Understanding the EVM: A Comprehensive Guide to the Ethereum Virtual Machine

Introduction to the Ethereum Virtual Machine (EVM)

What is the EVM?

The Ethereum Virtual Machine (EVM) is a stack-based computational engine that executes smart contract bytecode on the Ethereum blockchain. It processes instructions (opcodes) sequentially, managing state changes, gas fees, and contract interactions.

• Key Features:

  • Stack-based architecture: Operands are stored and retrieved via a last-in-first-out (LIFO) stack.
  • Deterministic execution: Ensures identical results across all nodes.
  • Gas mechanism: Charges fees for computational work to prevent spam.

Smart Contracts Explained

A smart contract is a self-executing program composed of opcodes (e.g., PUSH1, JUMP). When deployed, these contracts run on the EVM, enabling decentralized applications (dApps).

• Critical Notes:

  • Failed executions revert state changes (except gas consumption).
  • REVERT opcode refunds unused gas, unlike other failures.

EVM Execution Environment

Core Components

1. Code: Immutable bytecode stored on-chain. Accessed via CODESIZE/CODECOPY.
2. Program Counter (PC): Tracks the next instruction. Modified by JUMP/JUMPI.
3. Stack: Holds 32-byte values for opcode inputs/outputs (max 1024 entries). Managed by DUP1, SWAP1, etc.
4. Memory: Volatile byte array for runtime data (cleared post-execution).
5. Storage: Persistent key-value store (SLOAD/SSTORE).
6. Calldata: Immutable input data (CALLDATALOAD).
7. Return Data: Output from calls (RETURNDATACOPY).

Gas Costs and Optimization

Fee Structure

• Base Costs:

  • Intrinsic gas: 21,000 gas per transaction (+32,000 for contract creation).
  • Calldata: 4 gas per zero byte; 16 gas per non-zero byte.

• Dynamic Costs:

  • Memory expansion: Quadratic pricing based on accessed offsets.
  • Access lists: Cold vs. warm storage slots impact costs post-Berlin hardfork.

Memory Expansion Formula

memory_cost = (memory_size_word² / 512) + (3 × memory_size_word)

(Where memory_size_word = (byte_size + 31) / 32)

Advanced Topics

Gas Refunds

• Post-London Hardfork:

  • Max refund capped at 20% of total gas.
  • Only SSTORE (not SELFDESTRUCT) triggers refunds.

Access Lists

• Warm/Cold States: Addresses/slots accessed repeatedly cost less.
• Pre-warmed: Includes tx.sender, tx.to, and precompile addresses.

FAQ Section

1. How does the EVM handle transaction reversions?

• Answer: Reverted transactions undo state changes but consume gas (except for REVERT opcode refunds).

2. What’s the difference between memory and storage?

• Answer: Memory is temporary and cheaper; storage is persistent and costly.

3. Why do gas fees vary?

• Answer: Fees depend on opcode complexity, memory usage, and network demand.

4. How can I reduce gas costs?

• Answer: Optimize calldata, use warm access slots, and minimize storage writes.

👉 Explore EVM opcodes in-depth
👉 Master gas optimization techniques

For further reading, consult the official EVM documentation.