Skip to main content
SUBMIT A PRSUBMIT AN ISSUElast edit: Sep 04, 2024

Chain Release Notes

The following are the release notes for the subtensor on-chain software.

Mainnet​

Released 04 September 2024

New features​

  • Child hotkey take: With this feature, you can now set a take rate for child hotkey delegation that is different from the take rate for delegation by nominator. In addition, there is no maximum limit on a hotkey's total stake (i.e., no maximum limit on a hotkey's sum of delegated stake + child hotkey stake).

  • Schedule coldkey swap: Schedule a coldkey swap, and it always executes the actual swap five (5) days from the time of scheduling.

    No btcli support yet

    The Bittensor software does not yet support this schedule coldkey swap feature.

Enhancements​

  • Change in the emission schedule: To reduce the chain bloat, we are changing the emission schedule. The new emission schedule will be once every 24 hours (7200 blocks). The keys will continue to accumulate the emissions at every tempo as usual. With this new emission schedule, staking returns are auto-compounded every 24-hours (7200 blocks).

    Elaboration: Every time the emit_inflation_through_hotkey_account function is executed on-chain, it increases the size of the chain state. The chain currently runs this function every epoch, thereby significantly contributing to the chain bloat. To mitigate the above chain bloat issue, the OTF will make a change so that the above emit_inflation_through_hotkey_account function runs once every 20 epochs (every 7200 blocks), instead of every 360 blocks. As a result, this will reduce the chain bloat by a factor of 20.

  • Single tempo delay: We are introducing a single tempo delay between when you increase your stake and begin accumulating the corresponding staking returns. In specific, if you increased your stake during any block in the current tempo, you will start accumulating your staking returns only in the next tempo. For example, if you stake at tempo=0, then you will not receive any staking returns for this tempo=0. Your stake will only start to accumulate returns starting at tempo=1.

  • New proxy types: Added the transfer proxy type in two flavors for finer-grained control. This allows for separate proxies for small and large sum transactions.

  • Sensible runtime logging: Reduced the amount of logging and clarified the reduced output.

Fixed issues​

  • Better neuron pruning: A neuron with the worst pruning score is pruned first, but if multiple neurons have the same worst pruning score, then the oldest registered neuron is pruned first.

Mainnet 1.2.3​

Released 12 July 2024

Mainnet 1.1.3 hotfix​

Fixed issue​

A bug is fixed where, when a subnet's registration is turned off, the chain was not running the epoch for that subnet. The hotfix fixed this bug by running the epoch for the subnet and setting the emissions to zero, for that subnet.

Testnet 1.2.1-pre-release​

New features​

Consensus-based weights​

A new feature, called consensus-based weights is available in the testnet 1.2.1-pre-release. Currently while calculating the dividends to a subnet validator, a quantity called exponential moving average bond value of a subnet validator, weighted over the current epoch and several previous epochs, is used. See the below equation for how this moving average is computed.

Bij(t)=α⋅ΔBij(t)+(1−α)⋅Bij(t−1)B_{ij}^{(t)} = \alpha\cdot\Delta B_{ij}^{(t)} + (1-\alpha)\cdot B_{ij}^{(t-1)}

Until now, the value of α\alpha is set to 0.9. With this new feature, the α\alpha value is made into a variable. An optimium value for the variable α\alpha is determined based on the current consensus in a given subnet. Hence this feature is called consensus-based weights. A subnet owner can now experiment for the optimum value of α\alpha. See the below documentation links for how to use this feature:

Documentation links for consensus-based-weights feature

Detailed documentation and how to test the consensus-based weights feature (including a Python notebook for subnet owners): See Consensus-based Weights.

Tech blog post with analysis: https://blog.bittensor.com/consensus-based-weights-1c5bbb4e029b

Mathematical context: Validator bonding document section in the subtensor repo.

See the PR: https://github.com/opentensor/subtensor/pull/552

Safe Mode​

A new feature, called Safe Mode, is added to the subtensor. This feature is added in the form of a Rust pallet, called Safe Mode pallet, to the subtensor runtime. This pallet allows a highly privileged user such as root to put the entire chain into "safe mode" for a set period of time, whereby only a select few extrinsics can be executed, while all other extrinsics will be rejected. While in safe mode, the privileged user can also choose to extend the duration of the safe mode period.

See the PR: https://github.com/opentensor/subtensor/pull/553

Added freeze_struct attribute macro​

This feature is added to prevent accidental changes to storage structs. An attribute called freeze_struct is added to the storage structs and other change-sensitive Rust items. This attribute will take a single argument which is a hash code of the non-trivial (ignoring doc comments) AST nodes that make up the item the attribute is attached to.

If the hash code is not provided: A compiler error will be displayed providing the programmer with the proper hash code which they can then paste in.

If the hash code does not match: A compiler error will be displayed informing the programmer that they have made a non-trivial or structural change to a storage struct and that they will likely need to write a migration to enact this change. The correct hash code will also be displayed.

See the PR: https://github.com/opentensor/subtensor/pull/533

Fixed issues​

Current root members could not become a senate member​

The following bug is fixed: Current root members could not become a Senate member upon attaining the requirements after registering on the root network, initially. This is fixed in the below PR.

See the PR: https://github.com/opentensor/subtensor/pull/537

Subtensor error: Allocator ran out of space​

The following bug is fixed: When running a subtensor archive node, the node stops syncing blocks with the below error:

Failed to allocate memory: "Allocator ran out of space"

See the PR: https://github.com/opentensor/subtensor/pull/561

11 June 2024: Release notes for mainnet 1.1.1​

28 May 2024: Release notes for testnet 1.1.0​

The following are the release notes for the testchain 1.1.0 updates. These updates will be available on the testchain only. They are not yet available on the mainchain.

Breaking change​

If you are using the function set_weights() directly, then note that this function is no longer in use. In its place, use the new function set_root_weights() to set the root weights. This new function now requires that you sign it with your coldkey. The now-removed set_weights() function was signed by the hotkey. See line of code for the new set_root_weights().

Dependencies version change​

Polkadot SDK version is updated to 1.10 with this testchain update.

New feature​

Commit reveal​

Pull request: https://github.com/opentensor/subtensor/pull/396

A new feature, called, commit reveal, is made available on the testchain only, starting with testchain 1.1.0. This feature will address the weight copying issue.

For a detailed description of the feature and how to test it, see the Commit Reveal document.

Enhancement​

  • Emissions are now recycled for those subnets that have registration turned off.
  • The btcli root weights command now uses a new function set_root_weights() behind the scenes and uses the coldkey to sign the transaction.

Pull request: https://github.com/opentensor/subtensor/pull/474

Fixed issue​

Previousy, due to a bug in Yuma Consensus implementation, the weights were not normalized before calculating subnet emissions. This bug is fixed in this update, so that the weights are now normalized before calculating subnet emissions.

Pull request: https://github.com/opentensor/subtensor/pull/335