Necessary cookies
Free network tool
Gbps ↔ Mpps calculator
Convert a bit rate into packets per second—or work backwards from Mpps. See how frame size changes the theoretical load on an Ethernet port.
Compare physical port limits
For the selected frame size, these are theoretical maximum rates in one direction. The load column compares your calculation with each port.
Scroll horizontally to see all columns.
| Port | Theoretical maximum (Mpps) | Port load | Result |
|---|---|---|---|
| 10G | 14.881 | ≈1,000% | Exceeds physical limit |
| 25G | 37.202 | ≈400% | Exceeds physical limit |
| 40G | 59.524 | ≈250% | Exceeds physical limit |
| 100G | 148.81 | ≈100% | Within physical limit |
| 400G | 595.238 | ≈25% | Within physical limit |
Displayed rates are rounded. Calculations retain full precision.
A result below the port limit does not prove that a NIC, CPU, PCIe link or filter can sustain it. Adding ports does not by itself establish the capacity of a single flow.
Exactly what is counted
The frame value includes its FCS. The calculation then adds 8 bytes for preamble/SFD and 12 byte-times for the interframe gap: 20 additional byte-times per frame.
8Preamble + SFD
64Ethernet frame, including FCS
12Interframe gap
The formula
PPS = Gbps × 10⁹ ÷ ((N + 20) × 8)
Mpps = PPS ÷ 10⁶
Gbps = Mpps × (N + 20) × 8 ÷ 1000N = Ethernet frame size (bytes)
Gbps = 10⁹ bits/s; Mpps = 10⁶ packets/s. This uses the nominal Ethernet MAC rate; physical encoding and FEC are not added again.
Compare measurements using the same convention
- A 1,500-byte IP packet is not a 1,500-byte Ethernet frame.
- With an untagged Ethernet II header and FCS, it occupies a 1,518-byte frame. Add any VLAN tags or encapsulation to your actual frame size before calculating.
- Captures and interface counters can count different bytes.
- A packet capture often omits FCS, while counters can exclude preamble and gap. Check the counter definition before comparing a measured Gbps value with a wire-rate calculation.
- An average frame size can hide short bursts.
- Use matching measurement intervals for bytes and packets. Size for relevant small-packet and burst conditions as well as the average; no software capacity follows from this conversion alone.
Questions
Why add 20 bytes to each Ethernet frame?
The calculation reserves 8 bytes for the preamble/SFD and 12 byte-times for the interframe gap. The gap is time between frames, not extra packet payload. FCS is already included in the frame size you enter.
Can 100 Gbps mean either 148.8 Mpps or 8.1 Mpps?
Yes. The packet rate depends on frame size. Those values approximately correspond to 64-byte and 1,518-byte Ethernet frames at a theoretical 100 Gbps MAC rate.
Does this qualify a server for 100G or 400G mitigation?
No. It calculates the physical link budget. Forwarding and filtering need measurements with the actual hardware, software, policies, packet mix and acceptable loss criteria.
Does this tool send traffic to a network?
No. It performs arithmetic only. It does not probe an address, generate packets or change any network configuration.
Technical references
Turn a rate into a deployment plan
Port arithmetic is one input. CPU, queues, NUMA placement, packet mix, stateful rules and failure scenarios still need qualification on the intended platform.