Question about function fesm2_mul and ifma_amm52_psm2_

[YWRtaW4]

Hi,

It seems to me that function fesm2_mul from sources/ippcp/sm2/ifma_arith_psm2.c implements multiplication modular sm2_p using CIOS Montgomery multiplication, and that the function ifma_amm52_psm2_ from sources/ippcp/cryoto_mb/src/sm2/ifma_arith_psm2.c also implements multiplication modular multiplication modular sm2_p by first multiplying, then applying fast reduction. Both methods use AVX512IFMA instruction.

I'm confused about which implementation to use on the k1, namely IceLake platform. Could you provide any performance data or other suggestions? Thank you very much.

[paveldyakov]

Hi @YWRtaW4,

As I can see, one implementation is related to Single-buffer SM2 algorithm - https://www.intel.com/content/www/us/en/docs/ipp-crypto/developer-guide-reference/2021-9/functions-based-on-sm2.html The second one is for Multi-buffer SM2 algorithm - https://www.intel.com/content/www/us/en/docs/ipp-crypto/developer-guide-reference/2021-9/sm2-elliptic-curve-functions.html

Please review the documentation and API - probably it can help you to decide which function works best for your case

Please let us know if you have any more questions

[paveldyakov]

@YWRtaW4, could you please let us know if any help is required?

[paveldyakov]

Hope the provided information helped. Closing the issue because of no response.

[yellow123Nike]

Do you use ifma_mb8_to_HexStr8() please? I always get an error when I convert from mb8 to string8