RAM Sizing Calculator
Estimate how much system memory a machine needs from its operating system, virtual machines or containers, application datasets and GPUs, then round up to a configuration that populates every memory channel equally.
Results
- Workload demand
- 136 GBcalculated
- Required with headroom
- 163 GBcalculated
- Recommended configurationOne DIMM per channel
- 8 × 24 GB = 192 GBcalculated
What it calculates
- Memory demand
- Recommended capacity
- Balanced DIMM configuration
Worked example
Using the default inputs below; change any input above to recalculate.
| Operating system & services | 8 GB |
|---|---|
| VMs or containers | 0 |
| Memory per VM / container | 16 GB |
| In-memory dataset | 64 GB |
| Working-set multiplier | 2 |
| Total GPU memory in the system | 0 GB |
| Apply the 2 × GPU memory guideline | Yes |
| Headroom | 20 % |
| Memory channels to populate | 8 |
| Workload demand | 136 GB | calculated |
|---|---|---|
| Required with headroom | 163 GB | calculated |
| Recommended configuration | 8 × 24 GB = 192 GB | calculated |
Methodology
Demand is the sum of OS, VM/container and dataset working-set memory. For GPU systems, a common guideline is at least twice the total GPU memory in system RAM so data loaders and pinned buffers do not starve GPUs. The larger figure plus headroom is rounded up to identical modules across all channels.
Formulas
- Demand
- M = OS + n_VM × M_VM + dataset × factor
- Configuration
- channels × DIMMs_per_channel × module_size ≥ M × (1 + headroom)
Assumptions
- Standard module sizes from 8 GB to 256 GB.
Limitations
- Application-specific requirements (databases, CAE solvers) should override these estimates.