Parameter Description

• n_days: Loockback period. If set to 0, start_date and end_date are used.

• start_date: Data collection start date.

• end_date: Data collection end date.

• binning: It is used to bin the empirical PDFs. Higher binning means higher smoothness of the plots.

Compute Units

All the operations performed on-chain within a transaction require different amounts of computation resources be expended by validators when processing (compute cost). The smallest unit of measure for the consumption of these resources is called a "compute unit", ref.

Consensus has a limit of a maximum Compute Unit of 48M per slot.

Why the PDF matters?

Studying the empirical Probability Density Function (PDF) is crucial to understand the true distribution of the share of maximum compute units used per slots, identify patterns, and detect anomalies. Since the distribution is often skewed, relying solely on averages can be misleading, while the PDF reveals the underlying dynamics and variability of the data.

Dynamical Evolution

The compute unit used per slot for validators could be influenced by market conditions. Market conditions impact transaction volume and hence the possible size of a slot. This is reflected in the asymmetrical distribution of share of Max CU. Thus, sporadic high-value events can bias the metric. Analyzing the trend of the percentiles over time is crucial to account for these dependencies and identify patterns, outliers, or anomalies, enabling more accurate assessments of the ecosystem’s economic dynamics.

The Forbidden Region

Monitoring slots that exceed the 48M CU limit is crucial to ensure network stability. This helps to detect client modifications exploiting consensus failure points. Such changes may bypass execution limits, undermining fair resource allocation and network integrity. Identifying these cases ensures validators adhere to protocol rules, preventing unfair advantages and maintaining Solana’s security.