Ethereum block execution isn't a fixed-cost operation. For small blocks the state LRU cache handles nearly everything. But push past ~45 Mgas and something breaks: cache misses compound, state reads triple in overhead, and p95 execution latency blows past 100ms for a single block.
Nobody talks about this because mgas/s benchmarks measure throughput — not the hidden cost of cold cache reads. The gas limit doubling from 30M to 60M made this matter.
I started this looking for evidence that high-MEV blocks are harder for execution clients to process. The intuition is obvious: MEV blocks are full of complex DeFi interactions, sandwich attacks, arbitrage — all the state-thrashing stuff. Surely they're heavier to execute.
They're not. The correlation between MEV block value and newPayload execution time is r = −0.004. Essentially random noise.
What actually predicts execution latency is simpler and more boring: how much gas the block used.
Analyzing 52,104 blocks over 7 days: mean interval is 12.05s, median is 12s. Only 0.38% of blocks are delayed beyond 12 seconds. The network maintains remarkably tight timing.