1.000 秒 15.747 MB/s CHA-POLY 11 MB 用时 1.000 秒 10.522 MB/s 3DES 13 MB 用时 1.000 秒 12.988 MB/s MD5 41 MB 用时 1.000 秒 40.894 MB/s POLY1305 42 MB 用时 1.000 秒 41.846 MB/s SHA 38 MB 用时 1.004 秒 38.202 MB/s SHA-224 41 MB 用时 1.000 秒 41.309 MB/s SHA-256 39 MB 用时 1.000 秒 39.111 MB/s SHA-384 2 MB 用时 1.004 秒 2.310 MB/s SHA-512 2 MB 用时 1.004 秒 2.310 MB/s 参考: wolfSSL 与 STM32F7
STM32F77NI 返回页首 平台: Apple iPhone 8/8 Plus/X Apple A11 Bionic
2.39 GHz 六核 64 位 Benchmark: 在 Apple 最新 iPhone 8 / 8 Plus 和 iPhone X 所搭载的全新 A11 处理器上运行的基准测试。该处理器包含六个 ARMv8 CPU 核心(2 个 2.53GHz 高性能核心和 4 个 1.42GHz 高能效核心)。测试使用了 wolfSSL 针对 ARMv8 加密扩展和单精度数学运算的最新优化,充分展示了其卓越性能。使用 ARMv8 加密汇编优化后的对称算法 AES 和 SHA 表现如下: AES-128 CBC 加密:912.347 MB/s(36.58 倍) AES-128 CBC 解密:6,084.83 MB/s(256.15 倍) AES-128 GCM 加密:1,242.28 MB/s(193.65 倍) AES-128 GCM 解密:575.83 MB/s(90.26 倍) SHA-256:1,717.28 MB/s(56.11 倍) 该功能可通过 `./configure –enable-armasm` 或定义 `WOLFSSL_ARMASM` 启用。 使用单精度数学优化后的非对称算法 RSA、DH 和 ECC 表现如下: RSA 2048 公钥:1,211.27 ops/sec(1.50 倍) RSA 2048 私钥:32.59 ops/sec(1.18 倍) DH 2048 密钥生成:77.44 ops/sec(1.24 倍) DH 2048 密钥协商:77.45 ops/sec(1.29 倍) ECC 256 密钥生成:1670.65 ops/sec(8.67 倍) ECDHE 256 协商:396.88 ops/sec(2.05 倍) ECDSA 256 签名:1,212.33 ops/sec(6.42 倍) ECDSA 256 验证:331.02 ops/sec(2.38 倍) 该功能可通过 `./configure –enable-sp` 或定义 `WOLFSSL_HAVE_SP_RSA`、`WOLFSSL_HAVE_SP_DH` 和 `WOLFSSL_HAVE_SP_ECC` 启用。 以下是按密码套件统计的 TLS v1.2 性能测试结果: DHE-RSA-AES128-SHA256:CPS 22.5,读取 388 MB/s,写入 106 MB/s ECDHE-RSA-AES128-GCM-SHA256:CPS 26.2,读取 598 MB/s,写入 125 MB/s ECDHE-ECDSA-AES128-GCM-SHA256:CPS 83.4,读取 504.8 MB/s,写入 92.2 MB/s 测试在 iPhone X 上以单线程运行,使用 wolfCrypt 和 wolfSSL 自带的基准测试工具。
`X` = 相较于默认纯软件实现的性能提升倍数。
`CPS` = 每秒连接数 原始测试数据可在此下载:
https://www.wolfssl.com/files/benchmarks/iOSiPhoneXA11Benchmarks.pdf 参考
wolfSSL 博客文章 返回页首 平台: Intel SGX Intel(R) Core(TM) i5-6300U @ 2.4GHz Intel Compiler Benchmark: 以下是在 Intel SGX enclave 中运行 wolfCrypt 的基准测试结果。数据显示,在 SGX 内运行 wolfCrypt 加密操作对性能的影响极小。 非 SGX 环境测试结果 SHA-256 0.264s 189.078 MB/s
AES-GCM 3.142s 15.911 MB/s
RSA (2048) 加密 0.238 ms
RSA (2048) 解密 6.239 ms SGX 基准测试,enclave 内部 SHA-256 0.263s 190.436 MB/s -0.38% 差异
AES-GCM 3.128s 15.985 MB/s -0.45% 差异
RSA (2048) 加密 0.245 ms 2.94% 差异
RSA (2048) 解密 6.242 ms 0.05% 差异
参考资料 Intel Software Guard Extensions (SGX) SDK
返回页首 平台: Intel QuickAssist
DH895xCC Intel(R) Core(TM) i7-4790 CPU 3.60GHz 12GB 内存 基准测试: 加密基准测试: 使用 wolfSSL 的异步基准测试程序,在用户空间运行多线程并绑定 CPU 亲和性: RSA 2048 公钥:209,909 次/秒
RSA 2048 私钥:41,999 次/秒
DH 2048 密钥生成:112,491 次/秒
DH 2048 密钥协商:95,129 次/秒
ECDHE 256 协商:55,117 次/秒
ECDSA 256 签名:46,798 次/秒
ECDSA 256 验证:28,917 次/秒
AES-CBC 加密:2,932 MB/s
AES-CBC 解密:2,882 MB/s
AES-GCM:2,903 MB/s
3DES:1,511 MB/s
MD5:2,309 MB/s
SHA:5,068 MB/s
SHA-224:2,392 MB/s
SHA-256:1,275 MB/s
SHA-384:2,020 MB/s
SHA-512:1,908 MB/s 参考资料 wolfSSL 博客文章
Intel QuickAssist 返回页首 平台: HiKey LeMaker (ARMv8)
Kirin 620 SoC
ARM® Cortex™-A53 八核 64 位,主频最高 1.2GHz(ARM v8 指令集) 8GB eMMC 存储
1GB 内存 基准测试: 下面是硬件加速与普通软件加密对比的完整基准测试数据:
算法 软件加密 硬件加密 RNG 16.761 MB/s 82.599 MB/s AES-128-CBC 加密 26.491 MB/s 649.179 MB/s AES-128-CBC 解密 26.915 MB/s 607.407 MB/s AES-192-CBC 加密 22.796 MB/s 566.717 MB/s AES-192-CBC 解密 23.130 MB/s 553.092 MB/s AES-256-CBC 加密 20.004 MB/s 504.143 MB/s AES-256-CBC 解密 20.207 MB/s 491.374 MB/s AES-128-GCM 加密 6.224 MB/s 393.407 MB/s AES-128-GCM 解密 6.226 MB/s 182.279 MB/s AES-192-GCM 加密 5.895 MB/s 361.801 MB/s AES-192-GCM 解密 5.895 MB/s 175.676 MB/s AES-256-GCM 加密 5.609 MB/s 333.911 MB/s AES-256-GCM 解密 5.610 MB/s 169.085 MB/s CHACHA 60.510 MB/s 60.017 MB/s CHA-POLY 41.805 MB/s 41.410 MB/s MD5 156.310 MB/s 154.421 MB/s POLY1305 144.464 MB/s 143.058 MB/s SHA 89.874 MB/s 89.154 MB/s SHA-256 38.805 MB/s 533.139 MB/s HMAC-MD5 156.301 MB/s 154.083 MB/s HMAC-SHA 89.859 MB/s 89.045 MB/s HMAC-SHA256 38.814 MB/s 532.316 MB/s RSA,2048 位,公钥 171.995 次/秒 171.355 次/秒 RSA,2048 位,私钥 13.716 次/秒 13.686 次/秒 DH,2048 位,密钥生成 50.831 次/秒 50.575 次/秒 DH,2048 位,协商 41.826 次/秒 41.596 次/秒 返回页首 平台: AVX1:1.8GHz,Intel Core i5
AVX2:Intel Broadwell 基准测试: 加密基准测试: AVX2: SHA-256 处理 50 MB 用时 0.320 秒,156.118 MB/s
每字节周期数 = 9.75 = 47% AVX1: SHA-256 处理 50 MB 用时 0.272 秒,184.068 MB/s
每字节周期数 = 11.89 = 39% 普通模式:SHA-256 处理 50 MB 用时 0.376 秒,132.985 MB/s
每字节周期数 = 16.46 AVX2: SHA-384 处理 50 MB 用时 0.226 秒,221.318 MB/s
每字节周期数 = 6.88 = 42% AVX1: SHA-384 处理 50 MB 用时 0.192 秒,260.975 MB/s
每字节周期数 = 8.39 = 9% 普通模式:SHA-384 处理 50 MB 用时 0.209 秒,239.743 MB/s
每字节周期数 = 9.13 AVX2: SHA-512 处理 50 MB 用时 0.224 秒,223.120 MB/s
每字节周期数 = 6.82 = 75% AVX1: SHA-512 处理 50 MB 用时 0.188 秒,266.126 MB/s
每字节周期数 = 8.22 = 50% 普通模式:SHA-512 处理 50 MB 用时 0.281 秒,177.997 MB/s
每字节周期数 = 12.29 参考资料 wolfSSL 博客文章 返回页首 平台: Atmel SAMD21
ARM Cortex M0
48 MHz 基准测试: TLS 建立时间: 硬件加速 ATECC508A:平均 2.342 秒
纯软件实现:平均 13.422 秒 使用 ATECC508A 后,TLS 连接建立速度快了 5.73 倍。 纯软件实现(SAMD21 48MHz Cortex-M0,Fast Math TFM-ASM): ECC 256 密钥生成 3123.000 毫秒,5 次迭代平均值
EC-DHE 密钥协商 3117.000 毫秒,5 次迭代平均值
EC-DSA 签名耗时 1997.000 毫秒,5 次迭代平均值
EC-DSA 验签耗时 5057.000 毫秒,5 次迭代平均值 ATECC508A 硬件加速实现: ECC 256 密钥生成 144.400 毫秒,5 次迭代平均值
EC-DHE 密钥协商 134.200 毫秒,5 次迭代平均值
EC-DSA 签名耗时 293.400 毫秒,5 次迭代平均值
EC-DSA 验签耗时 208.400 毫秒,5 次迭代平均值 供参考,RNG、AES、MD5、SHA 和 SHA256 的基准测试结果如下: RNG 25 kB 耗时 0.784 秒,0.031 MB/s(来自 ATECC508A)
AES 25 kB 耗时 0.177 秒,0.138 MB/s
MD5 25 kB 耗时 0.050 秒,0.488 MB/s
SHA 25 kB 耗时 0.141 秒,0.173 MB/s
SHA-256 25 kB 耗时 0.352 秒,0.069 MB/s 参考资料 Microchip 硬件 TLS 平台
wolfSSL Microchip ATECC508A 概览与示例 返回页首 平台: STM32F221G-EVAL
ARM Cortex M3
120MHz
1 MB FLASH
128 KB SRAM 基准测试: 加密基准测试: 软件加密:wolfCrypt Benchmark,普通大整数运算库 AES 1024 kB 耗时 0.822 秒,1.22 MB/s
ARC4 1024 KB 耗时 0.219 秒,4.57 MB/s
DES 1024 KB 耗时 1.513 秒,0.66 MB/s
3DES 1024 KB 耗时 3.986 秒,0.25 MB/s MD5 1024 KB 耗时 0.119 秒,8.40 MB/s
SHA 1024 KB 耗时 0.279 秒,3.58 MB/s
SHA-256 1024 KB 耗时 0.690 秒,1.45 MB/s RSA 2048 加密耗时 111.17 毫秒,100 次迭代平均值
RSA 2048 解密耗时 1204.77 毫秒,100 次迭代平均值
DH 2048 密钥生成耗时 467.90 毫秒,100 次迭代平均值
DH 2048 密钥协商耗时 538.94 毫秒,100 次迭代平均值 STM32F2 硬件加密:wolfCrypt Benchmark,普通大整数运算库 AES 1024 kB 耗时 0.105 秒,9.52 MB/s
ARC4 1024 KB 耗时 0.219 秒,4.57 MB/s
DES 1024 KB 耗时 0.125 秒,8.00 MB/s
3DES 1024 KB 耗时 0.141 秒,7.09 MB/s MD5 1024 KB 耗时 0.045 秒,22.22 MB/s
SHA 1024 KB 耗时 0.047 秒,21.28 MB/s
SHA-256 1024 KB 耗时 0.690 秒,1.45 MB/s RSA 2048 加密耗时 111.09 毫秒(100 次迭代平均值)
RSA 2048 解密耗时 1204.88 毫秒(100 次迭代平均值)
DH 2048 密钥生成 467.56 毫秒(100 次迭代平均值)
DH 2048 密钥协商 542.11 毫秒(100 次迭代平均值) 参考 wolfSSL 与 STM32 返回页首 平台: Texas Instruments
Tiva C Series TM4C1294XL Connected Launchpad (www.ti.com)
ARM Cortex-M4
120 MHz
1 MB FLASH
256 KB SRAM
6 KB EEPROM 基准测试: 加密算法基准测试: AES 25 kB 耗时 0.038 秒,0.642 MB/s
Camellia 25 kB 耗时 0.032 秒,0.763 MB/s
ARC4 25 kB 耗时 0.006 秒,4.069 MB/s
RABBIT 25 kB 耗时 0.005 秒,4.883 MB/s
CHACHA 25 kB 耗时 0.007 秒,3.488 MB/s
3DES 25 kB 耗时 0.164 秒,0.149 MB/s MD5 25 kB 耗时 0.003 秒,8.138 MB/s
POLY1305 25 kB 耗时 0.004 秒,6.104 MB/s
SHA 25 kB 耗时 0.006 秒,4.069 MB/s
SHA-256 25 kB 耗时 0.014 秒,1.744 MB/s
SHA-512 25 kB 耗时 0.042 秒,0.581 MB/s RSA 2048 加密耗时 88.000 毫秒(1 次迭代平均值)
RSA 2048 解密耗时 1456.000 毫秒(1 次迭代平均值)
DH 2048 密钥生成 661.000 毫秒(1 次迭代平均值)
DH 2048 密钥协商 665.000 毫秒(1 次迭代平均值) ECC 256 密钥生成 130.400 毫秒(5 次迭代平均值)
EC-DHE 密钥协商 118.000 毫秒(5 次迭代平均值)
EC-DSA 签名耗时 136.800 毫秒(5 次迭代平均值)
EC-DSA 验证耗时 253.800 毫秒(5 次迭代平均值) 参考 CyaSSL 与 TI-RTOS 返回页首 平台: Freescale TWR-K70F120M (www.freescale.com)
Freescale Kinetis K70
120 MHz
2 GB FLASH
1 GB RAM 基准测试: 加密算法基准测试: AES 5120 kB 耗时 9.059 秒,0.55 MB/s
ARC4 5120 kB 耗时 2.190 秒,2.28 MB/s
DES 5120 kB 耗时 18.453 秒,0.27 MB/s MD5 5120 kB 耗时 1.396 秒,3.58 MB/s
SHA 5120 kB 耗时 3.635 秒,1.38 MB/s
SHA-256 5120 kB 耗时 9.145 秒,0.55 MB/s RSA 2048 加密耗时 73.99 毫秒(100 次迭代平均值)
RSA 2048 解密耗时 1359.09 毫秒(100 次迭代平均值)
DH 2048 密钥生成耗时 536.75 毫秒(100 次迭代平均值)
DH 2048 密钥协商耗时 540.99 毫秒(100 次迭代平均值) 构建细节 MQX RTOS,使用 fastmath 库并启用 TFM_TIMING_RESISTANT 在
Apple A4 (ARM Cortex-A8)
1 GHz
8 GB FLASH
256 MB RAM 基准测试: 加密基准测试: AES 5 MB 耗时 0.500 秒,9.99 MB/s
ARC4 5 MB 耗时 0.174 秒,28.66 MB/s
RABBIT 5 MB 耗时 0.126 秒,39.56 MB/s
3DES 5 MB 耗时 2.196 秒,2.28 MB/s MD5 5 MB 耗时 0.163 秒,30.73 MB/s
SHA 5 MB 耗时 0.137 秒,36.61 MB/s
SHA-256 5 MB 耗时 0.309 秒,16.20 MB/s RSA 1024 加密耗时 1.12 毫秒(100 次迭代平均值)
RSA 1024 解密耗时 17.81 毫秒(100 次迭代平均值)
DH 1024 密钥生成耗时 11.90 毫秒(100 次迭代平均值)
DH 1024 密钥协商耗时 11.22 毫秒(100 次迭代平均值) 构建细节 完整构建,使用 fastmath 编译(–enable-fastmath) 参考 博客文章:Running CyaSSL on the Apple TV 2 返回页面顶部 平台:
(www.mbed.org)
ARM Cortex-M3
96 MHz
512 kB FLASH
32 kB RAM 基准测试: 内存占用: RAM 占用:2.0 kB
Flash 占用*:64 kB
* 包含约 3 kB 的测试驱动代码。 加密基准测试: RSA 公钥运算:10 毫秒
RSA 私钥运算:165 毫秒 构建细节 – 完整构建,除 SHA-512、DH、DSA 和 HC-128 外全部包含
– 使用 mbed 云编译器编译 参考 http://mbed.org/users/toddouska/libraries/CyaSSL/lm43pv
http://mbed.org/users/toddouska/programs/cyassl-client/lm394s 返回页首 各密码算法性能对比 尽管具体密码算法的性能取决于主机平台,下图展示了 wolfCrypt 部分算法之间的相对性能。测试在 Macbook Pro(OS X 10.6.8,2.2 GHz Intel Core i7)上进行。 返回页首 如果你只想使用部分密码算法,可以自定义 wolfSSL 在建立 SSL/TLS 连接时使用的密码套件和/或算法。例如,要强制使用 128 位 AES,在调用 wolfSSL_CTX_new(SSL_CTX_new)之后添加如下代码: wolfSSL_CTX_set_cipher_list(ctx, “AES128-SHA”); 返回页首 后量子 Kyber 性能测试(Linux) 平台: 11th Gen Intel® Core™ i7-1185G7 @ 3.00GHz × 8 测试结果: 说明: – 仅使用 1 个核心 – 列出传统算法是为了便于对比 – 所用的 wolfSSL 配置为:
./configure --disable-psk --disable-shared --enable-intelasm --enable-aesni --enable-sp --enable-sp-math --enable-sp-asm --enable-kyber=wolfssl,all 'CFLAGS=-Os -DECC_USER_CURVES -DHAVE_ECC256 -DHAVE_ECC384' wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP384R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 953.534 ms Tx Total : 6.757 ms Rx : 2.753 MB/s Tx : 388.471 MB/s Connect : 55.076 ms Connect Avg : 2.503 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP384R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 956.304 ms Tx Total : 5.627 ms Rx : 2.745 MB/s Tx : 466.488 MB/s Connect : 54.827 ms Connect Avg : 2.492 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP521R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 948.864 ms Tx Total : 6.542 ms Rx : 2.766 MB/s Tx : 401.226 MB/s Connect : 59.490 ms Connect Avg : 2.704 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 ECC_SECP521R1): Total : 5505024 bytes Num Conns : 22 Rx Total : 951.101 ms Tx Total : 5.791 ms Rx : 2.760 MB/s Tx : 453.275 MB/s Connect : 59.598 ms Connect Avg : 2.709 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL1): Total : 5505024 bytes Num Conns : 22 Rx Total : 954.452 ms Tx Total : 7.089 ms Rx : 2.750 MB/s Tx : 370.297 MB/s Connect : 47.349 ms Connect Avg : 2.152 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL1): Total : 5505024 bytes Num Conns : 22 Rx Total : 956.931 ms Tx Total : 6.189 ms Rx : 2.743 MB/s Tx : 424.149 MB/s Connect : 46.129 ms Connect Avg : 2.097 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL3): Total : 5767168 bytes Num Conns : 23 Rx Total : 958.702 ms Tx Total : 7.431 ms Rx : 2.868 MB/s Tx : 370.070 MB/s Connect : 48.699 ms Connect Avg : 2.117 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL3): Total : 5767168 bytes Num Conns : 23 Rx Total : 961.281 ms Tx Total : 6.032 ms Rx : 2.861 MB/s Tx : 455.903 MB/s Connect : 47.497 ms Connect Avg : 2.065 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL5): Total : 5767168 bytes Num Conns : 23 Rx Total : 967.680 ms Tx Total : 6.709 ms Rx : 2.842 MB/s Tx : 409.906 MB/s Connect : 51.945 ms Connect Avg : 2.258 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 KYBER_LEVEL5): Total : 5767168 bytes Num Conns : 23 Rx Total : 969.719 ms Tx Total : 6.183 ms Rx : 2.836 MB/s Tx : 444.774 MB/s Connect : 49.456 ms Connect Avg : 2.150 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P256_KYBER_LEVEL1): Total : 2883584 bytes Num Conns : 12 Rx Total : 494.597 ms Tx Total : 2.225 ms Rx : 2.780 MB/s Tx : 618.065 MB/s Connect : 540.056 ms Connect Avg : 45.005 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P256_KYBER_LEVEL1): Total : 2883584 bytes Num Conns : 12 Rx Total : 495.337 ms Tx Total : 2.282 ms Rx : 2.776 MB/s Tx : 602.567 MB/s Connect : 538.472 ms Connect Avg : 44.873 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P384_KYBER_LEVEL3): Total : 2883584 bytes Num Conns : 12 Rx Total : 503.707 ms Tx Total : 3.427 ms Rx : 2.730 MB/s Tx : 401.222 MB/s Connect : 526.276 ms Connect Avg : 43.856 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P384_KYBER_LEVEL3): Total : 2883584 bytes Num Conns : 12 Rx Total : 505.002 ms Tx Total : 3.302 ms Rx : 2.723 MB/s Tx : 416.462 MB/s Connect : 523.324 ms Connect Avg : 43.610 ms wolfSSL 服务端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P521_KYBER_LEVEL5): Total : 2883584 bytes Num Conns : 12 Rx Total : 503.258 ms Tx Total : 3.171 ms Rx : 2.732 MB/s Tx : 433.556 MB/s Connect : 537.391 ms Connect Avg : 44.783 ms wolfSSL 客户端基准测试(TLS13-AES256-GCM-SHA384,密钥组 P521_KYBER_LEVEL5): Total : 2883584 bytes Num Conns : 12 Rx Total : 504.367 ms Tx Total : 3.316 ms Rx : 2.726 MB/s Tx : 414.636 MB/s Connect : 532.820 ms Connect Avg : 44.402 ms 返回页首 后量子 Kyber 基准测试(ARM Cortex-M4) 平台: STM NUCLEO-F446ZE 基准测试: 说明: – 项目中的 HCLK 设置为 168MHz – 仅使用 1 个核心
– wolfSSL 数学配置设为 "Single Precision ASM Cortex-M3+ Math"
– 优化选项:-Ofast
– 为便于对比,保留了传统算法 RSA 2048 public 82 ops took 1.020 sec, avg 12.439 ms, 80.392 ops/sec RSA 2048 private 4 ops took 1.827 sec, avg 456.750 ms, 2.189 ops/sec DH 2048 key gen 5 ops took 1.181 sec, avg 236.200 ms, 4.234 ops/sec DH 2048 agree 6 ops took 1.419 sec, avg 236.500 ms, 4.228 ops/sec ECC SECP256R1 key gen 118 ops took 1.012 sec, avg 8.576 ms, 116.601 ops/sec ECDHE SECP256R1 agree 56 ops took 1.016 sec, avg 18.143 ms, 55.118 ops/sec KYBER512 128 key gen 232 ops took 1.004 sec, avg 4.328 ms, 231.076 ops/sec KYBER512 128 encap 192 ops took 1.008 sec, avg 5.250 ms, 190.476 ops/sec KYBER512 128 decap 178 ops took 1.004 sec, avg 5.640 ms, 177.291 ops/sec KYBER768 192 key gen 146 ops took 1.008 sec, avg 6.904 ms, 144.841 ops/sec KYBER768 192 encap 118 ops took 1.008 sec, avg 8.542 ms, 117.063 ops/sec KYBER768 192 decap 110 ops took 1.000 sec, avg 9.091 ms, 110.000 ops/sec KYBER1024 256 key gen 92 ops took 1.011 sec, avg 10.989 ms, 90.999 ops/sec KYBER1024 256 encap 76 ops took 1.000 sec, avg 13.158 ms, 76.000 ops/sec KYBER1024 256 decap 72 ops took 1.000 sec, avg 13.889 ms, 72.000 ops/sec 返回页首 后量子 Kyber 基准测试(MacOS) 平台: Apple MacBook Pro 18,3,搭载 Apple M1 Pro 处理器,主频 3.09 GHz 基准测试: 说明: – 仅使用 1 个核心 数学库: 多精度:Wolf(SP) no-dyn-stack word-size=64 bits=4096 sp_int.c 单精度:ecc 256 384 521 rsa/dh 2048 3072 4096 asm sp_arm64.c wolfCrypt 基准测试(block bytes 1048576,每项至少运行 1.0 秒) DH 2048 key gen 3997 ops took 1.000 sec, avg 0.250 ms, 3996.812 ops/sec DH 2048 agree 4100 ops took 1.001 sec, avg 0.244 ms, 4097.522 ops/sec KYBER512 128 key gen 96100 ops took 1.001 sec, avg 0.010 ms, 96037.765 ops/sec KYBER512 128 encap 78000 ops took 1.000 sec, avg 0.013 ms, 77970.220 ops/sec KYBER512 128 decap 58900 ops took 1.001 sec, avg 0.017 ms, 58867.158 ops/sec KYBER768 192 key gen 58200 ops took 1.000 sec, avg 0.017 ms, 58192.314 ops/sec KYBER768 192 encap 48700 ops took 1.001 sec, avg 0.021 ms, 48664.334 ops/sec KYBER768 192 decap 38100 ops took 1.001 sec, avg 0.026 ms, 38059.656 ops/sec KYBER1024 256 key gen 37800 ops took 1.003 sec, avg 0.027 ms, 37704.299 ops/sec KYBER1024 256 encap 32600 ops took 1.001 sec, avg 0.031 ms, 32566.427 ops/sec KYBER1024 256 decap 26000 ops took 1.001 sec, avg 0.039 ms, 25967.020 ops/sec ECC [ SECP256R1] 256 key gen 84100 ops took 1.001 sec, avg 0.012 ms, 84013.469 ops/sec ECDHE [ SECP256R1] 256 agree 24400 ops took 1.004 sec, avg 0.041 ms, 24300.995 ops/sec 返回页首 平台: Raspberry Pi Pico-W
ARM Cortext M0+,125MHz 基准测试: wolfSSL/wolfCrypt v5.6.3 配置选项 – SP Math: 启用:WOLFSSL_HAVE_SP_RSA、WOLFSSL_HAVE_SP_DH、WOLFSSL_HAVE_SP_ECC、WOLFSSL_SP_ARM_THUMB_ASM 禁用:WOLFSSL_SP_SMALL、WOLFSSL_SP_ASM – AES-GCM: 启用:GCM_TABLE_4BIT 禁用:GCM_SMALL ``` (基准测试结果保持原样) ``` 返回页首 平台: Altera Agilex 5: 硬件:Terasic DE-25 Nano(Agilex 5 A5EB013BB23BE4SR1)。 主机系统:Linux 6.12.11-altera 分支版本。 CPU:Cortex-A55 核心(2 个 @1.25 GHz)和 Cortex-A76 核心(2 个 @1.40 GHz)。 基准测试软件:wolfSSL 5.9.1 benchmark.c,使用 GCC + O2 编译。 数据块大小:1048576 每个算法耗时:约 1.0 秒 基准测试结果:
返回页首 基准测试说明 处理器的原生寄存器位宽(32 位 vs 64 位)在执行 1000+ 位公钥运算时会产生显著影响。 fastmath(–enable-fastmath)可以降低动态内存占用,并加快公钥运算速度。如果在 32 位平台上开启 fastmath 编译遇到问题,可以禁用共享库,避免 PIC 占用寄存器(另见 README 中的说明): ./configure –enable-fastmath –disable-shared
make clean
make *注意:切换 configure 选项后,建议对 wolfSSL 执行一次 "make clean" 默认情况下,fastmath 会尽可能使用汇编优化。如果汇编优化不可用,也可以在编译 wolfSSL 时向 CFLAGS 添加 TFM_NO_ASM,在没有汇编优化的情况下使用 fastmath: ./configure –enable-fastmath CFLAGS=-DTFM_NO_ASM 对于非嵌入式平台的用户,可以尝试使用 fasthugemath 进一步压榨 fastmath 的性能: ./configure –enable-fasthugemath wolfSSL 默认构建在内存占用与性能之间做了较好的平衡。如果你更关注其中某一方面,请参阅 wolfSSL 手册第 2 章了解更多配置选项。 大块数据传输:wolfSSL 默认使用 128 字节的 I/O 缓冲区,因为约 80% 的 SSL 流量都在这个范围内,同时也能限制动态内存使用。如果需要大块数据传输,可以配置使用 16K 缓冲区(SSL 的最大值)。 出版物与宣传资料 与我们的 SSL/TLS 及加密库基准测试相关的出版物:
wolfSSL+NTRU:高性能 SSL 这份宣传资料介绍了 wolfSSL 嵌入式 SSL 库搭配 Security Innovation 的 NTRU 密码算法所能获得的性能提升。NTRU 与 RSA 公钥算法类似,但速度可以快 20 到 200 倍。 下载宣传资料
wolfSSL 安全 memcached 基准测试 wolfSSL 加密速度快、内存占用低,可以轻松部署在支持数千连接的大规模服务器上。这份宣传资料展示了使用 wolfSSL 的 memcached 基准测试结果。 下载宣传资料