Computing & Software
Roofline Arithmetic Intensity Performance Calculator
Compare a compute-throughput ceiling with the memory-bandwidth ceiling for an algorithm.
How the calculation works
Memory-limited performance ceiling = intensity*bandwidth GFLOP/s. Roofline ceiling at this intensity = min(peak,intensity*bandwidth) GFLOP/s. Ridge-point arithmetic intensity = peak/bandwidth FLOP/byte.
Worked example
Example inputs — Arithmetic intensity (FLOP/byte): 2; Sustained memory bandwidth (GB/s, decimal): 100; Compute performance ceiling (GFLOP/s): 1000. Expected results — Memory-limited performance ceiling: 200 GFLOP/s; Roofline ceiling at this intensity: 200 GFLOP/s; Ridge-point arithmetic intensity: 10 FLOP/byte.
Assumptions and limits
Analytical software/hardware scenario, not a benchmark or a live provider quote. Use the actual architecture, workload and billing definitions; overheads not named in the model are excluded. FLOP counting and transferred bytes must match the selected memory hierarchy. A roofline is an upper bound, not a prediction that all code will reach it.
Reference and further reading
Read the underlying subject reference. Calculith is an independent implementation; the reference does not endorse or certify this tool.
Related calculators
- Password Entropy & Brute-Force Time Estimate
- CPU Instruction Count & CPI Runtime Calculator
- Amdahl & Gustafson Parallel Speedup Calculator
- Memory Bus Theoretical Bandwidth Calculator
- DRAM Timing Cycles to Access Latency Calculator
Example values are editable. Results are estimates where assumptions apply. Calculations run in your browser. Updated 2026-11-16.