Skip to content
Research noteET-2026-0447

Four ways to price more than one resource

EIP-7999 asks how execution, state growth and data can carry separate base fees behind a gas interface that only knows one number.

3 minEthereum & Layer-2

Anders Elowsson posted a comparison of four designs for multidimensional gas accounting on 12 August, framed around EIP-7999 and the limits of how EIP-8037 meters state creation. The problem is narrow to state and awkward to solve: execution, state growth and data are separately scarce and should be separately priced, but the EVM exposes a single scalar gas value that contracts already depend on.

The four

  • Aggregate EVM gas — one shared meter, separate base fees, with the protocol recording after execution which resource was consumed. The cost is that funding must be checked conservatively at the highest of the three base fees.
  • Multidimensional subfee market — separate resource limits converted to a scalar at block inclusion using fixed rates. This removes the highest-price funding check but makes opcode costs move, so contracts have to read current rates.
  • Universal overflow — a dedicated limit per resource funded at its own base fee, plus an optional overflow funded at the highest. Legacy calls forward the overflow while dedicated budgets pass separately.
  • Updated EVM — separate budgets carried through execution frames, with direct multidimensional calls and introspection for new contracts and a compatibility layer for old ones.

The trade-off running through all four is the same. Keeping one number preserves every existing contract and forces conservative funding; splitting the number prices resources honestly and breaks the assumption that gas means one thing.

In the thread Vitalik Buterin points out that universal overflow combined with floating prices yields an efficiency neither gets alone, particularly where resources differ in how bursty they are — a case the fixed-rate conversion handles badly.

This is a design discussion rather than a decision, and none of the four is settled. It is worth following for anyone whose contracts do arithmetic on gas: two of the four would make the cost of an opcode a moving quantity, and code written on the assumption that it is fixed would need revisiting.

Retold from Ethereum Research. This is a summary in our own words; follow the link for the original reporting.

Read next

Across the network

Desks that share a zone with this one on the BITBRIEF coverage map.

Terms defined