Score:1

Comparison of Post Quantum Algorithms in Nist 3rd round

tl flag

I'm interested in the Post Quantum Algorithms of the 3rd round of the NIST standardization competition. The algorithms for KEM are Classic McEliece, CRYSTALS-KYBER, NTRU, SABER and their 5 alternatives. The signature algorithms are CRYSTALS-DILITHIUM, FALCON, Rainbow and the 3 alternatives. I was looking for a comparison of those algorithms regarding key length (mainly for level 1 NIST security) and runtime. Sadly I couldn't find a good source for that. Can you help me?

kelalaka avatar
in flag
There was a question about this a year ago... need to find...
kelalaka avatar
in flag
This covers half of your question [What are the public key and output sizes for the four remaining PQC KEM candidates?](https://crypto.stackexchange.com/q/87093/18298). Of course, there are some advancements. Your question is too broad in the sense that include two different subject to cover.
Score:2
de flag

You can find the full sets of parameters and run times for all propositions in their relevant algorithm specifications and supporting documents:

CRYSTALS-KYBER:

  • parameter sets - page 9
  • runtimes - page 15

Classic McEliece:

  • parameter sets - page 19
  • runtimes - page 33

NTRU:

  • parameter sets - page 5
  • runtimes - page 29

Saber:

  • parameter sets - page 11
  • runtimes - page 14

All of these and more files such as the test implementations these times are generated from can be found from the NIST round 3 submissions page.

edit: As @kelalaka has said in their comment, this question more succinctly answers the question regarding key length

edit 2: I found this paper which quite nicely compares the key length and performance of the round 3 submissions.

kelalaka avatar
in flag
Welcome to Cryptography.SE. This is not comparison.
supinie avatar
de flag
I don't understand how to give a comparison better than this - the runtime tables show cpu cycles taken for the given operations on the same processor, and the security of the schemes is not necessarily dependant on the length of the key but the parameters chosen.
kelalaka avatar
in flag
See my comment under the question...
supinie avatar
de flag
I see, I do agree that link does provide a more clear answer for key length and have added it as an edit, I still believe my answer gives comparison for runtimes.
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.