Validator Node Rewards
Sui uses a delegated proof-of-stake (DPoS) system, where validators receive voting power from SUI token holders who delegate their stake. At the end of each epoch, validators receive rewards based on their performance and commission rate. Rewards auto-compound because validator staking pools receive new delegated stakes.
The total voting power on Sui is always 10,000, regardless of the amount staked. Therefore, the quorum threshold is 6,667. There is no limit to the amount of SUI users can stake with a validator. Each validator has consensus voting power proportional to SUI in its staking pool, with one exception: the voting power of an individual validator is capped at 1,000 (10% of the total). If a validator accumulates more than 10% of total stake, the validator's voting power remains fixed at 10%, and the remaining voting power is spread across the rest of the validator set.
Staking on Sui compared to other blockchainsβ
The Sui staking mechanism includes features that might differ from other blockchains:
- Staking is self-custodial. Stakers keep their staked SUI tokens in an owned object.
- Staking rewards auto-compound because of the liquidity-pool-inspired design.
- A new stake begins contributing to a validator's voting power starting the epoch after it is created. Similarly, a withdrawn stake stops contributing starting the epoch after withdrawal.
Validator staking pool requirementsβ
There are minimum staking requirements a validator must satisfy to become active and to stay in the active validator set.
Stake requirementsβ
The Sui network is rolling out SIP-39, which will significantly lower the barrier to entry for validators. Instead of requiring a minimum amount of SUI tokens, validators will need a minimum amount of voting power.
When fully rolled out, SIP-39 will mean the following validator requirements:
- A validator candidate must accrue at least 3 voting power before they can request to join the validator set.
- If an active validator's stake falls below 2 voting power, they have seven epochs of grace period to gain back the stake before being removed from the validator set.
- If an active validator's stake falls below 1 voting power, they are removed from the validator set at the end of the current epoch boundary. Sui uses 24-hour epochs.
For more information on voting power, see Understanding the voting power formula.
Want to be a Sui validator?
If you have the required stake and plan to operate a validator on Sui, your participation is welcome and Sui is committed to supporting your onboarding. Kindly complete this form to be added to our Validator Discord and keep up with upcoming validator releases and technical support.
User staking and rewardsβ
When users stake SUI tokens, the tokens are wrapped into StakedSUI objects. A userβs share of the staking pool is calculated using the timestamp of their StakedSUI object, which records when the deposit occurred, along with the change in exchange rates between the deposit epoch and the withdrawal epoch.
Each staking pool maintains a time series of its exchange rates. This data allows the system to determine the correct withdrawal amount for any staker in the pool.
Find the exchange rateβ
Each epoch change emits a 0x2::validator_set::ValidatorEpochInfo event per validator with the exchange rate information. You can use the Events API to query events.
Reward distributionβ
A stake deposit request goes into a pending state immediately in the staking pool as soon as it is made. Wallets like [Slush] reflect any pending stake deposit requests for the user's account.
At the end of each epoch, gas fees collected and stake subsidies are distributed among validators and stakers as staking rewards. The rewards a validator receives depend on:
-
Performance: If a validator does not operate performantly, other validators can report them through the tallying rule. The validator's staking rewards for that epoch are then slashed.
-
Commission rate: The commission rate determines the portion of staking rewards the validator keeps. For example, if a validator has a 10% commission rate, then 10% of the staking rewards of their stakers go to the validator each epoch as new stake objects owned by the validator.
Rewards withdrawalβ
Validator rewards are distributed as regular stake objects, so withdrawal works the same for validators and stakers. A validator can call the request_withdraw_stake function in the sui_system module to withdraw stake and receive rewards.
Stake withdrawals are processed immediately with the exchange rate prevailing at the previous epoch's exchange rate. Withdrawals do not have to wait for the current epoch to close. Withdrawals include both the original stake the user deposited and all the stake rewards accumulated up to the previous epoch. Stakers do not earn the rewards accruing to their stake during the epoch at which they withdraw. Since there is no way to know how many stake rewards are accumulated during the current epoch until the epoch closes, these cannot be included in the withdrawal. Hence, any user can withdraw their stake immediately and receive:
Users can't withdraw a portion of their active stake. They must withdraw all staked SUI at the same time. Users can, however, stake using multiple StakedSui objects by splitting their SUI into multiple coins. They can then perform a partial withdrawal from a validator by un-staking only some of the StakedSUI objects.
Staker rewardsβ
Within a given validator staking pool, all stakers receive the same proportion of rewards through the pool's exchange rate appreciation. Because validators earn commissions over the stake they manage, they receive additional StakedSUI objects at the end of each epoch in proportion to the amount of commissions their staking pool earns.
Staking rewards are funded by transaction gas fees collected during the current epoch and by stake subsidies released at the end of the epoch.
Stake subsidies are intended to subsidize the network during its early phases and are funded by a 10% allocation of SUI tokens. After this allocation depletes, the entirety of stake rewards is made up of gas fees collected through regular network operations.
Stake rewards are made up of gas fees and stake subsidies. The total amount distributed throughout each epoch is determined as follows:
- Stake subsidies: The amount distributed in each epoch is determined prior to the beginning of the epoch according to a predefined schedule.
- Gas fees: Each epoch's amount depends on the total gas fees collected throughout the epoch. Each Sui transaction pays gas fees depending on 2 variables, the amount of executed gas units and the gas price:
The total amount of gas fees collected corresponds to the sum of gas fees across all transactions processed in the epoch. During regular market conditions, the vast majority of transactions should have a GasPrice equal to the ReferenceGasPrice.
Reference gas priceβ
Sui is designed such that end-users can expect the gas price to be stable and predictable during regular network operations. This is achieved by having validators set the network's reference gas price at the beginning of each epoch.
Operationally this is achieved through a gas price survey that occurs as follows:
- During each epoch E, each validator submits what they think the optimal reference gas price should be for the next epoch E+1.
- At the epoch boundary, when Sui transitions from epoch E to epoch E+1, the network observes the gas price quotes across the validator set and sets the 2/3 percentile weighted by stake as the epoch's reference gas price. Hence the reference gas price is constant throughout each epoch and is only updated when the epoch changes.
For example, assume that there are 7 validators with equal stake, and the price quotes they submit are {15, 1, 4, 2, 8, 3, 23}. The protocol sets the reference gas price at 8.
In practice, the process for submitting a gas price quote for the Gas Price Survey is a straightforward one. Each validator owns an object that contains their quote for the reference gas price. To change their response, they must update the value in that object.
For example, to set the price quote for the next epoch to 42, run:
Beginning with the Sui v1.24.1 release, the --gas-budget option is no longer required for CLI commands.
$ sui client call --package <PACKAGE-ID> --module sui_system --function request_set_gas_price --args 0x5 \"42\" --gas-budget <GAS-AMOUNT>
Importantly, the gas object's value persists across epochs so that a validator who does not update and submit a new quote uses the same quote from the previous epoch. Hence, a validator seeking to optimize its own operations should update its quote every epoch in response to changes in network operations and market conditions.
Validator slashing and tallying ruleβ
Sui is designed to encourage and enforce community monitoring of the validator set. This is done through the tallying rule by which each validator monitors and scores every other validator in order to ensure that everyone is operating efficiently and in the network's best interest. Validators that receive a low score can be penalized with slashed stake rewards.
The protocol only computes the global tallying rule score at the epoch boundary and so relies on validators monitoring actively and changing their individual scores whenever they detect changes in other validator behavior. In general, the tallying rule default option should always be a score of one for all validators and only be changed to zero upon determining bad operations. In practice, the tallying rule consists of a set of objects each validator owns that default to scores of one and thus a validator is generally passive and only update the object corresponding to another validator's score whenever needed.
For example, to report a validator whose Sui address is 0x44840a79dd5cf1f5efeff1379f5eece04c72db13512a2e31e8750f5176285446 as bad or non-performant, run:
$ sui client call --package <PACKAGE-ID> --module sui_system --function report_validator --args 0x5 0x44840a79dd5cf1f5efeff1379f5eece04c72db13512a2e31e8750f5176285446 --gas-budget <GAS-AMOUNT>
The tallying rule should be implemented through a social equilibrium. The validator set should actively monitor itself and if one validator is clearly non-performant, then the other validators should score that validator with a 0 and slash its rewards. Community members can launch public dashboards tracking validator performance and that can be used as further signal into a validator's operations. There is no limit on the number of validators that can receive a 0 tallying score in an epoch.