Computer performance
PC bottleneck and frame budget
Compare measured CPU and GPU processing times for one game or workload against your target frame rate.
Your inputs
Complete the fields for your selected method. Zero means intentionally zero, not unknown. Currency selection labels amounts; it does not convert exchange rates.
How this calculation works
frame budget = 1000 / target FPS; ideal overlapped CPU/GPU throughput ceiling = 1000 / max(CPU stage ms, GPU stage ms)
Use CPU execution time and GPU busy/render time from a profiler that distinguishes those stages, captured in the same repeatable workload, resolution and settings. Do not substitute total frame interval, CPU utilization percentage, GPU utilization percentage or input latency. If you do not have comparable stage measurements, no reliable bottleneck percentage can be derived from component names alone. Frame caps, synchronization, memory, storage, scheduling, thermal/power limits and frame generation are not modeled. The 5% similar-times label is a disclosed display tolerance, not a hardware benchmark standard.
What information do I need?
Use records for the same job, equipment or accounting period. Only the fields for your selected entry method are required. An unknown amount is not zero.
Input definitions and where to find them
- Game/workload, resolution and settings: Use matching records or an explicitly identified assumption for game/workload, resolution and settings. Unknown is not the same as zero.
- Measured CPU processing time per frame (ms/frame): Use matching records or an explicitly identified assumption for measured cpu processing time per frame. Unknown is not the same as zero.
- Measured GPU busy/render time per frame (ms/frame): Use matching records or an explicitly identified assumption for measured gpu busy/render time per frame. Unknown is not the same as zero.
- Target frame rate (FPS): Use matching records or an explicitly identified assumption for target frame rate. Unknown is not the same as zero.
Use the Where to find this value controls beside the inputs for available work-it-out helpers. Estimates remain your assumptions; a helper does not verify the source figures.
Worked example
Example amounts use US dollars (USD).
Example answer: Longer measured stage in this workload is GPU stage takes longer for the illustrative inputs below.
Example inputs
- Game/workload, resolution and settings: Example workload at 1080p, same settings
- Measured CPU processing time per frame: 8.00 ms/frame
- Measured GPU busy/render time per frame: 12.00 ms/frame
- Target frame rate: 60.00 FPS
- Longer measured stage in this workload
- GPU stage takes longer
- Simplified processing-throughput ceiling
- 83.33 FPS
- Target time budget per frame
- 16.67 ms
- CPU time minus target budget
- -8.67 ms
- GPU time minus target budget
- -4.67 ms
- The ideal throughput model assumes CPU and GPU stages overlap across frames. It is not input-to-display latency, measured delivered FPS or a guaranteed upgrade outcome. A negative budget difference means the supplied stage time is below the target budget.
These illustrative values are not recommended rates or industry averages. Replace every example input with your own figures.
Use the result well
Compare measured CPU and GPU processing times for one game or workload against your target frame rate.
Keep inputs on the same scope and time basis. Calculations retain precision internally; displayed values generally round to two decimal places, with extra precision for very small values and whole-unit rounding where required. Economic comparisons describe your supplied assumptions and do not guarantee future results.
This tool provides business arithmetic. It does not approve engineering designs, establish legal or tax treatment, select equipment or certify safe operating conditions.