Easy-to-update parallelizable hashes

Melab

7/18/24, 2:33 AM

Do any parallelizable cryptographic hash algorithms that allow for quick—preferably constant-time—recalculation of a hash result upon updating a portion of the data input exist?

I know that BLAKE3 is called "parallelizable", but a Merkle tree doesn't seem like it's what I'm looking for when the time taken to recalculate the final result changes according the input's size. I figured something like an unkeyed XOR MAC would suffice until I read this answer.

0 + 3

parallelization

Gilles 'SO- stop being evil'

7/18/24, 10:56 AM

Merkle trees are the obvious solution. They're updated in log(total_length) time. A logarithm is close enough to constant in practice. Why doesn't that solve your problem?

Maarten Bodewes

7/19/24, 2:27 PM

Here is [an earlier question of mine](https://crypto.stackexchange.com/questions/1558/counter-mode-secure-hash-algorithm). Unfortunately there seem to be main theoretical issues with combining input blocks using an algorithm with high performance. My main issue was that Merkle trees may have relatively large overhead. I guess you can tweak the data size that hashed in such a way that the time to hash, while keeping the tree itself relatively small as well.

Maarten Bodewes

7/19/24, 2:29 PM

@Gilles'SO-stopbeingevil' It would require to store the hash tree if you only want to update part of the input message. Otherwise you'd have to recalculate the entire tree.

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: Easy-to-update parallelizable hashes

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.