CPU Benchmark

Floating point · Integer · SIMD · Memory · Cache · Multi-core · Crypto

⏳ Initializing…
Overall Score
—
Ready to Benchmark
Starting…
FP32 Compute
Peak
—
GFLOPS
Sustained
—
GFLOPS
FP64 Compute
Peak
—
GFLOPS
Sustained
—
GFLOPS
SIMD (Vectorized)
Peak
—
GFLOPS
Sustained
—
GFLOPS
Integer Compute
Peak
—
GOPS
Sustained
—
GOPS
WASM Memory Bandwidth
Peak
—
GB/s
Sustained
—
GB/s
Random Walk Latency
32KB
—
ns
256KB
—
ns
8MB
—
ns
64MB
—
ns
lower is better
Multi-core
Aggregate
—
GFLOPS
Efficiency
—
%
Branch Prediction
Predictable
—
ns
Random
—
ns
lower is better
Cryptography
AES-GCM
—
GB/s
SHA-256
—
MH/s
Compression
Encode
—
MB/s
Decode
—
MB/s

History (local only)

    How it works

    BenchD runs intensive WebAssembly (WASM) kernels directly in your browser. The final score is a weighted average of individual tests, scaled using log1p(x) to reward performance gains while preventing numerical explosion.

    Scoring details
    FP32/64: 30% (Floating Point math)
    SIMD/Int: 20% (Vectorization & Logic)
    Memory: 20% (WASM memory throughput)
    Latency: 10% (32KB to 64MB random walks)
    Multi-core: 10% (Scaling efficiency)
    Advanced: 10% (Branch, Crypto, LZ77)

    *If a test fails on your hardware, its weight is automatically redistributed to ensure a fair overall score.