Introduce a new CachedMask for BDN #61
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
CountEnabled and IndexOfNthEnabled are both O(n) in the size of the mask, making this loop n^2. The BLS operations still tend to be the slow part, but the n^2 factor will start to show up with thousands of keys.
Co-authored-by: AnomalRoil <[email protected]>
Stebalien
force-pushed
the
steb/caching-mask
branch
from
f8fb3b4 to
8fd7d55
Compare
|
Benchmark: For 3000 keys (our use-case), this is 88x faster assuming the mask is cached. |
|
Well, that's 88x faster purely for aggregating public keys. For validating an aggregate signature with 3000 signers, it's 58x faster. |
This was referenced Aug 28, 2024
Closed
This new mask will pre-compute reusable values, speeding up repeated verification and aggregation of aggregate signatures (mostly the former).
Stebalien
force-pushed
the
steb/caching-mask
branch
from
c96a0e4 to
07398f9
Compare
Stebalien
commented
Aug 30, 2024
| // 2. Pre-computes the terms for public key aggregation. Once the CachedMask has been instantiated, | ||
| // distinct sets of public keys can be aggregated by simply summing the cached terms, ~2 orders | ||
| // of magnitude faster than aggregating from scratch. | ||
| func NewCachedMask(mask Mask) (*CachedMask, error) { |
There was a problem hiding this comment.
Happy to bikeshed on the name. Maybe just callit BDNMask?
| @@ -165,25 +162,23 @@ func (scheme *Scheme) AggregateSignatures(sigs [][]byte, mask *sign.Mask) (kyber | |||
| // AggregatePublicKeys aggregates a set of public keys (similarly to | |||
| // AggregateSignatures for signatures) using the hash function | |||
| // H: keyGroup -> R with R = {1, ..., 2^128}. | |||
| func (scheme *Scheme) AggregatePublicKeys(mask *sign.Mask) (kyber.Point, error) { | |||
| coefs, err := hashPointToR(mask.Publics()) | |||
| func (scheme *Scheme) AggregatePublicKeys(mask Mask) (kyber.Point, error) { | |||
There was a problem hiding this comment.
This is a breaking change for anyone abstracting over Scheme using an interface (e.g., for testing). If that's going to be an issue, I can look into alternative APIs (e.g., adding additional methods, putting the methods on the CachedMask, etc.
This was referenced Sep 3, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This new mask will pre-compute reusable values, speeding up repeated verification and aggregation of aggregate signatures (mostly the former).
Importantly, we noticed that F3 spends a lot of time multiplying public keys and coefficients:
kyber/sign/bdn/bdn.go
Line 184 in 94dae51
But, when we have a stable set of keys to draw from (which, IMO, is the usual case), we can pre-compute that product and save a bunch of time.
This PR is stacked on #60. Unfortunately, I can't change the base to that PR without making this PR against my fork, which kind of defeats the point. Only look at the top commit.
This is an early PR to get some quick feedback before I start writing tests.