Back to blog
June 17, 2024 · 7 min · SQD Team

Hype or Reality? A Look at Parallel EVMs

EVM Scalability Blockchain
Hype or Reality? A Look at Parallel EVMs

Introduction

The EVM dominates blockchain infrastructure, powering roughly 60% of all DeFi transactions. Yet sequential transaction processing creates bottlenecks during network congestion. Parallel processing emerges as a potential solution to increase throughput while reducing transaction competition.


Why Parallelization?

Parallel computing isn’t new — the concept dates to 1837 with technical implementations emerging in the 1950s. It divides complex tasks into manageable chunks using multiple processing elements simultaneously.

A Monad representative noted that parallel processing has become ubiquitous: “You can’t even buy a single-threaded computer today.”

Parallel EVMs simply run the Ethereum Virtual Machine in parallel, distinct from Layer 2s that operate separate state systems.


How Blockchain Transactions Currently Work

The standard process involves:

  1. User signs transaction
  2. Transaction enters mempool
  3. Block proposer includes it in a proposed block
  4. Validators verify and confirm
  5. EVM executes blocks sequentially for compatibility assurance

This sequential design prevents conflicts but creates congestion during peak activity, driving up fees.


Two Approaches to Parallel Execution

Deterministic Parallel Execution

Solana exemplifies this model. Transactions must declare all memory dependencies upfront. The network identifies independent transactions and reschedules dependent ones.

Solana’s Sealevel runtime enables validators to process multiple transactions simultaneously. A NEON EVM representative explained that with NEON EVM, they’re tapping into the parallelization of Solana by running the EVM as a smart contract.

Optimistic Parallel Execution

Chains like Canto and Monad assume all transactions are independent, only re-executing them if conflicts emerge.

Monad uses optimistic concurrency control. According to their team, transactions run in parallel but are not committed immediately. Inputs and outputs are compared, and dependencies are addressed accordingly.


The Real Bottleneck: State Access

Speed improvements don’t solely come from parallel processing. A Subsquid representative emphasized that the real bottleneck of TPS is not block size but transaction execution, specifically how fast you access required storage slots.

Monad invested nearly a year building a custom database supporting parallel execution. Their system uses advanced Linux kernel features enabling asynchronous disk operations — allowing the next transaction to process while waiting for prior transaction state data.

Canto splits its upgrade into three phases: optimistic execution, parallel processing, and finally state access optimization.


Parallelization vs. Sharding

A critical distinction exists between these approaches:

  • Parallelization: Maintains unified blockchain state while processing transactions in parallel
  • Sharding: Splits state across separate shards

Sharding introduces synchronization lag that harms composability and prevents certain DeFi applications. Industry observers strongly prefer parallelization, with one noting that sharding does not solve the issue of noisy neighbors where one successful contract clogs the network.


Why Choose EVM?

Despite its flaws, the EVM dominates because of:

  • Developer ecosystem: 5-6 times more developers on Ethereum than Solana
  • Existing tooling: Battle-tested frameworks and infrastructure
  • Network effects: Established community and research
  • Accessibility: Similar to JavaScript — not optimal but widely understood

A NEON EVM representative stated that they were following the trend when they chose to run EVM, noting there are already many devs using Solidity and all the tools needed to make it easier.


Current Status

Reality lags behind the hype:

  • Solana has faced congestion despite parallelization
  • SEI and similar networks haven’t gained significant traction
  • Monad remains pre-testnet
  • Canto hasn’t fully transitioned to parallel execution

However, industry observers expect parallelization to become standard infrastructure within five years, with focus shifting from the technology itself to applications it enables.


Conclusion

Parallel EVMs represent a natural evolution in blockchain architecture, mirroring how modern computing already operates. While current implementations remain immature, the fundamental approach addresses legitimate scaling challenges facing sequential blockchains.

Want to learn more about SQD?