GB/s to Mbps converter (and Mbps to GB/s)

Which is slower: the drive or the link?

Drive benchmarks and file copy windows report gigabytes per second; network gear is rated in megabits or gigabits. Putting both on one scale shows which one sets the pace of a copy.

Link figures are nominal signalling rates; drive figures are Samsung's rated sequential read speeds; check your own drive's data sheet. Read down the Mbps column and the pattern is clear: a gigabit network holds back almost any SSD, and even 10 GbE carries well under half of what a fast NVMe drive can read.

Copy-time examples

• 1 TB over 10 GbE. At 1.25 GB/s, 1,000 GB takes 800 seconds (13 min 20 s); at a realistic 90%, about 14 min 49 s, provided both drives keep up.
• The same 1 TB over Gigabit Ethernet. 8,000 seconds, or 2 h 13 min 20 s at full rate.
• A NAS behind a 2.5 GbE port. An array that reads at 0.8 GB/s (6,400 Mbps) is capped at 0.3125 GB/s by the port, well under half its speed.
• A 25,000 Mbps datacenter link. 3.125 GB/s at most; a single SATA array usually cannot fill it.

The file transfer time calculator has presets for these links and an efficiency setting. For steady transfers measured per month, such as offsite backups, TB per month to Mbps gives the sustained rate: 10 TB a month is about 30.4 Mbps around the clock.

Pitfall: GB/s, Gbps and GiB/s

Drive and copy tools write GB/s (gigabytes); network gear is labelled Gbps (gigabits). The factor between them is 8, so 1.25 GB/s = 10 Gbps, and mixing them makes a 10 Gbps link sound eight times faster than it is. Operating systems may also show GiB/s labelled as GB/s: 1 GiB/s is about 8,590 Mbps, so a full 1.25 GB/s appears as about 1.16 GiB/s.

A good 10 GbE copy that reports 1.1 GB/s is running at 8,800 Mbps, 88% of the nominal rate, which is normal once framing and protocol overhead are counted. To stay in bits, use Gbps to Mbps; for plan-sized speeds, Mbps to MB/s is easier to read, and the bytes to Mbps converter covers every unit from B/s up.