Skip to main content

Sharded Trees

The ShardedTreeRegistry manages 16 parallel Merkle trees, enabling 16x throughput compared to a single tree.

Architecture​

ShardedTreeRegistry
├── Shard 0  → Merkle Tree (depth 20)
├── Shard 1  → Merkle Tree (depth 20)
├── ...
└── Shard 15 → Merkle Tree (depth 20)

Each shard is an independent CommitmentTree with its own root, capacity (~1M leaves), and root history.

Shard assignment​

shard = uint256(commitment) % 16

Deterministic: the same commitment always maps to the same shard.

Proof generation​

When generating a proof for a sharded commitment:

  1. Identify the shard: shard = commitment % 16
  2. Get the shard's current root
  3. Build the Merkle proof within that shard's tree
  4. Generate the Groth16 proof as normal

The proof format is identical — the only difference is which root is used.

Trade-off: anonymity set​

Each shard has approximately 1/16th of the total commitments. This means the anonymity set per shard is smaller than the unified tree. For privacy-sensitive applications where anonymity set size is critical, use the standard single-tree CommitRevealVault.

Capacity​

ConfigurationTotal capacity
Single tree~1,048,576 commitments
16 shards~16,777,216 commitments