Data-Driven Insights into SupraSTM’s Hybrid Parallel Execution Approach

In our previous article on Specification-Aware SupraSTM, we introduced our groundbreaking approach to parallel transaction execution that promises to improve blockchain performance. But revolutionary claims require solid evidence. That’s why we conducted the industry’s first comprehensive analysis of historical transaction data across Ethereum and Solana — two of the world’s leading blockchain networks.

The results are in, and they don’t just support our approach — they make an overwhelming case for it. Our findings reveal untapped potential for parallelism that existing systems aren’t fully leveraging, along with critical insights into how transaction conflicts have evolved over time. Let’s dive into what we discovered and why it matters for the future of blockchain performance.

Why Transaction Conflicts Matter

Before we jump into the data, let’s recap why this matters. Blockchains typically process transactions in a strict sequence – one after another – to ensure all nodes reach identical states. This approach guarantees consistency but severely constrains throughput.

Consider a simple example: Two users interact with a popular DeFi pool.

• Transaction 1: Alice deposits 5 ETH into the pool, changing its total liquidity
• Transaction 2: Bob calculates his rewards based on the pool’s total liquidity

These transactions conflict because they both access the same state (the pool’s liquidity), and one modifies it. If executed in parallel without proper management, Bob might calculate rewards based on outdated information.

Meanwhile, a third transaction where Charlie transfers tokens to David involves completely different states and could safely run in parallel with either of the above transactions.

Identifying this simple pattern of which transactions truly depend on each other can help maximize blockchain performance. The key challenge is identifying which transactions can safely run in parallel without creating inconsistent states. This depends on understanding “conflicts” — when two transactions interact with the same data and at least one of them modifies it.

Ethereum and Solana take fundamentally different approaches to this problem:

• Ethereum processes transactions sequentially, with no advance knowledge of what state each transaction will access.
• Solana requires clients to specify upfront what states (accounts) a transaction will read or write.

Our analysis examined both networks to understand the true nature and distribution of these conflicts – critical information for optimizing execution.

A Historical Analysis of Ethereum and Solana

To truly understand transaction conflicts, we needed to go beyond theoretical models and look at real-world data. We analyzed number of blocks across three distinct periods for each network:

For Ethereum:

• The CryptoKitties launch period (Blocks 4604664-4605670) – The first major congestion event
• The Ethereum 2.0 Merge (Blocks 15536879-15537907) – A fundamental protocol transition
• Recent blocks (Blocks 21631000-21632001) – Current network behavior

For Solana:

• Old historical period (Blocks 61039000-61040210)
• Mid historical period (Blocks 205465000-205466007)
• Recent blocks (Blocks 293971000-293972009)

This approach allowed us to see not just the current state of transaction conflicts but how they’ve evolved alongside network growth and adoption and how it could evolve over time.

The Ethereum Revelation: 60%+ Transactions Can Be Parallelized

Our analysis of Ethereum blocks revealed something remarkable: consistently across all time periods, over 50% of transactions are completely independent and could theoretically be executed in parallel.

Some key findings:

• In recent blocks, 51.7% of transactions are independent (can be parallelized)
• The “longest chain” of dependent transactions (those that must run sequentially) is only about 16-17% of block content
• Smart contract transactions show higher conflict rates than simple ETH transfers
• More than 94% of recent blocks have at least 40% independent transactions
• Block-by-block variation is significant, suggesting the need for adaptive execution strategies

This data confirms what we suspected: traditional sequential execution is leaving enormous performance potential untapped. With proper conflict detection, Ethereum could theoretically achieve a significant speedup without any protocol changes.z

Solana’s Parallel Challenge: The Write-Write Problem

Solana’s architecture explicitly supports parallel execution, but our analysis revealed some surprising limitations:

• Only about 7-9% of transactions in recent blocks are truly independent
• Write-write conflicts dominate, accounting for over 95% of all conflicts
• Transaction success rates have declined from 85% to 54% over time
• Conflict patterns have become more granular, with conflict “families” increasing from 3 to 39

The most striking finding is that while Solana’s architecture enables parallel execution, the high rate of write-write conflicts significantly limits actual parallelism. This matches our experience that many transactions end up serialized in practice, creating bottlenecks.

What This Means for SupraSTM

These findings directly validate the approach we’ve taken with SupraSTM. Our hybrid model addresses exactly the challenges revealed by this historical data:

1. Pre-analysis of dependencies: Our Conflict Analyzer identifies which transactions can safely run in parallel before execution begins, leveraging the 50%+ independence rate in Ethereum-style workloads.
2. Multi-version approach: SupraSTM’s design specifically addresses the write-write conflict problem that hampers Solana’s performance.
3. Adaptive execution: The significant block-by-block variation we observed confirms the need for our adaptive approach that can dynamically select optimal execution strategies.

In essence, this historical analysis doesn’t just support our approach — it suggests we’re addressing exactly the right problems in exactly the right ways.

Real-World Implications: Beyond the Numbers

What does this mean for users, developers, and the broader ecosystem?

For users, these optimizations translate directly to faster transaction finality and reduced congestion, especially during high-activity periods when performance matters most.

For developers, our approach means they can build more complex applications without worrying about the explicit requirements of read-write specifications and conflict synchronization — the system handles this automatically.

For the ecosystem, these findings point toward a future where blockchain throughput is dramatically improved without sacrificing the security and determinism that make blockchains valuable in the first place.

Looking Ahead: Building the Future of Execution

This historical analysis represents a crucial step in our journey to optimize blockchain performance. The data provides clear direction for future development:

• Adaptive execution strategies that respond to the specific conflict patterns in each block
• Enhanced multi-version concurrency techniques to mitigate write-write conflicts
• Cross-VM compatibility to bring these optimizations to multiple blockchain ecosystems

We’re already incorporating these insights into our implementation of SupraSTM, and we’re excited to see how these optimizations will transform blockchain performance in the real world.

Join the Performance Revolution

As we continue to develop and refine our approach, we invite the community to engage with these findings and their implications. The data is clear: there’s enormous untapped potential for a hybrid approach to parallelism in blockchain execution, and SupraSTM is designed from the ground up to capture it.

Stay tuned for more updates as we move from research to implementation, and the theoretical speedups demonstrated in our analysis become real performance improvements for users and developers.

For those interested in the technical details, our full research paper “Blockchain Transaction Conflicts: A Historical Perspective” provides comprehensive methodology and findings. Together with our Specification-Aware SupraSTM whitepaper, these resources paint a complete picture of the future of blockchain execution performance.

---

Note: The implementation of these execution optimizations is still in the research and development phase. While we will ship Supra EVM soon, these specific optimizations will be incorporated in upcoming network upgrades after further validation and thorough testing.