Benchmark: add cycle counts for asym ops

[SparkiDev]

Description

Added million of cycles per op information.
Getting cycle count for Aarch64 now too.

Testing

Inte x64l and Aarch64 emulator.

Checklist

• added tests
• updated/added doxygen
• updated appropriate READMEs
• Updated manual and documentation

[devin-ai-integration]

🛟 Devin Lifeguard found 1 likely issues in this PR

• check-all-return-codes snippet: Verify whether ignoring the XSNPRINTF return value is acceptable in this codebase; if not, store the return value from XSNPRINTF in an int and handle negatives or length overflows accordingly in the new SHOW_*_CYCLES_OPS macros.

@SparkiDev
please take a look at the above issues which Devin flagged. Devin will not fix these issues automatically.

[dgarske]

I resolved the merge conflict tested and approve.

[JacobBarthelmeh]

The cycles per byte calculation seems off. What I see on the Linaro node:

./configure --enable-armasm --enable-sp --enable-sp-asm && make


------------------------------------------------------------------------------
 wolfSSL version 5.8.2
------------------------------------------------------------------------------
Math: 	Multi-Precision: Wolf(SP) word-size=64 bits=4096 sp_int.c
	Single Precision: ecc 256 384 521 rsa/dh 2048 3072 4096 asm sp_arm64.c
	Assembly Speedups: ARMASM ALIGN
wolfCrypt Benchmark (block bytes 1048576, min 1.0 sec each)
RNG                         85 MiB took 1.048 seconds,   81.094 MiB/s Cycles per byte =   0.01
AES-128-CBC-enc            650 MiB took 1.007 seconds,  645.666 MiB/s Cycles per byte =   0.00
AES-128-CBC-dec            605 MiB took 1.002 seconds,  603.849 MiB/s Cycles per byte =   0.00
AES-192-CBC-enc            570 MiB took 1.006 seconds,  566.793 MiB/s Cycles per byte =   0.00
AES-192-CBC-dec            550 MiB took 1.002 seconds,  548.960 MiB/s Cycles per byte =   0.00
AES-256-CBC-enc            505 MiB took 1.006 seconds,  502.021 MiB/s Cycles per byte =   0.00
AES-256-CBC-dec            495 MiB took 1.008 seconds,  490.868 MiB/s Cycles per byte =   0.00
AES-128-GCM-enc            555 MiB took 1.006 seconds,  551.961 MiB/s Cycles per byte =   0.00
AES-128-GCM-dec            555 MiB took 1.004 seconds,  552.964 MiB/s Cycles per byte =   0.00
AES-192-GCM-enc            515 MiB took 1.005 seconds,  512.671 MiB/s Cycles per byte =   0.00
AES-192-GCM-dec            515 MiB took 1.005 seconds,  512.650 MiB/s Cycles per byte =   0.00
AES-256-GCM-enc            485 MiB took 1.007 seconds,  481.773 MiB/s Cycles per byte =   0.00
AES-256-GCM-dec            485 MiB took 1.008 seconds,  481.231 MiB/s Cycles per byte =   0.00
GMAC Table 4-bit          1012 MiB took 1.000 seconds, 1011.610 MiB/s Cycles per byte =   0.00
CHACHA                     260 MiB took 1.010 seconds,  257.478 MiB/s Cycles per byte =   0.00
CHA-POLY                   175 MiB took 1.002 seconds,  174.628 MiB/s Cycles per byte =   0.01
POLY1305                   530 MiB took 1.006 seconds,  526.602 MiB/s Cycles per byte =   0.00
SHA                        110 MiB took 1.032 seconds,  106.555 MiB/s Cycles per byte =   0.01
SHA-224                    525 MiB took 1.000 seconds,  524.798 MiB/s Cycles per byte =   0.00
SHA-256                    525 MiB took 1.003 seconds,  523.473 MiB/s Cycles per byte =   0.00
SHA-384                    125 MiB took 1.022 seconds,  122.336 MiB/s Cycles per byte =   0.01
SHA-512                    125 MiB took 1.023 seconds,  122.210 MiB/s Cycles per byte =   0.01
SHA-512/224                125 MiB took 1.022 seconds,  122.290 MiB/s Cycles per byte =   0.01
SHA-512/256                125 MiB took 1.020 seconds,  122.555 MiB/s Cycles per byte =   0.01
SHA3-224                    85 MiB took 1.037 seconds,   82.000 MiB/s Cycles per byte =   0.01
SHA3-256                    80 MiB took 1.029 seconds,   77.765 MiB/s Cycles per byte =   0.01
SHA3-384                    65 MiB took 1.076 seconds,   60.427 MiB/s Cycles per byte =   0.02
SHA3-512                    45 MiB took 1.053 seconds,   42.734 MiB/s Cycles per byte =   0.03
HMAC-SHA                   110 MiB took 1.035 seconds,  106.322 MiB/s Cycles per byte =   0.01
HMAC-SHA224                525 MiB took 1.006 seconds,  521.686 MiB/s Cycles per byte =   0.00
HMAC-SHA256                530 MiB took 1.005 seconds,  527.105 MiB/s Cycles per byte =   0.00
HMAC-SHA384                125 MiB took 1.021 seconds,  122.440 MiB/s Cycles per byte =   0.01
HMAC-SHA512                125 MiB took 1.021 seconds,  122.459 MiB/s Cycles per byte =   0.01
PBKDF2                      30 KiB took 1.001 seconds,   29.855 KiB/s Cycles per byte =  39.25
RSA     2048   public      4300 ops took 1.019 sec, avg 0.237 ms,  4218.856 ops/sec, 1223080 cycles      284.4 Cycles/op
RSA     2048  private       200 ops took 1.747 sec, avg 8.733 ms,   114.508 ops/sec, 2095926 cycles    10479.6 Cycles/op
DH      2048  key gen       232 ops took 1.000 sec, avg 4.311 ms,   231.969 ops/sec, 1200162 cycles     5173.1 Cycles/op
DH      2048    agree       300 ops took 1.292 sec, avg 4.308 ms,   232.117 ops/sec, 1550939 cycles     5169.8 Cycles/op
ECC   [      SECP256R1]   256  key gen      7200 ops took 1.006 sec, avg 0.140 ms,  7158.320 ops/sec, 1206986 cycles      167.6 Cycles/op
ECDHE [      SECP256R1]   256    agree      1900 ops took 1.003 sec, avg 0.528 ms,  1894.778 ops/sec, 1203307 cycles      633.3 Cycles/op
ECDSA [      SECP256R1]   256     sign      4800 ops took 1.018 sec, avg 0.212 ms,  4713.935 ops/sec, 1221909 cycles      254.6 Cycles/op
ECDSA [      SECP256R1]   256   verify      1800 ops took 1.011 sec, avg 0.562 ms,  1780.586 ops/sec, 1213084 cycles      673.9 Cycles/op
Benchmark complete

A 0.01 cycle per byte would be blazingly fast.

[anhu]

It just so happens that this was requested in ZD 20619.

[SparkiDev]

Can't get the real frequency so I've put the clock frequency into the header and allowed the setting of actual frequency with -freq for Aarch64 only.

[SparkiDev]

retest this please

fipsv2:
�[1;38;5;243mdid NOT clone wolfSSL�[0m
R_VER = 5.8.2
�[1;32mRUNNING: run_checkout 432b-v2�[0m

[JacobBarthelmeh]

I'm still seeing some issue with using -freq. Sent a private note, maybe I am building incorrectly for collecting cycles/byte ?

[SparkiDev]

retest this please

PRB tests aborted

[JacobBarthelmeh]

Worked for me now when setting the -freq argument.