- Firedancer, a high-performance Solana validator client developed by Jump Crypto, proposed to remove the current 60 million Compute Unit (CU) per-block limit (SIMD-0370).
- The goal is to boost network throughput following the upcoming Alpenglow upgrade by allowing block size to scale dynamically based on validator hardware capacity.
- This dynamic scaling is intended to incentivise a hardware “arms race” among validators, where those with superior hardware earn higher fees, but raises concerns about potential centralisation risks.
Firedancer, Solana’s high-performance validator client developed by Jump Crypto, has put forward a proposal to remove Solana’s per-block compute unit (CU) limit, aiming to boost throughput following the network’s upcoming Alpenglow upgrade.
The proposal, labeled SIMD-0370, would scrap the current 60 million CU ceiling per block. A competing idea had suggested raising the cap to 100 million. Under Firedancer’s version, block size would scale dynamically with the hardware capacity of validators, and weaker validators would skip voting on oversized blocks under the skip-vote mechanism built into Alpenglow.
The team argues this would create a hardware “arms race” among block producers. Validators capable of handling larger blocks would earn higher fees, pushing others to upgrade their infrastructure. This cycle, according to the proposal, would steadily expand Solana’s average block capacity.
Related: Novogratz Declares ‘Season of Solana’ as Markets and Regulators Turn Favourable
Some Concerns and Support
Roger Wattenhofer, head of research at Anza, which developed Alpenglow, welcomed the idea but cautioned about centralisation risks, noting that a single validator with disproportionate resources could destabilise the network, though he maintained the challenges are solvable and reiterated support for removing the cap.
It would help tremendously if we made sure that the leader’s work (to put together the block) is bigger than the work of replaying/checking by the followers. E.g., the leader might give advice for parallel execution, or even stuff like rollups. That way the leader still has an incentive to go hard, but we have a sane system at the same time.
Roger Wattenhofer, AnzaAlpenglow itself passed governance with near-unanimous approval and is set to reduce block finality from 12.8 seconds to 150 milliseconds, alongside broader resilience and data upgrades. The idea is to achieve Web2 responsiveness while maintaining layer-1 security guarantees.
Related: CME Group to Launch Options on Solana and XRP Futures in October
Credit: Source link
