Skip to content

Latest commit

 

History

History
1969 lines (1735 loc) · 167 KB

RESULTS.md

File metadata and controls

1969 lines (1735 loc) · 167 KB

Comparative benchmarks results

Contents:

Results overview

This page compares the performannce of some cryptographic libraries on some processors.

Several types of results are presented:

  • Execution time in microseconds. Used to compare processors on the same operation.
  • Relative performance factor (the lower, the better). This the ratio between the execution time of an algorithm and a "reference test" on the same system.
  • Ratio between encryption and decryption (or signature and verification) on the same algorithm.

The relative performance factor values have no unit and their absolute value is meaningless. They are relative to the performance of the CPU. A standard reference test is run on each CPU before running the cryptographic operations. The presented number is the ratio of the execution time of the evaluated cryptographic operation over the execution time of the standard test.

Thus, if all CPU's had the same implementation and only differed in speed, the score of one operation would be the same on all CPU's.

A lower relative execution time on a CPU maybe due to several reasons:

  • Better performance of the instructions which are characteristic of that operation.
  • Dedicated accelerated instructions for that operation (e.g. AES and SHA-x).
  • More efficient compiler (gcc is used on Linux, clang on macOS).
  • More efficient code generation options to produce the binary of the library.

Due to the level of support of the different operating systems, it was not possible to use the same version of each library on all systems. Therefore, if it possible that some differences of performance are due to different implementations in different versions of the same library. See the corresponding report file for the exact version of each library.

The tested cryptographic libraries are:

  • OpenSSL (mostly version 3, version 1 on older systems).
  • MbedTLS
  • GnuTLS
  • Nettle
  • Libtomcrypt with libtommath backend
  • Libtomcrypt with GMP backend (only relevant for asymmetric cryptography).
  • Arm64, an implementation of AES using Arm64 specialized accelerated instructions through C/C++ intrinsics. It is run only on Arm64 processors supporting FEAT_AES. Note that some cryptographic libraries also uses such an implementation.

Additional results are provided for elementary operations on large numbers, as used in asymmetric cryptography.

In the result tables, Intel/AMD processors come first, them Arm processors.

For systems with known CPU core frequencies, the "execution time" ratio is presented. This is a performance ratio, compared to a reference system, with respect to their frequencies.

Reference system: AWS Graviton 3 (Neoverse V1)

Reference test

The "reference test" is a set of basic integer operations which are typically used in cryptography. There is no floating point operation. This test serves two purposes:

  • The absolute time of the reference test is a comparision between processors.
  • Each cryptographic algorithm is compared to the reference test on the same machine.

Execution time in microseconds (the lower, the better):

CPU OS Reference test
i7-5775R macOS 3518
i7-5775R Linux 5250
i7-8565U Linux 5058
i7-13700H Linux 2846
Xeon-6254 Linux 6846
Xeon-9460 Linux 5261
Cortex A53 (R.Pi3) Linux 75122
Cortex A72 (R.Pi4) Linux 17928
Apple M1 macOS 2892
Apple M1 Linux 5307
Apple M3 macOS 2430
Ampere Altra (Neoverse N1) Linux 10241
AWS Graviton 3 (Neoverse V1) Linux 8390
Nvidia Grace (Neoverse V2) Linux 6894
RISC-V (Qemu on Apple M1) Linux 19068
RISC-V (Qemu on i7-8565U) Linux 31487
s390x (Qemu on i7-8565U) Linux 42621

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS Reference test
Xeon-9460 Linux 0.47
Cortex A72 (R.Pi4) Linux 3.1
Apple M1 macOS 0.28
Apple M1 Linux 0.51
Apple M3 macOS 0.19
Ampere Altra (Neoverse N1) Linux 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0
Nvidia Grace (Neoverse V2) Linux 0.65

RSA-2048

RSA-2048 encryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 20 57 25 93 29
i7-5775R Linux 21 59 28 95 31
i7-8565U Linux 36 55 26 74 29
i7-13700H Linux 17.2 35 14.8 47 16.7
Xeon-6254 Linux 19.8 51 23 82
Xeon-9460 Linux 18.6 39 22 54
Cortex A53 (R.Pi3) Linux 775 1328 671 1927 698
Cortex A72 (R.Pi4) Linux 154 205 127 295 138
Apple M1 macOS 15.2 34 16.5 52 21
Apple M1 Linux 16.7 38 20.0 54 24
Apple M3 macOS 11.2 143 13.0 40 17.0
Ampere Altra (Neoverse N1) Linux 81 116 74 156 79
AWS Graviton 3 (Neoverse V1) Linux 26 61 29 74 32
Nvidia Grace (Neoverse V2) Linux 20.0 37 20 49
RISC-V (Qemu on Apple M1) Linux 298 690 143 393 171
RISC-V (Qemu on i7-8565U) Linux 470 970 235 650 285
s390x (Qemu on i7-8565U) Linux 870 0.000 942 2406 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 5.8 16.4 7.2 26 8.5
i7-5775R Linux 4.1 11.4 5.4 18.1 6.0
i7-8565U Linux 7.2 11.0 5.3 14.6 5.7
i7-13700H Linux 6.0 12.3 5.2 16.8 5.9
Xeon-6254 Linux 2.9 7.6 3.4 12.1
Xeon-9460 Linux 3.5 7.5 4.3 10.4
Cortex A53 (R.Pi3) Linux 10.3 17.7 8.9 25 9.3
Cortex A72 (R.Pi4) Linux 8.6 11.5 7.1 16.5 7.7
Apple M1 macOS 5.3 11.8 5.7 18.1 7.4
Apple M1 Linux 3.1 7.3 3.8 10.2 4.6
Apple M3 macOS 4.6 59 5.3 16.9 7.0
Ampere Altra (Neoverse N1) Linux 8.0 11.4 7.3 15.3 7.7
AWS Graviton 3 (Neoverse V1) Linux 3.2 7.4 3.5 8.9 3.9
Nvidia Grace (Neoverse V2) Linux 2.9 5.5 3.0 7.2
RISC-V (Qemu on Apple M1) Linux 15.7 36 7.5 20 9.0
RISC-V (Qemu on i7-8565U) Linux 15.0 30 7.5 20 9.1
s390x (Qemu on i7-8565U) Linux 20 0.000 22 56 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.51 0.47 0.58 0.55
Cortex A72 (R.Pi4) Linux 8.3 4.8 6.4 5.7 6.1
Apple M1 macOS 0.46 0.45 0.46 0.57 0.53
Apple M1 Linux 0.50 0.51 0.56 0.59 0.61
Apple M3 macOS 0.27 1.5 0.29 0.36 0.34
Ampere Altra (Neoverse N1) Linux 2.6 1.6 2.2 1.8 2.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.58 0.48 0.56 0.53

RSA-2048 decryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 551 2545 1661 2373 803
i7-5775R Linux 589 5625 2073 2300 878
i7-8565U Linux 598 5537 2021 1976 823
i7-13700H Linux 500 1479 979 1317 457
Xeon-6254 Linux 1483 2412 1548 2024
Xeon-9460 Linux 1169 1771 1543 1435
Cortex A53 (R.Pi3) Linux 26744 59438 35755 50136 23283
Cortex A72 (R.Pi4) Linux 4997 16383 6631 7686 4347
Apple M1 macOS 550 1409 1036 1407 593
Apple M1 Linux 601 3781 1240 1481 733
Apple M3 macOS 392 1108 773 1179 436
Ampere Altra (Neoverse N1) Linux 2857 4523 3617 4145 2569
AWS Graviton 3 (Neoverse V1) Linux 965 2538 1675 2046 952
Nvidia Grace (Neoverse V2) Linux 1548 1659 1172 1363
RISC-V (Qemu on Apple M1) Linux 11311 33374 10874 10760 5675
RISC-V (Qemu on i7-8565U) Linux 17508 46882 16979 16146 9388
s390x (Qemu on i7-8565U) Linux 20866 39536 62180 72203 22908

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 156 723 472 674 228
i7-5775R Linux 112 1071 394 438 167
i7-8565U Linux 118 1094 399 390 162
i7-13700H Linux 175 519 344 462 160
Xeon-6254 Linux 216 352 226 295
Xeon-9460 Linux 222 336 293 272
Cortex A53 (R.Pi3) Linux 356 791 475 667 309
Cortex A72 (R.Pi4) Linux 278 913 369 428 242
Apple M1 macOS 190 487 358 486 205
Apple M1 Linux 113 712 233 279 138
Apple M3 macOS 161 456 318 485 179
Ampere Altra (Neoverse N1) Linux 279 441 353 404 250
AWS Graviton 3 (Neoverse V1) Linux 115 302 199 243 113
Nvidia Grace (Neoverse V2) Linux 224 240 170 197
RISC-V (Qemu on Apple M1) Linux 593 1750 570 564 297
RISC-V (Qemu on i7-8565U) Linux 556 1488 539 512 298
s390x (Qemu on i7-8565U) Linux 489 927 1458 1694 537

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.90 0.52 0.68 0.52
Cortex A72 (R.Pi4) Linux 7.5 9.3 5.7 5.4 6.6
Apple M1 macOS 0.46 0.45 0.50 0.56 0.51
Apple M1 Linux 0.51 1.2 0.60 0.59 0.63
Apple M3 macOS 0.26 0.28 0.30 0.37 0.30
Ampere Altra (Neoverse N1) Linux 2.6 1.5 1.9 1.8 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 1.3 0.52 0.55 0.52

RSA-2048 decryption/encryption ratio

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 27 44 65 25 26
i7-5775R Linux 27 94 73 24 27
i7-8565U Linux 16.3 99 76 26 28
i7-13700H Linux 29 42 66 27 27
Xeon-6254 Linux 74 46 66 24
Xeon-9460 Linux 62 44 67 26
Cortex A53 (R.Pi3) Linux 34 44 53 26 33
Cortex A72 (R.Pi4) Linux 32 79 51 25 31
Apple M1 macOS 36 41 62 26 27
Apple M1 Linux 36 97 62 27 29
Apple M3 macOS 34 7.7 59 28 25
Ampere Altra (Neoverse N1) Linux 35 38 48 26 32
AWS Graviton 3 (Neoverse V1) Linux 35 40 57 27 29
Nvidia Grace (Neoverse V2) Linux 77 43 57 27
RISC-V (Qemu on Apple M1) Linux 37 48 75 27 33
RISC-V (Qemu on i7-8565U) Linux 37 48 72 24 32
s390x (Qemu on i7-8565U) Linux 23 0.000 65 30 0.000

RSA-2048 encryption with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 23 57 25 93 29
i7-5775R Linux 26 62 28 86 31
i7-8565U Linux 41 56 25 74 29
i7-13700H Linux 18.9 34 14.7 47 16.8
Xeon-6254 Linux 20 51 23 82
Xeon-9460 Linux 20 39 22 54
Cortex A53 (R.Pi3) Linux 826 1328 670 1927 698
Cortex A72 (R.Pi4) Linux 170 205 127 295 138
Apple M1 macOS 16.6 34 16.5 52 21
Apple M1 Linux 18.4 38 19.8 52 24
Apple M3 macOS 12.5 143 13.0 41 17.0
Ampere Altra (Neoverse N1) Linux 85 116 74 155 78
AWS Graviton 3 (Neoverse V1) Linux 29 61 29 74 32
Nvidia Grace (Neoverse V2) Linux 21 38 20 50
RISC-V (Qemu on Apple M1) Linux 357 689 143 392 171
RISC-V (Qemu on i7-8565U) Linux 612 992 234 769 301
s390x (Qemu on i7-8565U) Linux 1028 1090 844 2643 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 6.8 16.4 7.2 26 8.5
i7-5775R Linux 5.1 11.8 5.4 16.5 5.9
i7-8565U Linux 8.2 11.1 5.0 14.7 5.9
i7-13700H Linux 6.6 12.0 5.2 16.8 5.9
Xeon-6254 Linux 2.9 7.5 3.4 12.1
Xeon-9460 Linux 4.0 7.5 4.3 10.4
Cortex A53 (R.Pi3) Linux 11.0 17.7 8.9 25 9.3
Cortex A72 (R.Pi4) Linux 9.5 11.5 7.1 16.5 7.7
Apple M1 macOS 5.7 11.8 5.7 18.1 7.3
Apple M1 Linux 3.5 7.3 3.7 9.9 4.6
Apple M3 macOS 5.2 59 5.3 16.9 7.0
Ampere Altra (Neoverse N1) Linux 8.3 11.4 7.3 15.2 7.7
AWS Graviton 3 (Neoverse V1) Linux 3.6 7.4 3.5 8.9 3.9
Nvidia Grace (Neoverse V2) Linux 3.2 5.5 3.0 7.3
RISC-V (Qemu on Apple M1) Linux 18.7 36 7.5 20 9.0
RISC-V (Qemu on i7-8565U) Linux 19.4 31 7.4 24 9.6
s390x (Qemu on i7-8565U) Linux 24 25 19.8 62 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.52 0.47 0.58 0.55
Cortex A72 (R.Pi4) Linux 8.2 4.8 6.4 5.7 6.1
Apple M1 macOS 0.45 0.45 0.46 0.57 0.53
Apple M1 Linux 0.50 0.51 0.55 0.57 0.60
Apple M3 macOS 0.27 1.5 0.29 0.36 0.34
Ampere Altra (Neoverse N1) Linux 2.5 1.6 2.2 1.8 2.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.58 0.48 0.56 0.53

RSA-2048 decryption with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 558 2544 1657 2363 802
i7-5775R Linux 692 6200 2104 2292 904
i7-8565U Linux 552 5046 1821 2002 841
i7-13700H Linux 502 1515 977 1312 457
Xeon-6254 Linux 1481 2410 1550 2020
Xeon-9460 Linux 1172 1771 1543 1433
Cortex A53 (R.Pi3) Linux 26820 59495 35773 50100 23282
Cortex A72 (R.Pi4) Linux 5017 16382 6632 7686 4348
Apple M1 macOS 552 1427 1031 1406 592
Apple M1 Linux 601 3741 1253 1500 729
Apple M3 macOS 395 1109 776 1177 436
Ampere Altra (Neoverse N1) Linux 2862 4523 3617 4145 2555
AWS Graviton 3 (Neoverse V1) Linux 968 2538 1674 2045 952
Nvidia Grace (Neoverse V2) Linux 1551 1660 1172 1367
RISC-V (Qemu on Apple M1) Linux 11420 33344 10873 10763 5673
RISC-V (Qemu on i7-8565U) Linux 17576 46636 16914 24326 9631
s390x (Qemu on i7-8565U) Linux 22485 43794 53415 67351 19283

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 158 723 471 671 228
i7-5775R Linux 131 1180 400 436 172
i7-8565U Linux 109 997 360 395 166
i7-13700H Linux 176 532 343 461 160
Xeon-6254 Linux 216 352 226 295
Xeon-9460 Linux 222 336 293 272
Cortex A53 (R.Pi3) Linux 357 791 476 666 309
Cortex A72 (R.Pi4) Linux 279 913 369 428 242
Apple M1 macOS 190 493 356 486 204
Apple M1 Linux 113 704 236 282 137
Apple M3 macOS 162 456 319 484 179
Ampere Altra (Neoverse N1) Linux 279 441 353 404 249
AWS Graviton 3 (Neoverse V1) Linux 115 302 199 243 113
Nvidia Grace (Neoverse V2) Linux 225 240 170 198
RISC-V (Qemu on Apple M1) Linux 598 1748 570 564 297
RISC-V (Qemu on i7-8565U) Linux 558 1481 537 772 305
s390x (Qemu on i7-8565U) Linux 527 1027 1253 1580 452

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.90 0.52 0.68 0.52
Cortex A72 (R.Pi4) Linux 7.5 9.3 5.7 5.4 6.6
Apple M1 macOS 0.46 0.46 0.50 0.56 0.51
Apple M1 Linux 0.50 1.2 0.61 0.60 0.62
Apple M3 macOS 0.27 0.28 0.30 0.37 0.30
Ampere Altra (Neoverse N1) Linux 2.6 1.5 1.9 1.8 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 1.3 0.52 0.55 0.53

RSA-2048 decryption/encryption ratio, with rekeying

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 23 43 65 25 26
i7-5775R Linux 26 99 73 26 28
i7-8565U Linux 13.3 90 72 26 28
i7-13700H Linux 26 44 66 27 27
Xeon-6254 Linux 73 46 66 24
Xeon-9460 Linux 56 44 67 26
Cortex A53 (R.Pi3) Linux 32 44 53 25 33
Cortex A72 (R.Pi4) Linux 29 79 51 26 31
Apple M1 macOS 33 41 62 26 27
Apple M1 Linux 32 97 63 28 30
Apple M3 macOS 31 7.7 59 28 25
Ampere Altra (Neoverse N1) Linux 33 38 48 26 32
AWS Graviton 3 (Neoverse V1) Linux 32 40 57 27 29
Nvidia Grace (Neoverse V2) Linux 70 43 57 27
RISC-V (Qemu on Apple M1) Linux 31 48 75 27 33
RISC-V (Qemu on i7-8565U) Linux 28 47 72 31 31
s390x (Qemu on i7-8565U) Linux 21 40 63 25 0.000

RSA-2048 signature

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 547 2546 1652 2364 803
i7-5775R Linux 623 5633 2102 2363 1003
i7-8565U Linux 546 5356 1830 1958 846
i7-13700H Linux 506 1735 979 1322 456
Xeon-6254 Linux 533 2476 1558 1999
Xeon-9460 Linux 297 1771 1539 1441
Cortex A53 (R.Pi3) Linux 26762 59432 35750 50109 23280
Cortex A72 (R.Pi4) Linux 5005 16439 6626 7728 4344
Apple M1 macOS 549 1430 1036 1405 595
Apple M1 Linux 598 3761 1226 1465 723
Apple M3 macOS 390 1102 774 1177 435
Ampere Altra (Neoverse N1) Linux 2856 4523 3615 4144 2568
AWS Graviton 3 (Neoverse V1) Linux 964 2541 1674 2043 951
Nvidia Grace (Neoverse V2) Linux 732 1659 1170 1362
RISC-V (Qemu on Apple M1) Linux 11331 33357 10883 10770 5667
RISC-V (Qemu on i7-8565U) Linux 18653 47068 16817 24189 9752
s390x (Qemu on i7-8565U) Linux 21569 45935 65504 63075 19758

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 155 723 469 672 228
i7-5775R Linux 118 1072 400 450 191
i7-8565U Linux 108 1058 361 387 167
i7-13700H Linux 177 609 344 464 160
Xeon-6254 Linux 77 361 227 292
Xeon-9460 Linux 56 336 292 273
Cortex A53 (R.Pi3) Linux 356 791 475 667 309
Cortex A72 (R.Pi4) Linux 279 916 369 431 242
Apple M1 macOS 189 494 358 485 205
Apple M1 Linux 112 708 231 276 136
Apple M3 macOS 160 453 318 484 179
Ampere Altra (Neoverse N1) Linux 278 441 353 404 250
AWS Graviton 3 (Neoverse V1) Linux 114 302 199 243 113
Nvidia Grace (Neoverse V2) Linux 106 240 169 197
RISC-V (Qemu on Apple M1) Linux 594 1749 570 564 297
RISC-V (Qemu on i7-8565U) Linux 592 1494 534 768 309
s390x (Qemu on i7-8565U) Linux 506 1077 1536 1479 463

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.23 0.52 0.68 0.52
Cortex A72 (R.Pi4) Linux 7.5 9.3 5.7 5.5 6.6
Apple M1 macOS 0.46 0.46 0.50 0.56 0.51
Apple M1 Linux 0.50 1.2 0.60 0.58 0.62
Apple M3 macOS 0.26 0.28 0.30 0.37 0.30
Ampere Altra (Neoverse N1) Linux 2.6 1.5 1.9 1.8 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.51 0.55 0.53

RSA-2048 verification

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 19.1 56 27 92 28
i7-5775R Linux 21 58 30 96 33
i7-8565U Linux 25 53 27 71 27
i7-13700H Linux 16.6 35 15.9 46 16.4
Xeon-6254 Linux 18.0 50 25 82
Xeon-9460 Linux 17.2 38 22 53
Cortex A53 (R.Pi3) Linux 752 1316 685 1920 692
Cortex A72 (R.Pi4) Linux 140 202 131 294 135
Apple M1 macOS 14.4 33 17.9 50 20.0
Apple M1 Linux 15.3 37 19.5 51 22
Apple M3 macOS 10.7 144 16.1 39 17.9
Ampere Altra (Neoverse N1) Linux 79 115 75 155 77
AWS Graviton 3 (Neoverse V1) Linux 25 60 30 73 31
Nvidia Grace (Neoverse V2) Linux 18.9 37 20 48
RISC-V (Qemu on Apple M1) Linux 281 683 156 385 164
RISC-V (Qemu on i7-8565U) Linux 434 965 259 939 274
s390x (Qemu on i7-8565U) Linux 559 898 997 2492 647

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 5.4 16.2 7.7 26 8.2
i7-5775R Linux 4.1 11.2 5.8 18.4 6.5
i7-8565U Linux 5.1 10.6 5.4 14.2 5.5
i7-13700H Linux 5.8 12.4 5.6 16.4 5.8
Xeon-6254 Linux 2.6 7.3 3.7 12.0
Xeon-9460 Linux 3.3 7.3 4.3 10.2
Cortex A53 (R.Pi3) Linux 10.0 17.5 9.1 25 9.2
Cortex A72 (R.Pi4) Linux 7.8 11.3 7.3 16.4 7.6
Apple M1 macOS 5.0 11.6 6.2 17.5 6.9
Apple M1 Linux 2.9 7.0 3.7 9.7 4.2
Apple M3 macOS 4.4 59 6.6 16.4 7.3
Ampere Altra (Neoverse N1) Linux 7.7 11.2 7.4 15.2 7.6
AWS Graviton 3 (Neoverse V1) Linux 3.0 7.2 3.6 8.7 3.8
Nvidia Grace (Neoverse V2) Linux 2.7 5.4 3.0 7.1
RISC-V (Qemu on Apple M1) Linux 14.8 35 8.2 20 8.6
RISC-V (Qemu on i7-8565U) Linux 13.8 30 8.2 29 8.7
s390x (Qemu on i7-8565U) Linux 13.1 21 23 58 15.2

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.51 0.47 0.56 0.54
Cortex A72 (R.Pi4) Linux 8.1 4.8 6.3 5.8 6.2
Apple M1 macOS 0.46 0.45 0.48 0.56 0.51
Apple M1 Linux 0.49 0.50 0.52 0.57 0.58
Apple M3 macOS 0.28 1.5 0.34 0.35 0.37
Ampere Altra (Neoverse N1) Linux 2.7 1.6 2.2 1.8 2.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.59 0.49 0.54 0.53

RSA-2048 signature/verification ratio

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 28 44 60 25 28
i7-5775R Linux 29 95 69 24 29
i7-8565U Linux 21 99 67 27 30
i7-13700H Linux 30 49 61 28 27
Xeon-6254 Linux 29 49 61 24
Xeon-9460 Linux 17.3 45 67 26
Cortex A53 (R.Pi3) Linux 35 45 52 26 33
Cortex A72 (R.Pi4) Linux 35 81 50 26 32
Apple M1 macOS 38 42 57 27 29
Apple M1 Linux 39 101 63 28 32
Apple M3 macOS 36 7.6 48 29 24
Ampere Altra (Neoverse N1) Linux 36 39 47 26 32
AWS Graviton 3 (Neoverse V1) Linux 38 41 55 27 30
Nvidia Grace (Neoverse V2) Linux 38 44 56 27
RISC-V (Qemu on Apple M1) Linux 40 48 69 27 34
RISC-V (Qemu on i7-8565U) Linux 42 48 64 25 35
s390x (Qemu on i7-8565U) Linux 38 51 65 25 30

RSA-2048 signature with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 553 2544 1652 2362 804
i7-5775R Linux 597 5892 2033 2329 857
i7-8565U Linux 586 4923 1848 1965 851
i7-13700H Linux 508 1503 974 1301 460
Xeon-6254 Linux 534 2478 1550 1999
Xeon-9460 Linux 302 1771 1539 1441
Cortex A53 (R.Pi3) Linux 26844 59451 35753 50104 23279
Cortex A72 (R.Pi4) Linux 5034 16433 6626 7739 4344
Apple M1 macOS 551 1432 1032 1404 594
Apple M1 Linux 599 3785 1253 1501 742
Apple M3 macOS 399 1102 774 1175 435
Ampere Altra (Neoverse N1) Linux 2863 4523 3615 4145 2568
AWS Graviton 3 (Neoverse V1) Linux 969 2541 1674 2043 951
Nvidia Grace (Neoverse V2) Linux 736 1681 1180 1362
RISC-V (Qemu on Apple M1) Linux 11416 33356 10897 10767 5668
RISC-V (Qemu on i7-8565U) Linux 17846 46877 17130 25144 9521
s390x (Qemu on i7-8565U) Linux 23340 37137 57571 59882 20397

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 157 723 469 671 228
i7-5775R Linux 113 1122 387 443 163
i7-8565U Linux 115 973 365 388 168
i7-13700H Linux 178 528 342 457 161
Xeon-6254 Linux 78 362 226 292
Xeon-9460 Linux 57 336 292 273
Cortex A53 (R.Pi3) Linux 357 791 475 666 309
Cortex A72 (R.Pi4) Linux 280 916 369 431 242
Apple M1 macOS 190 495 357 485 205
Apple M1 Linux 112 713 236 282 139
Apple M3 macOS 164 453 318 483 179
Ampere Altra (Neoverse N1) Linux 279 441 353 404 250
AWS Graviton 3 (Neoverse V1) Linux 115 302 199 243 113
Nvidia Grace (Neoverse V2) Linux 106 243 171 197
RISC-V (Qemu on Apple M1) Linux 598 1749 571 564 297
RISC-V (Qemu on i7-8565U) Linux 566 1488 544 798 302
s390x (Qemu on i7-8565U) Linux 547 871 1350 1404 478

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.23 0.52 0.68 0.52
Cortex A72 (R.Pi4) Linux 7.5 9.3 5.7 5.5 6.6
Apple M1 macOS 0.46 0.46 0.50 0.56 0.51
Apple M1 Linux 0.50 1.2 0.61 0.60 0.63
Apple M3 macOS 0.27 0.28 0.30 0.37 0.30
Ampere Altra (Neoverse N1) Linux 2.6 1.5 1.9 1.8 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.52 0.56 0.53

RSA-2048 verification with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 23 56 27 91 28
i7-5775R Linux 25 70 29 86 29
i7-8565U Linux 41 53 27 73 27
i7-13700H Linux 19.0 30 16.0 46 16.2
Xeon-6254 Linux 18.0 49 25 82
Xeon-9460 Linux 21 38 22 53
Cortex A53 (R.Pi3) Linux 814 1316 685 1919 693
Cortex A72 (R.Pi4) Linux 159 203 131 294 135
Apple M1 macOS 16.1 33 17.9 50 20.0
Apple M1 Linux 17.7 37 19.5 51 22
Apple M3 macOS 12.4 144 16.0 39 17.9
Ampere Altra (Neoverse N1) Linux 83 115 75 154 77
AWS Graviton 3 (Neoverse V1) Linux 29 60 30 73 31
Nvidia Grace (Neoverse V2) Linux 22 37 20 49
RISC-V (Qemu on Apple M1) Linux 349 683 157 385 164
RISC-V (Qemu on i7-8565U) Linux 593 960 256 967 272
s390x (Qemu on i7-8565U) Linux 711 751 1218 2278 668

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 6.6 16.2 7.7 26 8.2
i7-5775R Linux 4.9 13.4 5.6 16.5 5.7
i7-8565U Linux 8.3 10.6 5.5 14.6 5.5
i7-13700H Linux 6.7 10.8 5.6 16.3 5.7
Xeon-6254 Linux 2.6 7.3 3.7 12.0
Xeon-9460 Linux 4.0 7.3 4.3 10.2
Cortex A53 (R.Pi3) Linux 10.8 17.5 9.1 25 9.2
Cortex A72 (R.Pi4) Linux 8.9 11.3 7.3 16.4 7.6
Apple M1 macOS 5.6 11.7 6.2 17.5 6.9
Apple M1 Linux 3.3 7.0 3.7 9.7 4.3
Apple M3 macOS 5.1 59 6.6 16.4 7.4
Ampere Altra (Neoverse N1) Linux 8.2 11.2 7.4 15.1 7.6
AWS Graviton 3 (Neoverse V1) Linux 3.5 7.2 3.6 8.7 3.8
Nvidia Grace (Neoverse V2) Linux 3.3 5.4 3.0 7.1
RISC-V (Qemu on Apple M1) Linux 18.4 35 8.2 20 8.6
RISC-V (Qemu on i7-8565U) Linux 18.8 30 8.1 30 8.7
s390x (Qemu on i7-8565U) Linux 16.7 17.6 28 53 15.7

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.54 0.47 0.56 0.54
Cortex A72 (R.Pi4) Linux 7.9 4.8 6.3 5.8 6.2
Apple M1 macOS 0.45 0.45 0.48 0.56 0.51
Apple M1 Linux 0.49 0.50 0.52 0.57 0.58
Apple M3 macOS 0.27 1.5 0.34 0.35 0.37
Ampere Altra (Neoverse N1) Linux 2.5 1.6 2.2 1.8 2.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.48 0.53 0.53

RSA-2048 signature/verification ratio, with rekeying

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 23 44 60 25 28
i7-5775R Linux 23 83 69 26 28
i7-8565U Linux 14.0 92 66 26 30
i7-13700H Linux 26 49 61 28 28
Xeon-6254 Linux 29 49 60 24
Xeon-9460 Linux 14.3 45 67 26
Cortex A53 (R.Pi3) Linux 32 45 52 26 33
Cortex A72 (R.Pi4) Linux 31 80 50 26 32
Apple M1 macOS 34 42 57 27 29
Apple M1 Linux 33 101 64 29 32
Apple M3 macOS 32 7.6 48 29 24
Ampere Altra (Neoverse N1) Linux 34 39 47 26 32
AWS Graviton 3 (Neoverse V1) Linux 33 41 55 27 30
Nvidia Grace (Neoverse V2) Linux 32 45 57 27
RISC-V (Qemu on Apple M1) Linux 32 48 69 27 34
RISC-V (Qemu on i7-8565U) Linux 30 48 66 25 34
s390x (Qemu on i7-8565U) Linux 32 49 47 26 30

RSA-2048 parse public key

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 60 1.1 4.1 16.0 7.2
i7-5775R Linux 149 0.86 3.7 15.4 6.8
i7-8565U Linux 192 0.56 2.8 12.3 5.9
i7-13700H Linux 72 0.45 2.0 6.2 3.0
Xeon-6254 Linux 1.2 0.62 2.8 14.2
Xeon-9460 Linux 87 0.56 1.8 8.6
Cortex A53 (R.Pi3) Linux 1504 8.1 34 144 43
Cortex A72 (R.Pi4) Linux 435 2.0 9.9 28 13.4
Apple M1 macOS 28 0.47 1.9 9.6 6.0
Apple M1 Linux 80 0.58 1.9 10.7 7.0
Apple M3 macOS 5.6 0.33 1.6 7.4 4.6
Ampere Altra (Neoverse N1) Linux 393 0.76 3.2 12.0 5.9
AWS Graviton 3 (Neoverse V1) Linux 378 0.66 2.6 9.6 4.8
Nvidia Grace (Neoverse V2) Linux 89 0.37 1.7 6.4
RISC-V (Qemu on Apple M1) Linux 2078 10.0 39 57 20
RISC-V (Qemu on i7-8565U) Linux 2787 16.5 66 149 31
s390x (Qemu on i7-8565U) Linux 3222 0.000 0.000 140 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 17.3 0.30 1.2 4.6 2.0
i7-5775R Linux 28 0.16 0.71 2.9 1.3
i7-8565U Linux 38 0.11 0.55 2.4 1.2
i7-13700H Linux 25 0.16 0.72 2.2 1.1
Xeon-6254 Linux 0.18 0.091 0.40 2.1
Xeon-9460 Linux 16.7 0.11 0.34 1.6
Cortex A53 (R.Pi3) Linux 20 0.11 0.47 1.9 0.58
Cortex A72 (R.Pi4) Linux 24 0.11 0.55 1.6 0.75
Apple M1 macOS 10.0 0.16 0.67 3.3 2.1
Apple M1 Linux 15.1 0.11 0.37 2.0 1.3
Apple M3 macOS 2.3 0.14 0.68 3.0 1.9
Ampere Altra (Neoverse N1) Linux 38 0.075 0.31 1.2 0.58
AWS Graviton 3 (Neoverse V1) Linux 45 0.079 0.31 1.1 0.57
Nvidia Grace (Neoverse V2) Linux 12.9 0.054 0.25 0.93
RISC-V (Qemu on Apple M1) Linux 109 0.52 2.1 3.0 1.1
RISC-V (Qemu on i7-8565U) Linux 88 0.52 2.1 4.7 1.0
s390x (Qemu on i7-8565U) Linux 75 0.000 0.000 3.3 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.17 0.62 0.51 0.67
Cortex A72 (R.Pi4) Linux 1.7 4.3 5.5 4.4 4.1
Apple M1 macOS 0.062 0.58 0.60 0.81 1.0
Apple M1 Linux 0.17 0.71 0.60 0.91 1.2
Apple M3 macOS 0.010 0.33 0.41 0.50 0.63
Ampere Altra (Neoverse N1) Linux 0.90 0.99 1.0 1.1 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.19 0.44 0.52 0.53

RSA-2048 parse private key

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 65 39 35 34 5.1
i7-5775R Linux 151 36 41 34 4.9
i7-8565U Linux 198 27 30 28 5.5
i7-13700H Linux 74 18.2 26 13.2 2.1
Xeon-6254 Linux 3.7 30 24 34
Xeon-9460 Linux 90 21 19.9 18.5
Cortex A53 (R.Pi3) Linux 1562 662 317 381 39
Cortex A72 (R.Pi4) Linux 445 97 87 67 11.3
Apple M1 macOS 31 22 17.8 15.0 2.6
Apple M1 Linux 83 21 20 16.9 3.0
Apple M3 macOS 8.5 18.0 15.3 11.6 2.2
Ampere Altra (Neoverse N1) Linux 400 35 32 26 4.5
AWS Graviton 3 (Neoverse V1) Linux 383 24 23 20 3.4
Nvidia Grace (Neoverse V2) Linux 90 18.0 17.9 13.6
RISC-V (Qemu on Apple M1) Linux 2061 421 307 170 26
RISC-V (Qemu on i7-8565U) Linux 2863 651 549 418 37
s390x (Qemu on i7-8565U) Linux 3194 926 1417 336 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 18.7 11.2 10.2 9.8 1.4
i7-5775R Linux 28 7.0 7.9 6.6 0.93
i7-8565U Linux 39 5.4 6.0 5.6 1.1
i7-13700H Linux 26 6.4 9.4 4.6 0.74
Xeon-6254 Linux 0.54 4.5 3.6 5.0
Xeon-9460 Linux 17.2 4.0 3.8 3.5
Cortex A53 (R.Pi3) Linux 20 8.8 4.2 5.1 0.53
Cortex A72 (R.Pi4) Linux 24 5.5 4.9 3.7 0.63
Apple M1 macOS 11.0 7.9 6.1 5.2 0.89
Apple M1 Linux 15.8 4.0 3.8 3.2 0.56
Apple M3 macOS 3.5 7.4 6.3 4.8 0.89
Ampere Altra (Neoverse N1) Linux 39 3.5 3.1 2.6 0.44
AWS Graviton 3 (Neoverse V1) Linux 45 3.0 2.8 2.4 0.40
Nvidia Grace (Neoverse V2) Linux 13.1 2.6 2.6 2.0
RISC-V (Qemu on Apple M1) Linux 108 22 16.1 9.0 1.4
RISC-V (Qemu on i7-8565U) Linux 90 20 17.5 13.3 1.2
s390x (Qemu on i7-8565U) Linux 74 21 33 7.9 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.18 0.63 0.63 0.67
Cortex A72 (R.Pi4) Linux 1.7 5.7 5.4 4.7 4.8
Apple M1 macOS 0.067 0.74 0.62 0.59 0.61
Apple M1 Linux 0.18 0.69 0.71 0.67 0.72
Apple M3 macOS 0.014 0.47 0.43 0.37 0.42
Ampere Altra (Neoverse N1) Linux 0.91 1.3 1.2 1.1 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.19 0.57 0.60 0.52

RSA-4096

RSA-4096 encryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 62 198 82 310 90
i7-5775R Linux 69 214 91 276 98
i7-8565U Linux 129 185 84 245 92
i7-13700H Linux 60 130 54 158 57
Xeon-6254 Linux 61 179 80 276
Xeon-9460 Linux 64 159 80 174
Cortex A53 (R.Pi3) Linux 2856 4936 2245 6881 2296
Cortex A72 (R.Pi4) Linux 480 738 393 1015 412
Apple M1 macOS 53 126 60 179 69
Apple M1 Linux 58 143 79 167 79
Apple M3 macOS 40 583 45 134 52
Ampere Altra (Neoverse N1) Linux 253 433 220 563 227
AWS Graviton 3 (Neoverse V1) Linux 94 223 94 238 101
Nvidia Grace (Neoverse V2) Linux 71 138 65 160
RISC-V (Qemu on Apple M1) Linux 847 2445 481 1062 531
RISC-V (Qemu on i7-8565U) Linux 1304 3373 785 3477 863
s390x (Qemu on i7-8565U) Linux 1425 4839 3546 7484 2226

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 17.9 56 23 88 25
i7-5775R Linux 13.3 40 17.5 52 18.9
i7-8565U Linux 25 36 16.7 48 18.3
i7-13700H Linux 21 45 19.1 55 20
Xeon-6254 Linux 8.9 26 11.7 40
Xeon-9460 Linux 12.3 30 15.2 33
Cortex A53 (R.Pi3) Linux 38 65 29 91 30
Cortex A72 (R.Pi4) Linux 26 41 21 56 23
Apple M1 macOS 18.5 43 20 62 23
Apple M1 Linux 11.0 27 14.9 31 14.9
Apple M3 macOS 16.6 239 18.7 55 21
Ampere Altra (Neoverse N1) Linux 24 42 21 55 22
AWS Graviton 3 (Neoverse V1) Linux 11.3 26 11.3 28 12.1
Nvidia Grace (Neoverse V2) Linux 10.4 20 9.5 23
RISC-V (Qemu on Apple M1) Linux 44 128 25 55 27
RISC-V (Qemu on i7-8565U) Linux 41 107 24 110 27
s390x (Qemu on i7-8565U) Linux 33 113 83 175 52

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.51 0.53 0.63 0.55
Cortex A72 (R.Pi4) Linux 7.3 4.8 6.0 6.2 5.9
Apple M1 macOS 0.46 0.46 0.52 0.61 0.56
Apple M1 Linux 0.50 0.52 0.68 0.57 0.64
Apple M3 macOS 0.28 1.7 0.31 0.37 0.34
Ampere Altra (Neoverse N1) Linux 2.3 1.7 2.0 2.1 1.9
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.59 0.49 0.55 0.53

RSA-4096 decryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 3640 14755 9000 12502 5556
i7-5775R Linux 4473 23882 10585 13504 6024
i7-8565U Linux 8661 20292 9796 10669 5483
i7-13700H Linux 3647 9209 5601 7107 3208
Xeon-6254 Linux 7260 13781 9852 10608
Xeon-9460 Linux 7407 11113 9232 7468
Cortex A53 (R.Pi3) Linux 186535 378405 214376 312961 158954
Cortex A72 (R.Pi4) Linux 34813 72849 49520 45186 29517
Apple M1 macOS 3497 8798 6179 7253 3732
Apple M1 Linux 3878 14981 7477 7644 4860
Apple M3 macOS 2603 7021 5023 5638 2910
Ampere Altra (Neoverse N1) Linux 20248 28560 28455 25751 17497
AWS Graviton 3 (Neoverse V1) Linux 6295 15070 11590 10845 6575
Nvidia Grace (Neoverse V2) Linux 7935 10243 8444 7296
RISC-V (Qemu on Apple M1) Linux 61093 192521 59195 48199 33403
RISC-V (Qemu on i7-8565U) Linux 96142 267158 82392 140225 53760
s390x (Qemu on i7-8565U) Linux 96877 269989 345333 369270 142696

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 1034 4193 2557 3553 1579
i7-5775R Linux 852 4548 2015 2571 1147
i7-8565U Linux 1712 4011 1936 2109 1084
i7-13700H Linux 1281 3235 1968 2497 1127
Xeon-6254 Linux 1060 2013 1439 1549
Xeon-9460 Linux 1408 2112 1754 1419
Cortex A53 (R.Pi3) Linux 2483 5037 2853 4166 2115
Cortex A72 (R.Pi4) Linux 1941 4063 2762 2520 1646
Apple M1 macOS 1209 3041 2136 2507 1290
Apple M1 Linux 730 2822 1408 1440 915
Apple M3 macOS 1071 2888 2066 2319 1197
Ampere Altra (Neoverse N1) Linux 1977 2788 2778 2514 1708
AWS Graviton 3 (Neoverse V1) Linux 750 1796 1381 1292 783
Nvidia Grace (Neoverse V2) Linux 1150 1485 1224 1058
RISC-V (Qemu on Apple M1) Linux 3203 10096 3104 2527 1751
RISC-V (Qemu on i7-8565U) Linux 3053 8484 2616 4453 1707
s390x (Qemu on i7-8565U) Linux 2272 6334 8102 8663 3347

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.87 0.55 0.59 0.51
Cortex A72 (R.Pi4) Linux 8.0 7.0 6.2 6.0 6.5
Apple M1 macOS 0.45 0.47 0.43 0.54 0.46
Apple M1 Linux 0.50 0.81 0.52 0.57 0.60
Apple M3 macOS 0.27 0.30 0.28 0.34 0.29
Ampere Altra (Neoverse N1) Linux 2.8 1.6 2.1 2.1 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.99 0.54 0.57 0.53

RSA-4096 decryption/encryption ratio

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 57 74 109 40 61
i7-5775R Linux 64 111 115 48 60
i7-8565U Linux 67 109 116 43 59
i7-13700H Linux 60 70 102 44 55
Xeon-6254 Linux 118 76 122 38
Xeon-9460 Linux 114 69 115 42
Cortex A53 (R.Pi3) Linux 65 76 95 45 69
Cortex A72 (R.Pi4) Linux 72 98 125 44 71
Apple M1 macOS 65 69 102 40 53
Apple M1 Linux 66 104 94 45 61
Apple M3 macOS 64 12.0 110 41 55
Ampere Altra (Neoverse N1) Linux 79 65 128 45 77
AWS Graviton 3 (Neoverse V1) Linux 66 67 122 45 64
Nvidia Grace (Neoverse V2) Linux 111 73 128 45
RISC-V (Qemu on Apple M1) Linux 72 78 122 45 62
RISC-V (Qemu on i7-8565U) Linux 73 79 104 40 62
s390x (Qemu on i7-8565U) Linux 67 55 97 49 64

RSA-4096 encryption with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 66 197 81 310 90
i7-5775R Linux 77 217 90 313 99
i7-8565U Linux 188 181 84 242 90
i7-13700H Linux 63 130 54 158 57
Xeon-6254 Linux 60 178 80 276
Xeon-9460 Linux 67 159 79 174
Cortex A53 (R.Pi3) Linux 2908 4934 2245 6916 2296
Cortex A72 (R.Pi4) Linux 497 738 392 1015 412
Apple M1 macOS 54 126 60 179 69
Apple M1 Linux 60 143 81 174 82
Apple M3 macOS 41 546 45 134 52
Ampere Altra (Neoverse N1) Linux 257 433 219 563 227
AWS Graviton 3 (Neoverse V1) Linux 98 223 94 237 101
Nvidia Grace (Neoverse V2) Linux 73 136 65 161
RISC-V (Qemu on Apple M1) Linux 916 2424 481 1058 532
RISC-V (Qemu on i7-8565U) Linux 1418 3361 783 4922 858
s390x (Qemu on i7-8565U) Linux 1792 2872 3365 7866 2225

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 18.9 56 23 88 25
i7-5775R Linux 14.7 41 17.3 59 19.0
i7-8565U Linux 37 35 16.7 47 18.0
i7-13700H Linux 22 45 19.2 55 20
Xeon-6254 Linux 8.9 26 11.7 40
Xeon-9460 Linux 12.8 30 15.2 33
Cortex A53 (R.Pi3) Linux 38 65 29 92 30
Cortex A72 (R.Pi4) Linux 27 41 21 56 23
Apple M1 macOS 18.9 43 20 61 23
Apple M1 Linux 11.5 27 15.3 32 15.4
Apple M3 macOS 17.2 224 18.6 55 21
Ampere Altra (Neoverse N1) Linux 25 42 21 55 22
AWS Graviton 3 (Neoverse V1) Linux 11.7 26 11.3 28 12.1
Nvidia Grace (Neoverse V2) Linux 10.6 19.9 9.5 23
RISC-V (Qemu on Apple M1) Linux 48 127 25 55 27
RISC-V (Qemu on i7-8565U) Linux 45 106 24 156 27
s390x (Qemu on i7-8565U) Linux 42 67 78 184 52

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.51 0.53 0.63 0.55
Cortex A72 (R.Pi4) Linux 7.3 4.8 6.0 6.2 5.9
Apple M1 macOS 0.45 0.46 0.52 0.61 0.56
Apple M1 Linux 0.50 0.52 0.70 0.60 0.66
Apple M3 macOS 0.28 1.6 0.31 0.37 0.34
Ampere Altra (Neoverse N1) Linux 2.3 1.7 2.0 2.1 1.9
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.59 0.48 0.55 0.54

RSA-4096 decryption with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 3641 14754 8994 12526 5556
i7-5775R Linux 4176 24461 10711 11867 5861
i7-8565U Linux 7029 23341 9711 10505 5566
i7-13700H Linux 3659 9280 5575 7095 3221
Xeon-6254 Linux 7243 13776 9884 10626
Xeon-9460 Linux 7410 11110 9241 7460
Cortex A53 (R.Pi3) Linux 187251 379170 214359 311307 158941
Cortex A72 (R.Pi4) Linux 34867 72833 49520 45177 29516
Apple M1 macOS 3499 8798 6179 7253 3732
Apple M1 Linux 3916 15090 7411 7499 4770
Apple M3 macOS 2650 7045 5024 5609 2920
Ampere Altra (Neoverse N1) Linux 20263 28559 28557 25629 17501
AWS Graviton 3 (Neoverse V1) Linux 6299 15072 11591 10844 6574
Nvidia Grace (Neoverse V2) Linux 7925 10130 8412 7296
RISC-V (Qemu on Apple M1) Linux 61343 192813 59115 48175 33408
RISC-V (Qemu on i7-8565U) Linux 93131 269093 82632 137372 53696
s390x (Qemu on i7-8565U) Linux 107973 241173 351911 346849 142998

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 1034 4193 2556 3559 1579
i7-5775R Linux 795 4658 2040 2260 1116
i7-8565U Linux 1389 4614 1919 2076 1100
i7-13700H Linux 1285 3260 1958 2493 1131
Xeon-6254 Linux 1057 2012 1443 1552
Xeon-9460 Linux 1408 2111 1756 1417
Cortex A53 (R.Pi3) Linux 2492 5047 2853 4143 2115
Cortex A72 (R.Pi4) Linux 1944 4062 2762 2519 1646
Apple M1 macOS 1209 3041 2135 2507 1290
Apple M1 Linux 737 2843 1396 1412 898
Apple M3 macOS 1090 2898 2067 2307 1201
Ampere Altra (Neoverse N1) Linux 1978 2788 2788 2502 1708
AWS Graviton 3 (Neoverse V1) Linux 750 1796 1381 1292 783
Nvidia Grace (Neoverse V2) Linux 1149 1469 1220 1058
RISC-V (Qemu on Apple M1) Linux 3217 10111 3100 2526 1752
RISC-V (Qemu on i7-8565U) Linux 2957 8546 2624 4362 1705
s390x (Qemu on i7-8565U) Linux 2533 5658 8256 8137 3355

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.87 0.55 0.59 0.51
Cortex A72 (R.Pi4) Linux 8.0 7.0 6.2 6.0 6.5
Apple M1 macOS 0.45 0.47 0.43 0.54 0.46
Apple M1 Linux 0.51 0.81 0.52 0.56 0.59
Apple M3 macOS 0.27 0.30 0.28 0.34 0.29
Ampere Altra (Neoverse N1) Linux 2.8 1.6 2.1 2.0 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.99 0.53 0.57 0.53

RSA-4096 decryption/encryption ratio, with rekeying

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 54 74 109 40 61
i7-5775R Linux 54 112 117 37 58
i7-8565U Linux 37 128 115 43 61
i7-13700H Linux 58 71 102 44 55
Xeon-6254 Linux 119 77 123 38
Xeon-9460 Linux 110 69 115 42
Cortex A53 (R.Pi3) Linux 64 76 95 45 69
Cortex A72 (R.Pi4) Linux 70 98 126 44 71
Apple M1 macOS 63 69 102 40 53
Apple M1 Linux 64 105 91 42 58
Apple M3 macOS 63 12.9 110 41 55
Ampere Altra (Neoverse N1) Linux 78 65 129 45 77
AWS Graviton 3 (Neoverse V1) Linux 64 67 122 45 64
Nvidia Grace (Neoverse V2) Linux 108 73 127 45
RISC-V (Qemu on Apple M1) Linux 66 79 122 45 62
RISC-V (Qemu on i7-8565U) Linux 65 80 105 27 62
s390x (Qemu on i7-8565U) Linux 60 83 104 44 64

RSA-4096 signature

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 3636 14716 8973 12516 5534
i7-5775R Linux 4464 23671 10773 11314 5775
i7-8565U Linux 7816 27146 9718 10258 5632
i7-13700H Linux 4037 8906 5588 7073 3216
Xeon-6254 Linux 3724 13886 9806 10610
Xeon-9460 Linux 3864 11145 9223 7471
Cortex A53 (R.Pi3) Linux 187189 379076 214369 311659 158949
Cortex A72 (R.Pi4) Linux 34822 72850 49495 45169 29500
Apple M1 macOS 3495 8806 6183 7250 3741
Apple M1 Linux 3929 15137 7496 7722 4709
Apple M3 macOS 2618 7040 5022 5641 2924
Ampere Altra (Neoverse N1) Linux 20251 28563 28445 25749 17499
AWS Graviton 3 (Neoverse V1) Linux 6292 15072 11591 10853 6569
Nvidia Grace (Neoverse V2) Linux 4770 10126 8406 7284
RISC-V (Qemu on Apple M1) Linux 61217 192723 59295 48211 33395
RISC-V (Qemu on i7-8565U) Linux 94018 267837 82538 125473 53400
s390x (Qemu on i7-8565U) Linux 114340 256951 354929 341691 143775

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 1033 4182 2550 3557 1572
i7-5775R Linux 850 4508 2051 2154 1100
i7-8565U Linux 1545 5366 1921 2027 1113
i7-13700H Linux 1418 3129 1963 2485 1130
Xeon-6254 Linux 543 2028 1432 1549
Xeon-9460 Linux 734 2118 1753 1420
Cortex A53 (R.Pi3) Linux 2491 5046 2853 4148 2115
Cortex A72 (R.Pi4) Linux 1942 4063 2760 2519 1645
Apple M1 macOS 1208 3044 2137 2506 1293
Apple M1 Linux 740 2851 1412 1454 887
Apple M3 macOS 1077 2896 2066 2320 1203
Ampere Altra (Neoverse N1) Linux 1977 2788 2777 2514 1708
AWS Graviton 3 (Neoverse V1) Linux 749 1796 1381 1293 783
Nvidia Grace (Neoverse V2) Linux 692 1468 1219 1056
RISC-V (Qemu on Apple M1) Linux 3210 10107 3109 2528 1751
RISC-V (Qemu on i7-8565U) Linux 2985 8506 2621 3984 1695
s390x (Qemu on i7-8565U) Linux 2682 6028 8327 8016 3373

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.46 0.55 0.59 0.51
Cortex A72 (R.Pi4) Linux 8.0 7.0 6.2 6.0 6.5
Apple M1 macOS 0.45 0.47 0.43 0.54 0.46
Apple M1 Linux 0.51 0.82 0.53 0.58 0.58
Apple M3 macOS 0.27 0.30 0.28 0.34 0.29
Ampere Altra (Neoverse N1) Linux 2.8 1.6 2.1 2.1 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.53 0.57 0.53

RSA-4096 verification

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 61 195 85 306 87
i7-5775R Linux 69 220 96 268 92
i7-8565U Linux 132 195 87 245 90
i7-13700H Linux 67 130 56 156 56
Xeon-6254 Linux 58 175 84 273
Xeon-9460 Linux 62 157 79 171
Cortex A53 (R.Pi3) Linux 2827 4921 2277 6868 2284
Cortex A72 (R.Pi4) Linux 464 733 399 1009 407
Apple M1 macOS 51 124 64 177 68
Apple M1 Linux 57 142 75 169 83
Apple M3 macOS 39 544 45 130 49
Ampere Altra (Neoverse N1) Linux 250 431 222 561 225
AWS Graviton 3 (Neoverse V1) Linux 92 221 96 234 99
Nvidia Grace (Neoverse V2) Linux 68 135 66 158
RISC-V (Qemu on Apple M1) Linux 819 2412 502 1049 519
RISC-V (Qemu on i7-8565U) Linux 1237 3368 823 2863 850
s390x (Qemu on i7-8565U) Linux 1346 4026 3646 7198 2317

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 17.4 55 24 87 25
i7-5775R Linux 13.1 41 18.4 51 17.7
i7-8565U Linux 26 38 17.3 48 17.9
i7-13700H Linux 23 45 19.8 55 19.7
Xeon-6254 Linux 8.6 25 12.3 39
Xeon-9460 Linux 11.9 29 15.1 32
Cortex A53 (R.Pi3) Linux 37 65 30 91 30
Cortex A72 (R.Pi4) Linux 25 40 22 56 22
Apple M1 macOS 18.0 43 22 61 23
Apple M1 Linux 10.9 26 14.3 31 15.7
Apple M3 macOS 16.4 224 18.9 53 20
Ampere Altra (Neoverse N1) Linux 24 42 21 54 21
AWS Graviton 3 (Neoverse V1) Linux 11.0 26 11.5 27 11.8
Nvidia Grace (Neoverse V2) Linux 10.0 19.6 9.6 23
RISC-V (Qemu on Apple M1) Linux 42 126 26 55 27
RISC-V (Qemu on i7-8565U) Linux 39 106 26 90 27
s390x (Qemu on i7-8565U) Linux 31 94 85 168 54

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.50 0.53 0.61 0.54
Cortex A72 (R.Pi4) Linux 7.2 4.8 6.0 6.2 5.9
Apple M1 macOS 0.46 0.46 0.54 0.61 0.56
Apple M1 Linux 0.51 0.52 0.64 0.59 0.68
Apple M3 macOS 0.28 1.6 0.31 0.36 0.32
Ampere Altra (Neoverse N1) Linux 2.3 1.7 2.0 2.1 2.0
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.59 0.48 0.54 0.53

RSA-4096 signature/verification ratio

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 59 75 105 40 62
i7-5775R Linux 64 107 111 42 62
i7-8565U Linux 58 139 111 41 62
i7-13700H Linux 59 68 99 45 57
Xeon-6254 Linux 63 79 116 38
Xeon-9460 Linux 61 70 115 43
Cortex A53 (R.Pi3) Linux 66 77 94 45 69
Cortex A72 (R.Pi4) Linux 74 99 124 44 72
Apple M1 macOS 67 70 95 40 54
Apple M1 Linux 67 106 99 45 56
Apple M3 macOS 65 12.9 109 43 59
Ampere Altra (Neoverse N1) Linux 80 66 127 45 77
AWS Graviton 3 (Neoverse V1) Linux 67 67 119 46 66
Nvidia Grace (Neoverse V2) Linux 69 74 127 45
RISC-V (Qemu on Apple M1) Linux 74 79 117 45 64
RISC-V (Qemu on i7-8565U) Linux 75 79 100 43 62
s390x (Qemu on i7-8565U) Linux 84 63 97 47 62

RSA-4096 signature with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 3644 14713 8977 12508 5533
i7-5775R Linux 4249 23898 11284 11382 5623
i7-8565U Linux 6083 21523 9845 10463 5765
i7-13700H Linux 4097 8865 5614 7055 3219
Xeon-6254 Linux 3700 13923 9808 10619
Xeon-9460 Linux 3868 11113 9224 7474
Cortex A53 (R.Pi3) Linux 187283 378629 214372 311857 158939
Cortex A72 (R.Pi4) Linux 34884 72824 49495 45175 29507
Apple M1 macOS 3497 8802 6186 7271 3743
Apple M1 Linux 3943 15398 7429 7523 4620
Apple M3 macOS 2638 7045 5029 5632 2921
Ampere Altra (Neoverse N1) Linux 20254 28563 28549 25757 17499
AWS Graviton 3 (Neoverse V1) Linux 6297 15074 11591 10845 6573
Nvidia Grace (Neoverse V2) Linux 4806 10127 8417 7288
RISC-V (Qemu on Apple M1) Linux 61290 192573 59315 48211 33538
RISC-V (Qemu on i7-8565U) Linux 93959 270152 82190 121945 52905
s390x (Qemu on i7-8565U) Linux 103769 388794 361854 344166 150272

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 1035 4181 2551 3554 1572
i7-5775R Linux 809 4551 2148 2167 1070
i7-8565U Linux 1202 4254 1946 2068 1139
i7-13700H Linux 1439 3114 1972 2478 1131
Xeon-6254 Linux 540 2033 1432 1551
Xeon-9460 Linux 735 2112 1753 1420
Cortex A53 (R.Pi3) Linux 2493 5040 2853 4151 2115
Cortex A72 (R.Pi4) Linux 1945 4062 2760 2519 1645
Apple M1 macOS 1209 3042 2138 2513 1293
Apple M1 Linux 743 2901 1399 1417 870
Apple M3 macOS 1085 2898 2069 2317 1202
Ampere Altra (Neoverse N1) Linux 1977 2788 2787 2514 1708
AWS Graviton 3 (Neoverse V1) Linux 750 1796 1381 1292 783
Nvidia Grace (Neoverse V2) Linux 697 1469 1220 1057
RISC-V (Qemu on Apple M1) Linux 3214 10099 3110 2528 1758
RISC-V (Qemu on i7-8565U) Linux 2984 8579 2610 3872 1680
s390x (Qemu on i7-8565U) Linux 2434 9121 8489 8074 3525

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.46 0.55 0.59 0.51
Cortex A72 (R.Pi4) Linux 8.0 7.0 6.2 6.0 6.5
Apple M1 macOS 0.45 0.47 0.43 0.54 0.46
Apple M1 Linux 0.51 0.83 0.52 0.56 0.57
Apple M3 macOS 0.27 0.30 0.28 0.34 0.29
Ampere Altra (Neoverse N1) Linux 2.8 1.6 2.1 2.1 2.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.53 0.57 0.53

RSA-4096 verification with rekeying

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 65 196 85 305 88
i7-5775R Linux 74 211 101 270 100
i7-8565U Linux 112 184 88 243 90
i7-13700H Linux 70 117 56 156 56
Xeon-6254 Linux 58 176 83 272
Xeon-9460 Linux 66 157 79 171
Cortex A53 (R.Pi3) Linux 2888 4921 2277 6866 2285
Cortex A72 (R.Pi4) Linux 484 733 399 1010 406
Apple M1 macOS 53 124 64 177 68
Apple M1 Linux 60 142 75 170 83
Apple M3 macOS 41 544 45 130 49
Ampere Altra (Neoverse N1) Linux 255 431 222 561 225
AWS Graviton 3 (Neoverse V1) Linux 96 221 96 235 99
Nvidia Grace (Neoverse V2) Linux 73 135 66 158
RISC-V (Qemu on Apple M1) Linux 893 2412 503 1050 521
RISC-V (Qemu on i7-8565U) Linux 1428 3371 836 2830 864
s390x (Qemu on i7-8565U) Linux 1868 3329 3508 7246 2456

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 18.7 55 24 86 25
i7-5775R Linux 14.2 40 19.3 51 19.1
i7-8565U Linux 22 36 17.5 48 17.9
i7-13700H Linux 24 41 19.8 54 19.8
Xeon-6254 Linux 8.6 25 12.2 39
Xeon-9460 Linux 12.7 29 15.2 32
Cortex A53 (R.Pi3) Linux 38 65 30 91 30
Cortex A72 (R.Pi4) Linux 27 40 22 56 22
Apple M1 macOS 18.6 43 22 61 23
Apple M1 Linux 11.4 26 14.3 32 15.7
Apple M3 macOS 17.2 224 18.9 53 20
Ampere Altra (Neoverse N1) Linux 24 42 21 54 21
AWS Graviton 3 (Neoverse V1) Linux 11.5 26 11.5 28 11.8
Nvidia Grace (Neoverse V2) Linux 10.7 19.6 9.6 22
RISC-V (Qemu on Apple M1) Linux 46 126 26 55 27
RISC-V (Qemu on i7-8565U) Linux 45 107 26 89 27
s390x (Qemu on i7-8565U) Linux 43 78 82 170 57

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.51 0.53 0.61 0.54
Cortex A72 (R.Pi4) Linux 7.2 4.8 6.0 6.2 5.9
Apple M1 macOS 0.45 0.46 0.54 0.61 0.56
Apple M1 Linux 0.51 0.52 0.64 0.59 0.68
Apple M3 macOS 0.28 1.6 0.31 0.36 0.32
Ampere Altra (Neoverse N1) Linux 2.3 1.7 2.0 2.1 2.0
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.60 0.48 0.54 0.53

RSA-4096 signature/verification ratio, with rekeying

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 55 74 105 40 62
i7-5775R Linux 56 112 111 42 56
i7-8565U Linux 54 116 111 42 63
i7-13700H Linux 58 75 99 45 57
Xeon-6254 Linux 62 79 117 38
Xeon-9460 Linux 57 70 115 43
Cortex A53 (R.Pi3) Linux 64 76 94 45 69
Cortex A72 (R.Pi4) Linux 72 99 124 44 72
Apple M1 macOS 64 70 95 41 54
Apple M1 Linux 65 108 97 44 55
Apple M3 macOS 63 12.9 109 43 59
Ampere Altra (Neoverse N1) Linux 79 66 128 45 77
AWS Graviton 3 (Neoverse V1) Linux 64 67 119 46 66
Nvidia Grace (Neoverse V2) Linux 65 74 127 46
RISC-V (Qemu on Apple M1) Linux 68 79 117 45 64
RISC-V (Qemu on i7-8565U) Linux 65 80 98 43 61
s390x (Qemu on i7-8565U) Linux 55 116 103 47 61

RSA-4096 parse public key

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 60 1.1 4.7 44 12.6
i7-5775R Linux 173 0.87 6.8 44 12.4
i7-8565U Linux 144 0.60 4.5 33 10.9
i7-13700H Linux 81 0.43 3.4 16.2 5.5
Xeon-6254 Linux 1.4 0.69 3.3 39
Xeon-9460 Linux 88 0.58 2.1 22
Cortex A53 (R.Pi3) Linux 1515 8.4 40 407 75
Cortex A72 (R.Pi4) Linux 434 2.0 12.2 74 24
Apple M1 macOS 29 0.51 2.3 25 11.1
Apple M1 Linux 85 0.61 2.5 24 13.3
Apple M3 macOS 5.9 0.35 1.9 19.4 8.4
Ampere Altra (Neoverse N1) Linux 395 0.75 3.8 30 10.8
AWS Graviton 3 (Neoverse V1) Linux 379 0.67 3.1 24 8.7
Nvidia Grace (Neoverse V2) Linux 89 0.42 2.0 16.7
RISC-V (Qemu on Apple M1) Linux 2006 12.0 44 126 32
RISC-V (Qemu on i7-8565U) Linux 2768 19.0 78 266 54
s390x (Qemu on i7-8565U) Linux 4776 0.000 0.000 339 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 17.2 0.30 1.3 12.7 3.6
i7-5775R Linux 33 0.16 1.3 8.5 2.4
i7-8565U Linux 28 0.12 0.89 6.5 2.2
i7-13700H Linux 28 0.15 1.2 5.7 1.9
Xeon-6254 Linux 0.20 0.10 0.48 5.8
Xeon-9460 Linux 16.7 0.11 0.41 4.2
Cortex A53 (R.Pi3) Linux 20 0.11 0.54 5.4 1.0
Cortex A72 (R.Pi4) Linux 24 0.11 0.68 4.1 1.3
Apple M1 macOS 10.1 0.18 0.78 8.7 3.8
Apple M1 Linux 16.1 0.11 0.47 4.7 2.5
Apple M3 macOS 2.4 0.14 0.80 8.0 3.5
Ampere Altra (Neoverse N1) Linux 38 0.073 0.37 3.0 1.1
AWS Graviton 3 (Neoverse V1) Linux 45 0.080 0.37 2.9 1.0
Nvidia Grace (Neoverse V2) Linux 13.0 0.061 0.28 2.4
RISC-V (Qemu on Apple M1) Linux 105 0.63 2.3 6.6 1.7
RISC-V (Qemu on i7-8565U) Linux 87 0.60 2.5 8.4 1.7
s390x (Qemu on i7-8565U) Linux 112 0.000 0.000 8.0 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.17 0.64 0.51 0.69
Cortex A72 (R.Pi4) Linux 1.7 4.4 5.7 4.5 4.0
Apple M1 macOS 0.063 0.62 0.59 0.85 1.0
Apple M1 Linux 0.18 0.74 0.66 0.85 1.2
Apple M3 macOS 0.010 0.34 0.41 0.53 0.63
Ampere Altra (Neoverse N1) Linux 0.90 0.97 1.1 1.1 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.19 0.49 0.50 0.55

RSA-4096 parse private key

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 66 116 65 113 7.7
i7-5775R Linux 168 88 99 111 7.7
i7-8565U Linux 185 70 56 86 8.9
i7-13700H Linux 87 47 44 39 3.2
Xeon-6254 Linux 4.5 75 44 106
Xeon-9460 Linux 91 54 35 54
Cortex A53 (R.Pi3) Linux 1567 1824 622 1170 60
Cortex A72 (R.Pi4) Linux 451 265 164 191 18.0
Apple M1 macOS 33 62 34 49 4.3
Apple M1 Linux 89 71 35 43 4.7
Apple M3 macOS 9.9 51 26 37 3.6
Ampere Altra (Neoverse N1) Linux 406 100 67 75 7.2
AWS Graviton 3 (Neoverse V1) Linux 384 71 40 59 5.3
Nvidia Grace (Neoverse V2) Linux 91 53 30 40
RISC-V (Qemu on Apple M1) Linux 2055 1118 422 386 32
RISC-V (Qemu on i7-8565U) Linux 2838 1917 772 606 50
s390x (Qemu on i7-8565U) Linux 4178 2199 1896 833 0.000

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
i7-5775R macOS 19.0 32 18.5 32 2.2
i7-5775R Linux 32 16.8 18.9 21 1.5
i7-8565U Linux 36 14.0 11.2 17.0 1.8
i7-13700H Linux 30 16.6 15.6 13.9 1.1
Xeon-6254 Linux 0.65 11.1 6.4 15.6
Xeon-9460 Linux 17.4 10.4 6.8 10.4
Cortex A53 (R.Pi3) Linux 20 24 8.3 15.6 0.81
Cortex A72 (R.Pi4) Linux 25 14.8 9.2 10.7 1.0
Apple M1 macOS 11.4 21 11.8 17.2 1.5
Apple M1 Linux 16.8 13.5 6.8 8.3 0.89
Apple M3 macOS 4.1 21 10.8 15.6 1.5
Ampere Altra (Neoverse N1) Linux 39 9.8 6.6 7.4 0.70
AWS Graviton 3 (Neoverse V1) Linux 45 8.5 4.9 7.1 0.64
Nvidia Grace (Neoverse V2) Linux 13.3 7.7 4.4 5.8
RISC-V (Qemu on Apple M1) Linux 107 58 22 20 1.7
RISC-V (Qemu on i7-8565U) Linux 90 60 24 19.3 1.6
s390x (Qemu on i7-8565U) Linux 98 51 44 19.5 0.000

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Libtomcrypt Libtomcrypt / GMP
Xeon-9460 Linux 0.18 0.57 0.66 0.69
Cortex A72 (R.Pi4) Linux 1.7 5.4 5.8 4.7 4.9
Apple M1 macOS 0.070 0.71 0.68 0.68 0.65
Apple M1 Linux 0.19 0.81 0.72 0.60 0.72
Apple M3 macOS 0.017 0.47 0.42 0.42 0.44
Ampere Altra (Neoverse N1) Linux 0.92 1.2 1.4 1.1 1.2
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.19 0.58 0.59 0.53

AES-128-CBC

AES-128-CBC encryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 1180 1679 1177 1723 4491
i7-5775R Linux 1244 2290 1227 1553 3692
i7-8565U Linux 667 1436 677 956 3587
i7-13700H Linux 429 932 447 782 2994
Xeon-6254 Linux 678 1411 679 947 3474
Xeon-9460 Linux 595 1289 592 879 3201
Cortex A53 (R.Pi3) Linux 21010 50739 56339 54874 37195
Cortex A72 (R.Pi4) Linux 19761 10909 12656 12638 12095
Apple M1 macOS 636 1159 3254 3265 3430 975
Apple M1 Linux 723 4849 720 1317 4179 1126
Apple M3 macOS 422 729 2602 2573 2729 717
Ampere Altra (Neoverse N1) Linux 424 4648 423 4976 4738 730
AWS Graviton 3 (Neoverse V1) Linux 488 4266 492 4605 4273 850
Nvidia Grace (Neoverse V2) Linux 382 3131 374 728 3220 602
RISC-V (Qemu on Apple M1) Linux 19301 28468 27597 27583 38513
RISC-V (Qemu on i7-8565U) Linux 30125 41928 38709 38793 57255
s390x (Qemu on i7-8565U) Linux 50705 54921 60871 58705 48596

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 335 477 334 489 1276
i7-5775R Linux 237 436 233 295 703
i7-8565U Linux 131 283 133 189 709
i7-13700H Linux 150 327 157 275 1051
Xeon-6254 Linux 99 206 99 138 507
Xeon-9460 Linux 113 245 112 167 608
Cortex A53 (R.Pi3) Linux 279 675 749 730 495
Cortex A72 (R.Pi4) Linux 1102 608 705 704 674
Apple M1 macOS 220 400 1125 1128 1186 337
Apple M1 Linux 136 913 135 248 787 212
Apple M3 macOS 173 300 1070 1058 1123 295
Ampere Altra (Neoverse N1) Linux 41 453 41 485 462 71
AWS Graviton 3 (Neoverse V1) Linux 58 508 58 548 509 101
Nvidia Grace (Neoverse V2) Linux 55 454 54 105 467 87
RISC-V (Qemu on Apple M1) Linux 1012 1493 1447 1446 2019
RISC-V (Qemu on i7-8565U) Linux 956 1331 1229 1232 1818
s390x (Qemu on i7-8565U) Linux 1189 1288 1428 1377 1140

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
Xeon-9460 Linux 0.91 0.22 0.89 0.14 0.56
Cortex A72 (R.Pi4) Linux 58 3.7 37 4.0 4.1
Apple M1 macOS 1.1 0.22 5.4 0.58 0.65 0.93
Apple M1 Linux 1.2 0.92 1.2 0.23 0.79 1.1
Apple M3 macOS 0.56 0.11 3.4 0.36 0.42 0.55
Ampere Altra (Neoverse N1) Linux 0.75 0.94 0.75 0.94 0.96 0.74
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.62 0.58 0.60 0.12 0.59 0.56

AES-128-CBC decryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 168 915 168 226 4136
i7-5775R Linux 176 1512 174 234 3761
i7-8565U Linux 163 1190 161 209 3469
i7-13700H Linux 66 792 68 116 2541
Xeon-6254 Linux 161 1019 161 279 3333
Xeon-9460 Linux 92 851 90 116 2652
Cortex A53 (R.Pi3) Linux 22138 45732 47236 47203 39899
Cortex A72 (R.Pi4) Linux 14152 11032 11686 11682 12878
Apple M1 macOS 77 478 2619 2624 3190 142
Apple M1 Linux 97 4676 97 150 3643 190
Apple M3 macOS 50 139 1869 1881 2346 107
Ampere Altra (Neoverse N1) Linux 234 4495 259 4668 4900 390
AWS Graviton 3 (Neoverse V1) Linux 114 4009 124 3993 4336 260
Nvidia Grace (Neoverse V2) Linux 89 2917 95 105 3173 154
RISC-V (Qemu on Apple M1) Linux 19085 35403 16119 16127 57693
RISC-V (Qemu on i7-8565U) Linux 29662 49837 27297 26578 76965
s390x (Qemu on i7-8565U) Linux 31146 55083 39750 39514 98029

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 47 260 47 64 1175
i7-5775R Linux 33 288 33 44 716
i7-8565U Linux 32 235 31 41 685
i7-13700H Linux 23 278 24 40 893
Xeon-6254 Linux 23 148 23 40 486
Xeon-9460 Linux 17.7 161 17.2 22 504
Cortex A53 (R.Pi3) Linux 294 608 628 628 531
Cortex A72 (R.Pi4) Linux 789 615 651 651 718
Apple M1 macOS 26 165 905 907 1102 49
Apple M1 Linux 18.3 881 18.4 28 686 35
Apple M3 macOS 20 57 769 774 965 44
Ampere Altra (Neoverse N1) Linux 22 438 25 455 478 38
AWS Graviton 3 (Neoverse V1) Linux 13.7 477 14.9 475 516 31
Nvidia Grace (Neoverse V2) Linux 13.0 423 13.9 15.3 460 22
RISC-V (Qemu on Apple M1) Linux 1000 1856 845 845 3025
RISC-V (Qemu on i7-8565U) Linux 942 1582 866 844 2444
s390x (Qemu on i7-8565U) Linux 730 1292 932 927 2299

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
Xeon-9460 Linux 0.60 0.16 0.54 0.022 0.45
Cortex A72 (R.Pi4) Linux 177 4.0 135 4.2 4.3
Apple M1 macOS 0.55 0.097 17.0 0.53 0.60 0.45
Apple M1 Linux 0.69 0.95 0.64 0.031 0.68 0.60
Apple M3 macOS 0.29 0.023 9.7 0.31 0.35 0.27
Ampere Altra (Neoverse N1) Linux 1.8 0.97 1.8 1.0 0.98 1.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.62 0.57 0.60 0.021 0.58 0.47

AES-128-CBC encryption/decryption ratio

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 7.0 1.8 7.0 7.6 1.1
i7-5775R Linux 7.1 1.5 7.0 6.6 0.98
i7-8565U Linux 4.1 1.2 4.2 4.6 1.0
i7-13700H Linux 6.4 1.2 6.5 6.7 1.2
Xeon-6254 Linux 4.2 1.4 4.2 3.4 1.0
Xeon-9460 Linux 6.4 1.5 6.5 7.5 1.2
Cortex A53 (R.Pi3) Linux 0.95 1.1 1.2 1.2 0.93
Cortex A72 (R.Pi4) Linux 1.4 0.99 1.1 1.1 0.94
Apple M1 macOS 8.2 2.4 1.2 1.2 1.1 6.8
Apple M1 Linux 7.4 1.0 7.4 8.7 1.1 5.9
Apple M3 macOS 8.3 5.2 1.4 1.4 1.2 6.7
Ampere Altra (Neoverse N1) Linux 1.8 1.0 1.6 1.1 0.97 1.9
AWS Graviton 3 (Neoverse V1) Linux 4.2 1.1 3.9 1.2 0.99 3.3
Nvidia Grace (Neoverse V2) Linux 4.3 1.1 3.9 6.9 1.0 3.9
RISC-V (Qemu on Apple M1) Linux 1.0 0.80 1.7 1.7 0.67
RISC-V (Qemu on i7-8565U) Linux 1.0 0.84 1.4 1.5 0.74
s390x (Qemu on i7-8565U) Linux 1.6 1.0 1.5 1.5 0.50

AES-256-CBC

AES-256-CBC encryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 1643 2139 1646 2142 6009
i7-5775R Linux 1742 2676 1719 2081 5112
i7-8565U Linux 979 1737 927 1199 4814
i7-13700H Linux 583 1089 613 981 4011
Xeon-6254 Linux 935 1661 934 1204 4604
Xeon-9460 Linux 811 1504 808 1095 4202
Cortex A53 (R.Pi3) Linux 27673 65286 72289 70831 47898
Cortex A72 (R.Pi4) Linux 27705 13703 16892 16825 15757
Apple M1 macOS 890 1411 4323 4319 4498 1276
Apple M1 Linux 1020 5845 1026 1613 5373 1470
Apple M3 macOS 574 920 3429 3449 3562 997
Ampere Altra (Neoverse N1) Linux 593 5957 592 6641 6144 949
AWS Graviton 3 (Neoverse V1) Linux 679 5567 682 6147 5580 1081
Nvidia Grace (Neoverse V2) Linux 523 4081 523 875 4200 790
RISC-V (Qemu on Apple M1) Linux 24467 33076 33149 33143 43845
RISC-V (Qemu on i7-8565U) Linux 37893 50247 47783 47619 65983
s390x (Qemu on i7-8565U) Linux 64662 65498 71239 72580 62154

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 466 608 467 608 1707
i7-5775R Linux 331 509 327 396 973
i7-8565U Linux 193 343 183 237 951
i7-13700H Linux 205 382 215 344 1409
Xeon-6254 Linux 136 242 136 175 672
Xeon-9460 Linux 154 285 153 208 798
Cortex A53 (R.Pi3) Linux 368 869 962 942 637
Cortex A72 (R.Pi4) Linux 1545 764 942 938 878
Apple M1 macOS 307 487 1494 1493 1554 441
Apple M1 Linux 192 1101 193 303 1012 277
Apple M3 macOS 236 378 1411 1419 1465 410
Ampere Altra (Neoverse N1) Linux 57 581 57 648 599 92
AWS Graviton 3 (Neoverse V1) Linux 80 663 81 732 665 128
Nvidia Grace (Neoverse V2) Linux 75 592 75 127 609 114
RISC-V (Qemu on Apple M1) Linux 1283 1734 1738 1738 2299
RISC-V (Qemu on i7-8565U) Linux 1203 1595 1517 1512 2095
s390x (Qemu on i7-8565U) Linux 1517 1536 1671 1702 1458

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
Xeon-9460 Linux 0.89 0.20 0.88 0.13 0.56
Cortex A72 (R.Pi4) Linux 58 3.6 35 4.0 4.1
Apple M1 macOS 1.1 0.21 5.1 0.57 0.65 0.96
Apple M1 Linux 1.2 0.85 1.2 0.21 0.78 1.1
Apple M3 macOS 0.55 0.11 3.3 0.36 0.41 0.60
Ampere Altra (Neoverse N1) Linux 0.76 0.93 0.75 0.94 0.95 0.76
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.61 0.58 0.60 0.11 0.59 0.58

AES-256-CBC decryption

Execution time in microseconds (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 234 1186 234 315 5730
i7-5775R Linux 246 1668 245 328 4982
i7-8565U Linux 230 1342 224 269 4746
i7-13700H Linux 96 938 121 154 2859
Xeon-6254 Linux 225 1159 226 353 4431
Xeon-9460 Linux 129 1023 126 154 3616
Cortex A53 (R.Pi3) Linux 28485 58836 63194 63130 51241
Cortex A72 (R.Pi4) Linux 19975 13948 15877 15883 16647
Apple M1 macOS 143 558 3680 3947 4254 189
Apple M1 Linux 171 5860 153 174 5035 252
Apple M3 macOS 116 176 2711 2968 3176 131
Ampere Altra (Neoverse N1) Linux 317 5786 307 6313 6220 537
AWS Graviton 3 (Neoverse V1) Linux 155 5283 175 5683 5621 347
Nvidia Grace (Neoverse V2) Linux 121 3850 123 133 4141 202
RISC-V (Qemu on Apple M1) Linux 24193 40043 21814 21842 71284
RISC-V (Qemu on i7-8565U) Linux 37198 57469 35277 34749 94678
s390x (Qemu on i7-8565U) Linux 42516 66216 51480 53434 111634

Relative performance score (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 66 337 66 89 1628
i7-5775R Linux 46 317 46 62 948
i7-8565U Linux 45 265 44 53 938
i7-13700H Linux 33 329 42 54 1004
Xeon-6254 Linux 32 169 33 51 647
Xeon-9460 Linux 24 194 24 29 687
Cortex A53 (R.Pi3) Linux 379 783 841 840 682
Cortex A72 (R.Pi4) Linux 1114 778 885 885 928
Apple M1 macOS 49 193 1272 1364 1470 65
Apple M1 Linux 32 1104 28 32 948 47
Apple M3 macOS 47 72 1115 1221 1306 53
Ampere Altra (Neoverse N1) Linux 31 565 30 616 607 52
AWS Graviton 3 (Neoverse V1) Linux 18.6 629 20 677 669 41
Nvidia Grace (Neoverse V2) Linux 17.6 558 18.0 19.4 600 29
RISC-V (Qemu on Apple M1) Linux 1268 2099 1144 1145 3738
RISC-V (Qemu on i7-8565U) Linux 1181 1825 1120 1103 3006
s390x (Qemu on i7-8565U) Linux 997 1553 1207 1253 2619

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
Xeon-9460 Linux 0.61 0.14 0.54 0.020 0.48
Cortex A72 (R.Pi4) Linux 185 3.8 130 4.0 4.3
Apple M1 macOS 0.75 0.086 17.0 0.56 0.62 0.44
Apple M1 Linux 0.90 0.90 0.71 0.025 0.73 0.59
Apple M3 macOS 0.48 0.022 10.0 0.34 0.37 0.25
Ampere Altra (Neoverse N1) Linux 1.8 0.95 1.5 0.96 0.96 1.3
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.61 0.57 0.56 0.019 0.58 0.46

AES-256-CBC encryption/decryption ratio

CPU OS OpenSSL MbedTLS GnuTLS Nettle Libtomcrypt Arm64 accel
i7-5775R macOS 7.0 1.8 7.0 6.8 1.0
i7-5775R Linux 7.1 1.6 7.0 6.3 1.0
i7-8565U Linux 4.2 1.3 4.1 4.5 1.0
i7-13700H Linux 6.0 1.2 5.0 6.3 1.4
Xeon-6254 Linux 4.2 1.4 4.1 3.4 1.0
Xeon-9460 Linux 6.3 1.5 6.4 7.1 1.2
Cortex A53 (R.Pi3) Linux 0.97 1.1 1.1 1.1 0.93
Cortex A72 (R.Pi4) Linux 1.4 0.98 1.1 1.1 0.95
Apple M1 macOS 6.2 2.5 1.2 1.1 1.1 6.7
Apple M1 Linux 5.9 1.0 6.7 9.3 1.1 5.8
Apple M3 macOS 4.9 5.2 1.3 1.2 1.1 7.6
Ampere Altra (Neoverse N1) Linux 1.9 1.0 1.9 1.1 0.99 1.8
AWS Graviton 3 (Neoverse V1) Linux 4.4 1.1 3.9 1.1 0.99 3.1
Nvidia Grace (Neoverse V2) Linux 4.3 1.1 4.2 6.5 1.0 3.9
RISC-V (Qemu on Apple M1) Linux 1.0 0.83 1.5 1.5 0.62
RISC-V (Qemu on i7-8565U) Linux 1.0 0.87 1.4 1.4 0.70
s390x (Qemu on i7-8565U) Linux 1.5 0.99 1.4 1.4 0.56

2048-bit modular arithmetic

2048-bit modular arithmetic basic operations

Execution time in microseconds (the lower, the better):

CPU OS Add Multiply Multiply
(Montgomery)		 Multiply
(reciprocal)		 Square Divide
(reciprocal)		 Inverse Square
Root
i7-5775R macOS 0.32 4.1 0.97 2.8 4.0 0.030 255 1626
i7-5775R Linux 0.28 4.2 1.0 3.0 3.9 0.036 253 3424
i7-8565U Linux 0.26 3.8 1.5 2.7 3.6 0.031 259 2664
i7-13700H Linux 0.14 1.7 0.93 1.5 1.7 0.019 147 1581
Xeon-6254 Linux 0.26 4.0 0.92 2.5 3.7 0.029 238 1616
Xeon-9460 Linux 0.18 2.6 0.99 2.0 2.4 0.021 194 3372
Cortex A53 (R.Pi3) Linux 3.5 130 35 87 123 0.43 3180 160771
Cortex A72 (R.Pi4) Linux 0.84 16.3 8.1 14.6 15.0 0.10 679 15433
Apple M1 macOS 0.22 4.8 0.88 3.2 4.8 0.000 228 2869
Apple M1 Linux 0.24 5.3 1.1 3.7 5.2 0.022 252 1793
Apple M3 macOS 0.17 3.7 0.59 2.3 3.6 0.000 172 1131
Ampere Altra (Neoverse N1) Linux 0.38 8.3 4.9 7.5 7.6 0.034 302 9234
AWS Graviton 3 (Neoverse V1) Linux 0.30 5.9 1.5 4.3 5.8 0.025 273 2624
Nvidia Grace (Neoverse V2) Linux 0.19 3.6 1.1 2.9 3.6 0.018 173 3714
RISC-V (Qemu on Apple M1) Linux 2.2 33 16.0 43 28 0.62 1878 27965
RISC-V (Qemu on i7-8565U) Linux 3.4 53 25 70 44 0.78 2557 42683
s390x (Qemu on i7-8565U) Linux 0.000 0.000 0.000 0.000 0.000 0.000 2966 76952

Relative performance score (the lower, the better):

CPU OS Add Multiply Multiply
(Montgomery)		 Multiply
(reciprocal)		 Square Divide
(reciprocal)		 Inverse Square
Root
i7-5775R macOS 0.091 1.2 0.27 0.81 1.1 0.009 72 462
i7-5775R Linux 0.054 0.79 0.20 0.57 0.74 0.007 48 652
i7-8565U Linux 0.051 0.76 0.29 0.54 0.71 0.006 51 526
i7-13700H Linux 0.049 0.61 0.33 0.54 0.61 0.007 51 555
Xeon-6254 Linux 0.038 0.58 0.13 0.36 0.54 0.004 34 236
Xeon-9460 Linux 0.035 0.49 0.19 0.37 0.46 0.004 36 640
Cortex A53 (R.Pi3) Linux 0.046 1.7 0.48 1.2 1.6 0.006 42 2140
Cortex A72 (R.Pi4) Linux 0.047 0.91 0.45 0.82 0.84 0.006 37 860
Apple M1 macOS 0.077 1.7 0.30 1.1 1.7 0.000 78 991
Apple M1 Linux 0.045 1.0 0.20 0.70 0.98 0.004 47 337
Apple M3 macOS 0.070 1.5 0.24 0.95 1.5 0.000 70 465
Ampere Altra (Neoverse N1) Linux 0.037 0.81 0.48 0.73 0.74 0.003 29 901
AWS Graviton 3 (Neoverse V1) Linux 0.036 0.71 0.17 0.52 0.70 0.003 32 312
Nvidia Grace (Neoverse V2) Linux 0.028 0.53 0.16 0.41 0.52 0.003 25 538
RISC-V (Qemu on Apple M1) Linux 0.12 1.8 0.84 2.3 1.5 0.033 98 1466
RISC-V (Qemu on i7-8565U) Linux 0.11 1.7 0.82 2.3 1.4 0.025 81 1355
s390x (Qemu on i7-8565U) Linux 0.000 0.000 0.000 0.000 0.000 0.000 69 1805

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS Add Multiply Multiply
(Montgomery)		 Multiply
(reciprocal)		 Square Divide
(reciprocal)		 Inverse Square
Root
Xeon-9460 Linux 0.45 0.32 0.50 0.34 0.31 0.64 0.53 0.95
Cortex A72 (R.Pi4) Linux 4.1 4.0 8.0 4.9 3.7 5.9 3.6 8.5
Apple M1 macOS 0.60 0.66 0.49 0.61 0.67 0.000 0.68 0.89
Apple M1 Linux 0.65 0.73 0.59 0.70 0.72 0.72 0.75 0.56
Apple M3 macOS 0.37 0.40 0.26 0.35 0.41 0.000 0.41 0.28
Ampere Altra (Neoverse N1) Linux 1.1 1.2 2.9 1.5 1.1 1.2 0.96 3.0
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.51 0.48 0.59 0.52 0.48 0.58 0.50 1.1

2048-bit modular arithmetic exponentiations on public (short) exponents

Execution time in microseconds (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
i7-5775R macOS 21 14.3 12.7 53 71
i7-5775R Linux 22 14.9 13.3 57 72
i7-8565U Linux 28 22 19.5 49 62
i7-13700H Linux 16.7 13.4 10.9 28 29
Xeon-6254 Linux 20 14.2 12.5 46 64
Xeon-9460 Linux 19.5 15.2 12.6 38 44
Cortex A53 (R.Pi3) Linux 857 708 636 1523 2117
Cortex A72 (R.Pi4) Linux 152 128 104 246 261
Apple M1 macOS 19.2 12.8 10.9 60 83
Apple M1 Linux 22 15.4 12.9 65 92
Apple M3 macOS 15.2 9.8 8.6 44 63
Ampere Altra (Neoverse N1) Linux 87 76 61 125 131
AWS Graviton 3 (Neoverse V1) Linux 30 22 19.0 78 101
Nvidia Grace (Neoverse V2) Linux 21 16.7 14.0 51 63
RISC-V (Qemu on Apple M1) Linux 314 230 183 690 486
RISC-V (Qemu on i7-8565U) Linux 487 361 294 1156 789
s390x (Qemu on i7-8565U) Linux 0.000 349 0.000 764 669

Relative performance score (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
i7-5775R macOS 6.2 4.1 3.6 15.2 20
i7-5775R Linux 4.3 2.8 2.5 11.0 13.7
i7-8565U Linux 5.7 4.4 3.9 9.9 12.3
i7-13700H Linux 5.9 4.7 3.8 10.1 10.4
Xeon-6254 Linux 3.0 2.1 1.8 6.7 9.5
Xeon-9460 Linux 3.7 2.9 2.4 7.2 8.4
Cortex A53 (R.Pi3) Linux 11.4 9.4 8.5 20 28
Cortex A72 (R.Pi4) Linux 8.5 7.2 5.8 13.7 14.6
Apple M1 macOS 6.6 4.4 3.8 21 28
Apple M1 Linux 4.2 2.9 2.4 12.4 17.5
Apple M3 macOS 6.2 4.0 3.5 18.4 26
Ampere Altra (Neoverse N1) Linux 8.5 7.5 6.1 12.3 12.9
AWS Graviton 3 (Neoverse V1) Linux 3.6 2.7 2.3 9.4 12.1
Nvidia Grace (Neoverse V2) Linux 3.1 2.4 2.0 7.4 9.1
RISC-V (Qemu on Apple M1) Linux 16.5 12.1 9.6 36 25
RISC-V (Qemu on i7-8565U) Linux 15.5 11.5 9.4 36 25
s390x (Qemu on i7-8565U) Linux 0.000 8.2 0.000 17.9 15.7

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
Xeon-9460 Linux 0.47 0.50 0.49 0.36 0.33
Cortex A72 (R.Pi4) Linux 7.2 8.3 7.9 4.5 3.7
Apple M1 macOS 0.51 0.46 0.47 0.63 0.67
Apple M1 Linux 0.60 0.56 0.55 0.68 0.74
Apple M3 macOS 0.32 0.28 0.29 0.37 0.41
Ampere Altra (Neoverse N1) Linux 2.5 3.0 2.8 1.4 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.56 0.58 0.58 0.51 0.49

2048-bit modular arithmetic exponentiations on private (large) exponents

Execution time in microseconds (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
i7-5775R macOS 1652 1640 1660 6630 9672
i7-5775R Linux 1851 1705 1832 7138 9391
i7-8565U Linux 2610 2684 2388 6300 8666
i7-13700H Linux 1593 1582 1417 3893 4069
Xeon-6254 Linux 1595 1597 1639 5935 9039
Xeon-9460 Linux 1783 1783 1627 5017 6030
Cortex A53 (R.Pi3) Linux 84912 84760 78166 196706 295110
Cortex A72 (R.Pi4) Linux 15397 15371 13614 32410 36296
Apple M1 macOS 1471 1464 1429 7894 11609
Apple M1 Linux 1742 1737 1701 8719 13003
Apple M3 macOS 1119 1113 1103 5735 8706
Ampere Altra (Neoverse N1) Linux 9189 9180 8052 16628 18248
AWS Graviton 3 (Neoverse V1) Linux 2617 2609 2441 10329 14145
Nvidia Grace (Neoverse V2) Linux 1951 1949 1808 6793 9234
RISC-V (Qemu on Apple M1) Linux 27633 27523 24490 92583 67942
RISC-V (Qemu on i7-8565U) Linux 42673 42464 38362 153789 109387
s390x (Qemu on i7-8565U) Linux 42387 42138 42543 101567 94984

Relative performance score (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
i7-5775R macOS 469 466 471 1884 2748
i7-5775R Linux 352 324 348 1359 1788
i7-8565U Linux 516 530 472 1245 1713
i7-13700H Linux 559 556 498 1367 1429
Xeon-6254 Linux 233 233 239 866 1320
Xeon-9460 Linux 339 339 309 953 1146
Cortex A53 (R.Pi3) Linux 1130 1128 1040 2618 3928
Cortex A72 (R.Pi4) Linux 858 857 759 1807 2024
Apple M1 macOS 508 506 494 2728 4013
Apple M1 Linux 328 327 320 1642 2449
Apple M3 macOS 460 458 454 2359 3582
Ampere Altra (Neoverse N1) Linux 897 896 786 1623 1781
AWS Graviton 3 (Neoverse V1) Linux 311 310 290 1231 1685
Nvidia Grace (Neoverse V2) Linux 283 282 262 985 1339
RISC-V (Qemu on Apple M1) Linux 1449 1443 1284 4855 3563
RISC-V (Qemu on i7-8565U) Linux 1355 1348 1218 4884 3474
s390x (Qemu on i7-8565U) Linux 994 988 998 2382 2228

Execution time ratio, compared to reference system, frequencies included (the lower, the better):

CPU OS Exponent Exponent
(Montgomery)		 Exponent
(Montgomery)
(word)		 Exponent
(reciprocal)		 Exponent
(simple)
Xeon-9460 Linux 0.51 0.51 0.50 0.36 0.32
Cortex A72 (R.Pi4) Linux 8.5 8.5 8.1 4.5 3.7
Apple M1 macOS 0.46 0.46 0.48 0.62 0.67
Apple M1 Linux 0.54 0.54 0.57 0.69 0.75
Apple M3 macOS 0.28 0.28 0.29 0.36 0.40
Ampere Altra (Neoverse N1) Linux 3.0 3.0 2.9 1.4 1.1
AWS Graviton 3 (Neoverse V1) Linux 1.0 1.0 1.0 1.0 1.0
Nvidia Grace (Neoverse V2) Linux 0.59 0.59 0.58 0.52 0.51