FWSS Gas & Cost Calculator

Gas fee parameters

Base fee (attoFIL/gas)

Set by network. Today ≈ 100.

Gas premium (attoFIL/gas)

Tip to SP. Min ~100,000. Avg 30d: 252,105.

Gas fee cap (attoFIL/gas)

Max sender pays. Must be ≥ BaseFee + Premium.

Effective gas price

= BaseFee + min(Premium, FeeCap − BaseFee)

—

attoFIL/gas

Formula (Lotus source): effective_gas_price = BaseFee + min(GasPremium, GasFeeCap − BaseFee)
Total tx cost = GasUsed × BaseFee (burned to f099) + GasUsed × min(GasPremium, GasFeeCap − BaseFee) (to SP) + overestimation burn (penalty for over-declaring GasLimit by >10%).
Today (pre FIP-0115): SP premium dominates (~99.9% of fee). Under FIP-0115 (Draft): BaseFee rises to ~90% of EGP.

Regime: BaseFee ≈ 0.04% of EGP today

Historical effective gas price (from real FWSS transactions)

24h avg

—

30d avg

—

90d avg

—

Min observed

—

Max observed

—

CDN dataset

CDN enabled

3 FilecoinPay rails instead of 1.
createDataSet: +45M (~840M total)
terminateService: +105M (~241M total)

FIL price (USD)

$ / FIL live

Observation period

days

Minimum floor price — minimum a client must pay per dataset per 30 days to cover SP daily proving gas.
Formula: (gas_PP(N) + gas_NPP) × effective_gas_price ÷ 10¹⁸ × FIL_price × 30
—

Gas per operation (30d avg · live from FOC Observer)

addPieces batching savings (observed mainnet data)

Batch size	Gas/tx	Gas/piece	Saving vs ×1	Source

Beyond ×10: no mainnet observations. Model extrapolated: 228.72M + 78.20M × K. Beyond ×310: tx exceeds tipset budget → stuck forever.

Usage parameters — define your volume · gas values above are unit costs · costs below = unit × volume

Datasets → proving costs

Number of active FWSS datasets

Pieces per dataset → provePossession gas

Drives gas via log₂(N) model

Pieces to add → addPieces cost

Total uploads during period

Batch size → pieces per addPieces tx

Larger = cheaper per piece. K>10: extrapolated. K>368: stuck.

⚠ Beyond ×10: no mainnet data — linear extrapolation (228.72M + 78.20M×K). Beyond ×310: tx exceeds tipset budget → stuck forever.

Pieces to remove → removePieces cost

Removals per period (per tx)

Observation period → multiplies all costs

How many days to model

Cost breakdown

—

Operation	Txns	Gas/tx	Total gas	FIL cost	USD
Total			—	—	—

Burn + SP tip breakdown per operation (per single tx)

Daily gas (all datasets)

—

Daily cost (FIL)

—

Per dataset / month

—

proving only

Cost per piece added

—

USD cost by operation

Complete gas reference — all operations

30d avg from FOC Observer

Operation	Avg gas (30d)	FIL/tx	USD/tx	Who pays	When
Loading...

Methodology — how the models work click to expand

provePossession model

Gas scales logarithmically with piece count N (number of Merkle roots). provePossession traverses a Sum Tree of depth log₂(N) on-chain — 5 challenges × O(log₂N) SLOAD operations. Piece size (leafCount) is irrelevant — it's a value in Sum Tree nodes for proportional sampling, not tree depth.

gas = 158.67M + 8.485M × log₂(pieces)
R² = 0.9551 · MAE = 10.4M gas · 254 real mainnet datasets · refitted daily

Binned averages from real mainnet datasets (FOC Observer). Note: the model tends to overestimate gas for medium piece counts — the real distribution is flatter than the log₂ curve suggests, possibly due to SP-specific optimizations.

removePieces model

Gas scales sublinearly with K (pieces removed per tx) — multiple pieces share Sum Tree ancestor nodes in the same tx (warm access: ~200 gas vs ~5,000). O(K × log₂N) theoretical formula not clearly confirmed by observed data (N not tracked at removal time in FOC Observer). Model: observed values for K≤144, linear interpolation between points, K^0.614 extrapolation beyond K=144.

Gas fee model

From Lotus source: effective_gas_price = BaseFee + min(GasPremium, GasFeeCap − BaseFee). Total cost = GasUsed × BaseFee (burned) + GasUsed × effective_premium (to SP) + overestimation burn. Today: BaseFee ≈ 100 attoFIL, SP tip dominates (99.9%). Under FIP-0115 (Draft): BaseFee rises to ~90% of EGP.