How does Solayer achieve 340,000 TPS with RDMA under hardware support?

I heard that Solayer has launched the InfiniSVM Devnet, and it immediately showcased a peak of over 340,000 TPS in a usable testing environment? To be honest, this is not a gradual optimization in the Solana ecosystem where it is still at 4,000 TPS and PumpFun users are still suffering from transaction failures; rather, Solayer's solution is a direct leap in magnitude.

Where does the confidence behind this come from?

1. The confidence of InfiniSVM comes first from a reality in the industry: the pure software optimization route has nearly reached its limit, and hardware acceleration is timely.

In the past few years, improvements in blockchain performance have mainly relied on architectural innovations—from Bitcoin's UTXO to Ethereum's account model, from PoW to PoS, and from single chain to L1+Layer2 modular stacking. These are all breakthroughs at the software level. But now this path is becoming increasingly narrow.

The reason why most high-performance chains are stuck at the threshold of tens of thousands of TPS is due to this.

Aptos claims that 160,000 TPS is mostly a theoretical value, and Sui also finds it difficult to maintain full capacity in practical applications, with the fundamental reason being the physical bottleneck under a general hardware architecture.

Traditional CPU serial processing, network I/O latency, and memory access overhead are becoming the biggest obstacles to performance improvement due to these hardware-level limitations.

The RDMA (Remote Direct Memory Access) hardware acceleration technology route chosen by InfiniSVM essentially bypasses the CPU bottleneck, allowing for direct memory-level communication between nodes. In addition, InfiniSVM has also introduced a multi-actor parallel processing model, combined with SDN (Software Defined Networking) for real-time traffic optimization. These are all possibilities for seeking new breakthroughs from the hardware level and represent an important shift in industry development.

2. InfiniSVM is fully compatible with the Solana Virtual Machine, which makes migrating to InfiniSVM basically a matter of "changing RPC endpoints" for developers who are already deeply involved in the Solana ecosystem. The commercial value of this compatible design is self-evident. Traditional blockchains are limited by the TPS upper limit, and many application scenarios can only stay at the concept stage. For example, high-frequency algorithmic trading, real-time game state synchronization, etc. The finality of InfiniSVM's confirmation (0.01 second confirmation time) is a breakthrough in many application scenarios: real-time interaction in chain games, millisecond-level settlement in DEX, and high-frequency automated trading with AI Agent. This is key to breaking through the limitations of current Solana application scenarios.

3. The hybrid POAS consensus model adopted by InfiniSVM attempts to find a balance between performance and decentralization. Daily transactions are processed quickly through a network of validators, and in cases of disputes or anomalies, the Solana mainnet serves as the final arbiter. This design concept of "fast track + insurance mechanism" is quite pragmatic.

In fact, this design idea is not new—early designs of Polygon and various sidechain solutions have similar logic. The problem is that hardware acceleration schemes naturally raise the operational threshold for nodes. While RDMA and InfiniBand technologies offer strong performance, their costs and technical complexity also increase accordingly, which will inevitably lead to the "centralization" issue of validation node networks, similar to most high-performance chains.

This logic is very realistic: since the hardware acceleration route will inevitably lead to the centralization of the validator network, we can borrow a mature decentralized network as the ultimate security guarantee. It is equivalent to layering the two requirements of "performance" and "security"—InfiniSVM is responsible for extreme performance, while Solana is responsible for ultimate security.

The above. It should be noted that the current Devnet is still in the internal testing phase, the blockchain state may be reset from time to time, and network data may be intermittently unstable, all of which indicate that there is still a considerable amount of engineering work to be done before production readiness. This is especially true when truly facing the extreme pressure of 1 million+ TPS.

Overall, InfiniSVM represents an important directional shift in blockchain infrastructure—from software optimization to hardware acceleration, from theoretical innovation to engineering implementation. The confidence of Solayer can simply be described as a forward-looking choice of "technical pathway". (As shown in the figure below, the real-time TPS is nearly 100,000)