Recently, discussions around Solana's 100K TPS have resurfaced. The reason is that @cavemanloverboy indeed achieved over 100,000 TPS on the Solana mainnet. However, most people haven't fully grasped the significance behind this data:
-
First and foremost, cavey's experiment was essentially a stress test under "ideal conditions." This means it does not represent the normal performance of the Solana mainnet. While there might be slight differences compared to lab data from testnet environments, they are generally similar.
He used a "noop" (no operation) test program. As the name suggests, it only performs the most basic signature verification and immediately returns success—it executes no computations, changes no account states, invokes no other programs. Furthermore, each transaction was only 200 bytes, far less than the 1kb+ of a normal transaction.
This means that this 100K TPS test was calculated in an abnormal transaction environment. It tested the极限吞吐量 (limit throughput) of Solana's network and consensus layers, not the actual processing capability at the application layer.
-
Another key to the experiment's success was the Frankendancer validator client. Simply put, Frankendancer is a "hybrid test version" of the Firedancer validator being developed by Jump Crypto—it grafts the high-performance components already completed for Firedancer onto the existing Solana validator.
Essentially, it restructures Solana's node system using the high-frequency trading (HFT) tech stack from Wall Street, achieving performance improvements through low-level optimizations like refined memory management and custom thread scheduling. Merely replacing some components resulted in a 3-5x performance boost.
-
This test experiment shows that Solana can achieve 100k+ TPS under extremely ideal conditions. So why is the daily TPS only around 3000-4000? Simply put, there are roughly three reasons:
-
Solana's Proof of History (PoH) consensus mechanism requires validators to constantly vote to maintain it. These vote transactions alone occupy over 70% of the block space, narrowing the performance channel left for normal transactions.
-
Ecosystem activities on Solana often involve significant state contention. For example, when minting a new NFT or launching a new MEME coin, thousands of transactions might compete for write access to the same account, leading to a relatively high proportion of failed transactions.
-
Arbitrage bots within the Solana ecosystem, aiming to capture MEV (Maximal Extractable Value), may send a large number of invalid transactions, constituting resource wastage.
-
However, the upcoming full deployment of Firedancer and the Alpenglow consensus upgrade will systematically address these issues.
A key point of the Alpenglow consensus upgrade is moving vote transactions off-chain, effectively freeing up ~70% of the space for normal transactions, while also reducing confirmation time to 150 milliseconds. This will make the experience of using Solana DEXs infinitely closer to that of CEXs. Additionally, enabling local fee markets can prevent a single program's FOMO-induced popularity from causing network-wide congestion.
Besides performance optimization and improvements, a key benefit of Firedancer is achieving client diversity. It allows Solana to have multiple clients (like Geth, Nethermind for Ethereum), directly improving decentralization and resilience against single-node failures.
That's all.
Therefore, those who understand the industry see the discussion about Solana's 100K TPS as confidence in its future client and consensus protocol upgrades. Those who don't try to use the TPS arms race to garner attention for Solana (even though competing on TPS is already outdated). But understanding the meaning behind the experiment is quite insightful. Just sharing some knowledge with everyone.
