vote cost

[tao-stones]

Problem

Be good to know how the vote cost in mainnet-beta look like,

• the percentage of 2 sig vs 1 sig
• the min/max/median of tx data size
• the min/max/median of account sizes

Proposed Solution

[tao-stones]

replaying mainnet-beta ledger, looks like vote transactions load static amount of accounts:

• first signer account (node account): 0 bytes
• vote account: 3731 bytes
• SysvarC1ock111..., 40 bytes
• SysvarS1otHashes111..., 20488 bytes
• Vote111..., 19 bytes.
  So that total to 24_278 bytes (less than 32K page size)

Didn't see 2 signature votes. (edited)

It gives me confidence to set a static CU for vote transactions: 1 * signatures + 2 * write locks + 1 vote builtin ix + 24_278 bytes cost = 3428 CU, pretty much same as they are counted today. (edited)

Hopefully I didn't miss anything, wdyt @t-nelson @AshwinSekar @apfitzge

[7layermagik]

Hmm I think that is wrong actually -- most votes just pay 0.000005 sol and have one signer as far as I know. Having two signers for all those votes is a pretty huge cost. Here's an example transaction from us: https://explorer.solana.com/tx/5UCfUbvNmU2KMSZzoptEg64n9BAb6mTNcZ7YrohQg3zP2vG2tDKwzTYA4n7kyCY6yXN8SMv7GaCSCLkt434AW8EX

[tao-stones]

thanks for pointing out, I'll rerun with larger data set.

[tao-stones]

Ah, my script mis-labeled write-locks to "signatures". updated comment.

[t-nelson]

this is just the cost of votes that fit is_simple_vote(), right?

[tao-stones]

this is just the cost of votes that fit is_simple_vote(), right?

correct.

Are there other vote transactions allowed other than "simple-vote"? Does it not drop/ignore vote if it comes with part of normal transaction (eg a normal transaction includes a vote instruction)?

[t-nelson]

shouldn't be dropped, no. but won't be prioritized as a vote either, just treated as normal transaction