Coin Brief ENDE

SIMD-0567 would cut the compute cost of Solana's token-account program by 81%

The Solana Improvement Documents repository merged SIMD-0567, "p-ATA: CU-optimized ATA Program", on 1 November. Its status is Review and its activation feature is still to be assigned. The authors, from Anza and Ergonia, propose replacing the Associated Token Account program at its existing address, ATokenGPvbdGVxr1b2hvZbsiqW5xWH25efTNsLJA8knL, with a Pinocchio-based implementation: no-std, no heap allocation, zero-copy parsing of instructions and accounts, minimal dependencies.

The motivation is volume. On a 30,000-block stratified mainnet sample, the ATA program appears in 11.85% of all transactions, 31.3% of non-vote ones, and accounts for 13.28% of total compute unit use. The proposal measures an average reduction of 81% per instruction, weighted by frequency: the most common case, an idempotent create of a new SPL token account, falls from 22,940 to 4,171 CU, and an idempotent call on an existing account from 3,710 to 530. Given the program's share, the authors expect a cluster-wide CU reduction of about 10 to 11%. They cite p-token, which cut the token program's CU use by 93%, as the precedent.

The replacement keeps the same instruction discriminants, accounts, signer statuses, address and bump derivation, and error codes. It adds one instruction, CreateWithArgs, a seven-byte payload in which callers may pass the bump, the account length and the rent sysvar as hints; the proposal's own table shows these hints help most cases but cost a little extra in some Token-2022 creates.

SIMD-0567 would cut the compute cost of Solana's token-account program by 81%
SIMD-0567 would cut the compute cost of Solana's token-account program by 81% — Coin Brief

What it means

Like p-token, this is capacity freed without anyone changing their code: wallets and programs that call the ATA program keep working, and blocks get room back. The open steps are the review the status names and the feature gate that would switch it on.

Written by Victoria Shinder.