SNOVA Optimization

[pqclab-zero]

Dear All,

I hope this message finds you well. I am pleased to announce that the latest optimized code for the SNOVA post-quantum cryptography digital signature scheme has been updated on GitHub. This update brings some significant improvements and optimizations, which we believe will enhance the security and performance of our signature scheme.

You can find the latest code at the following link: https://github.com/PQCLAB-SNOVA/SNOVA

Here are the latest benchmarks for efficiency:

            name   generate_keys_ssk   generate_keys_esk   sign_digest_ssk   sign_digest_esk   verify_signture
SNOVA(37, 17, 2)             801,339             828,048           690,663           338,723           146,738
 SNOVA(25, 8, 3)             433,186             442,031           713,438           370,046           218,801
 SNOVA(24, 5, 4)             202,843             212,256           444,236           306,736           163,805
SNOVA(56, 25, 2)           4,207,799           4,302,330         3,332,244           964,716           507,009
SNOVA(49, 11, 3)           2,758,759           2,808,062         3,702,911         1,365,463         1,004,519
 SNOVA(37, 8, 4)             891,056             933,377         1,795,054         1,188,690           544,395
SNOVA(75, 33, 2)          11,844,511          12,109,075         8,871,223         2,304,920         1,165,161
SNOVA(66, 15, 3)           8,635,966           8,779,555        10,685,501         3,546,746         2,460,059
SNOVA(60, 10, 4)           3,877,397           4,013,756         6,206,765         3,110,898         1,504,945

Our team remains committed to continuously improving and optimizing our digital signature scheme to provide better security in the post-quantum era. If you have any questions or suggestions, please feel free to contact us anytime.

Thank you once again for your attention and support to our work!

Best Regards,
SNOVA TEAM

[thomwiggers]

Thanks for the update---it's pretty tough to keep up with what's happening so I'm afraid many scheme's data got a little bit stale. This helps.

Just to clarify what I'm looking at, what's the difference between ssk and esk?

[pqclab-zero]

ssk uses the seed as the private key, esk uses the private key that has already been expanded, so when signing with ssk, the private key will be expanded by seed first, but this is not needed when using esk, so it is faster.

[pqclab-zero]

However, the storage space of the esk private key is larger.

[pqclab-zero]

SNOVA parameter

SL	Name	V	O	L	sk size (esk)	sk size (ssk)	pk size	sign size
1	SNOVA_37_17_2	37	17	2	60008(+48)	48	9826(+16)	108(+16)
1	SNOVA_25_8_3	25	8	3	37962(+48)	48	2304(+16)	148.5(+16)
1	SNOVA_24_5_4	24	5	4	34112(+48)	48	1000(+16)	232(+16)
3	SNOVA_56_25_2	56	25	2	202132(+48)	48	31250(+16)	162(+16)
3	SNOVA_49_11_3	49	11	3	174798(+48)	48	5989.5(+16)	270(+16)
3	SNOVA_37_8_4	37	8	4	128384(+48)	48	4096(+16)	360(+16)
5	SNOVA_75_33_2	75	33	2	515360(+48)	48	71874(+16)	216(+16)
5	SNOVA_66_15_3	66	15	3	432297(+48)	48	15187.5(+16)	364.5(+16)
5	SNOVA_60_10_4	60	10	4	389312(+48)	48	8000(+16)	560(+16)

esk vs ssk