Roofline Performance Calculator
Calculate an ideal compute-or-memory throughput roof from entered operations, bytes moved and hardware ceilings.
YOUR CALCULATION
Result
—
Calculation steps
Work through one step at a time or reveal the complete calculation. The steps use your current inputs.
How to calculate
Choose one memory boundary and matching peak ceilings, then take the larger ideal compute/transfer time.
Intensity=operations/bytes; roof=min(peak FLOP/s,bandwidth bytes/s×intensity)
Worked example
1e12 FLOPs moving 1e11 bytes at 500 GB/s takes at least 0.2 seconds under this roof.
Assumptions and input details
Analytical lower time bound, not a benchmark or product specification. Excludes latency, launch cost, dependence, occupancy, instruction mix, cache hierarchy and overlap limits. FLOP counting and precision must match the entered peak.
References: NVIDIA: roofline model
Using this calculator
Can I enter my own problem?
Yes. Replace the example values with your own inputs. The result, calculation steps, tables and applicable diagrams update from those inputs. Blank or invalid values clear the result; example values are never silently substituted for missing inputs.
Are the results exact?
Results use floating-point arithmetic and displayed rounding. Read the formula, assumptions and any stated data precision before interpreting the result.
Is my input uploaded?
The calculation runs in your browser. No account, file upload or external AI service is required. Inputs are not saved between page reloads.
Estimate fixed-workload speedup from parallel fraction, worker count and entered constant overhead.
Find transfer time, required nominal speed or data size. Distinguish bits, bytes, decimal and binary units and include an explicit effective-throughput percentage.
Combine operation latency, queue-depth budget, fixed block size and transfer ceiling into a steady-state IOPS estimate.
Reserved advertising space.