SCIENTIFIC RIGOR & METRICS

Testing Methodology & Protocol

Formal specification of the CoinMiners browser benchmark suite (schema coinminers-device-check-1.0) and our transparent mathematical economic model.

1. Benchmark Workload Protocol

Universal Schema 1.0

To maintain strict reproducibility across Windows, macOS, and Linux without installing native software, CoinMiners employs a standardized, sandboxed in-browser workload.

CPU Workload: Multi-Thread WebCrypto SHA-256

The CPU test launches dedicated Web Workers executing SHA-256 hashing loops using the browser's hardware-accelerated SubtleCrypto.digest() API.

Worker Matrix: Evaluated at 1, 2, and 4 worker threads (capped at navigator.hardwareConcurrency).
Rounds: 3 independent 5-second sampling intervals per worker group.
Statistical Metric: Median throughput (operations/second) and Coefficient of Variation (CV = σ / μ) to quantify thermal throttling or background system interference.

GPU Workload: WGSL Compute Shader with CPU Reference Verification

If the browser exposes a valid WebGPU adapter, a custom WebGPU compute pipeline is instantiated executing a 1024-step xorshift pseudorandom transformation across 65,536 parallel work items per batch.

Execution: Repeated over 3 distinct 5-second rounds.
Buffer Readback: Entire storage buffer is mapped back to CPU memory using mapAsync(GPUMapMode.READ).
Verification: Three pseudo-random output indices (first element, second element, last element) are independently computed on the CPU and strictly matched against the GPU readback buffer. Any mismatch aborts with a validation failure.

2. Why Browser Synthetic Scores Cannot Equal Mining H/s

A core ethical principle of CoinMiners is that we never deceive users by claiming a WebCrypto score translates to cryptocurrency revenue.

Algorithmic Architecture Discrepancy

Real mining algorithms (such as Monero's RandomX) rely on dedicated virtual machines and intensive CPU L3 cache interactions (≥2MB per thread). WebCrypto SHA-256 only exercises standard crypto instruction sets, not cache latency or memory bandwidth.

Sandbox & Memory Limitations

GPU algorithms like KawPow or Etchash require allocating huge directed acyclic graphs (DAGs) ranging from 3GB to 6GB directly in VRAM. Web browsers restrict WebGPU buffer allocation sizes for security, making browser-based mining impractical.

3. Published Mathematical Economic Formulas

Our Profit Calculator implements standard financial cost accounting without hidden coefficients or invented fees.

// 1. Adjusted Daily Gross Revenue adjusted_gross = gross_24h × (operating_hours / 24) // 2. Fees (Platform fee is charged after the pool fee) pool_fee = adjusted_gross × pool_fee_pct / 100 platform_fee = (adjusted_gross − pool_fee) × platform_fee_pct / 100 total_fees = pool_fee + platform_fee // 3. Power Consumption & Utility Cost daily_kwh = (incremental_wall_watts / 1000) × operating_hours electricity_cost = daily_kwh × electricity_price_usd_per_kwh // 4. Illustrative Net Margin illustrative_net = adjusted_gross − total_fees − electricity_cost − conversion_costs − payout_costs − other_costs // 5. Break-Even Electricity Cost Threshold ($/kWh) break_even_electricity = (adjusted_gross − total_fees − conversion_costs − payout_costs − other_costs) / daily_kwh * If daily_kwh == 0: Status = ZERO_POWER (N/A) * If break_even < 0: Status = ALWAYS_UNPROFITABLE (Unprofitable even with free electricity)

Put the methodology to work

Run a benchmark on your hardware or simulate your utility cost scenario.

Start Benchmark → Open Calculator