By KarenZ, Foresight News

This week, Solana ecosystem re-staking project Solayer unveiled its 2025 roadmap, with the core highlight being the upcoming hardware-accelerated SVM blockchain, “Solayer InfiniSVM”. For Solayer, “Solayer InfiniSVM” is undoubtedly an important part of realizing its long-term vision.

Solayer's chief engineer Chaofan Shou joined the Solayer team after dropping out of his PhD studies at the University of California, Berkeley. He revealed that the team gained a lot of inspiration from the Solana validator client Firedancer during the development process and decided to offload most of Solana's components to SmartNICs and programmable switches. This approach can greatly improve the network's processing power and make transaction processing more efficient.

In Solayer Chain, each transaction follows a set of workflows. The transaction first enters a scalable entry cluster consisting of hundreds of thousands to millions of nodes, which clean and pre-execute the transaction based on probabilistic predictions of future states. Subsequently, all execution snapshots are sent to a sorter built with Intel Tofino switches and additional FPGAs. It is worth noting that most transactions have been confirmed as valid in the pre-execution stage, so there is no need to execute them again on the sorter. For the remaining conflicting transactions, Solayer Chain's sorter will re-execute using the SOTA (State-of-the-Art) scheduling algorithm based on the fine-grained account access patterns collected in the pre-execution to ensure the fairness and efficiency of the transaction.

In terms of performance, Chaofan Shou said that for simple workloads, Solayer Chain can achieve a transaction processing capacity of more than 16 billion transactions per second (TPS); and for conflicting workloads, it can also reach 890,000 TPS. This means that on Solayer Chain, billions of people's USDC transfer requests can be processed every second, and millions of people's transaction needs for the same memecoin on Raydium can be processed.

So how is Solayer InfiniSVM implemented?

How to implement Solayer Chain?

According to the Solayer Chain Lightpaper, Solayer Chain achieves unlimited scalability of single-state blockchains by distributing workloads across dedicated hardware and clusters while retaining the global atomic state.

Solayer said that through SDN (Software Defined Network) and RDMA (Remote Direct Memory Access) connections, 100 Gbps can be achieved while maintaining atomic state. Solayer InfiniSVM achieves 1ms transaction confirmation by offloading to hardware circuits and kernels, spanning incoming, sorting, scheduling, banking, and storage.

Here is a brief overview of the Solayer Chain workflow:

1. Transaction incoming: Each transaction will enter an initial entry point as shown in the upper left corner of the figure below, which will perform signature verification (sigverify) and local deduplication operations (to avoid duplicate transactions);

2. Pre-execution phase: The verified transactions are then sent to the pre-execution cluster for pre-execution.

3. Sorting and Scheduling: Transaction results and intermediate snapshots are sent to the sorter via InfiniBand (which provides a high-speed, low-latency network architecture tailored for high-performance computing and data center environments). The sorter uses SDN switches and FPGAs to decide whether the transaction takes a simple or complex path.

Simple path: If all accounts are up to date at the time of transaction pre-execution, state changes are applied directly via RDMA (Remote Direct Memory Access), using local cache on the SDN, avoiding further processing by the sequencer.

Complex path: If at least one account has an updated version, the transaction will enter the local memory pool. The sorter schedules transactions in the local memory pool to achieve fair and optimal parallel execution of all transactions.

4. Status update: The transaction status changes after execution will be updated to the shard database. The shard database uses the RDMA protocol to achieve efficient cross-node data access.

5. Transaction broadcast: After the transaction is executed and the state change is written, the transaction will be broadcast through global PoPs (Point of Presence, network access points).

Is the era of millisecond transactions coming? A quick overview of Solayer Chain

In terms of consensus mechanism, Solayer Chain adopts Proof-of-Authority-and-Stake hybrid consensus protocol to batch transactions into fragments (shreds), each of which contains slot number, transaction vector, version metadata of access account and link hash. Trusted entities act as sorters and publish fragments, and provers all stake and vote to decide whether the fragments can be accepted.

It is worth mentioning that Solayer Chain not only focuses on performance, but also introduces a number of user experience improvements, especially chain-level support, such as:

  • Hooks: Allow developers to embed post-trade logic such as arbitrage, liquidation, and accounting directly into the chain.
  • Large transactions: Support larger transaction sizes and allow cross-program calls.
  • Cross-chain call: Atomic cross-chain operations are achieved through built-in system programs.
  • Built-in OAuth support: Allows users to use OAuth services like Google, X, or Reddit as wallets.

Here we emphasize Hooks. Hooks allow developers to embed logic such as arbitrage, liquidation, and accounting directly into the chain. In addition, Solayer Chain has set up an incentive and fee model for hooks. The execution of Hooks adopts a bidding model similar to the Dutch auction. If a developer or user wants to attach a Hook to a specific program, he or she needs to bid (bid epoch by epoch for the right to execute the hook in the next epoch). The bid price determines whether the Hook can be executed and the priority of execution. The first 16 bidders with the highest bids win.

Each time a Hook is executed, its bid amount is distributed in the following proportions:

  • 40% is allocated to the transaction initiator.
  • 40% is allocated to program owners to encourage them to develop and maintain high-quality programs.
  • 20% is allocated to the network to offset additional on-chain computational overhead.

The above model of allocating bidding fees to transaction initiators and program owners will encourage more developers and users to participate in the use of Hooks. This not only improves the functionality of the platform, but also increases the activity of the network. Through this, Solayer Chain can also effectively prevent spam transactions or malicious MEV utilization off-chain, providing an additional layer of protection for the network.

The vision of Solayer Chain is not only to improve performance, but also to integrate more user experience and developer-friendly features into blockchain technology.