Storage & networking

MB/s vs Mb/s: one capital letter changes the unit by a factor of eight

A 1,000 Mb/s network link and a drive copying at 125 MB/s can describe roughly the same raw data rate. The difference is bits versus bytes: one byte contains eight bits.

The fast conversionDivide megabits per second (Mb/s or Mbps) by 8 to get megabytes per second (MB/s) at the same raw bit rate. Multiply MB/s by 8 to get Mb/s. Real transfers are usually lower because protocols, encoding, storage behavior, and system overhead consume part of the theoretical rate.

Capital B means byte; lowercase b means bit

The unit symbols matter. A byte is eight bits. NIST's binary-prefix reference gives the relationship directly: one byte (B) equals eight bits (bit). That means 8 Mb of data is 1 MB when both use the same decimal prefix.

Network products are commonly marketed in bits per second, such as 100 Mb/s, 1 Gb/s, or 10 Gb/s. Storage applications often report file-transfer rates in bytes per second, such as MB/s or GB/s. Comparing the printed numbers without converting the units can make one product appear eight times faster or slower than it really is.

Bit/byte rate converter

Why 1,000 Mb/s becomes 125 MB/s on paper

The arithmetic is simple:

1,000 Mb/s ÷ 8 = 125 MB/s

That 125 MB/s is an ideal unit conversion, not a promise that a real file copy will hold exactly 125 MB/s. Ethernet framing, TCP/IP headers, Wi-Fi airtime, encryption, file-system work, drive speed, cache behavior, and application overhead can all reduce useful payload throughput.

Decimal prefixes are another source of confusion

NIST defines the SI prefixes as powers of ten: kilo is 10³, mega is 10⁶, giga is 10⁹, and tera is 10¹². NIST also warns that SI prefixes should not be repurposed for powers of two. Binary multiples have separate prefixes such as Ki, Mi, Gi, and Ti.

So 1 MB is 1,000,000 bytes in the SI system, while 1 MiB is 1,048,576 bytes. Some software has historically displayed binary quantities while labeling them with decimal-looking symbols, which makes storage and transfer comparisons harder than they need to be.

Example: downloading a 10 GB file over a 1 Gb/s link

Using decimal units for a clean theoretical example, 10 GB is 80 gigabits because there are eight bits per byte. A 1 Gb/s link would therefore need at least 80 seconds to carry 80 Gb if every transmitted bit were useful file data and the link stayed at exactly 1 Gb/s.

Real download time will be longer. The remote server may be slower, the route may be congested, protocol overhead consumes bandwidth, and a Wi-Fi connection may fluctuate. The calculation is still useful because it provides a lower-bound sanity check.

Where people commonly mix the units

ContextCommon unitWhat to watch
Internet planMb/s or Gb/sUsually bits per second
Ethernet link speedMb/s or Gb/sRaw link rate, not guaranteed file throughput
SSD file copyMB/s or GB/sUsually bytes per second
Browser download displayMB/sOften bytes per second; software varies
Storage capacityGB or TBCapacity unit, not a per-second rate

Do not use the ÷8 rule to predict every real transfer

The factor of eight converts the units exactly. It does not model a protocol stack or a device. If a network interface reports 2.5 Gb/s, the ideal decimal conversion is 312.5 MB/s. A file transfer can still be limited by the sending storage, receiving storage, CPU, wireless conditions, protocol overhead, or the other endpoint.

The same caution applies in reverse. A fast SSD benchmark does not imply that an internet connection can feed it at that rate. The slowest stage in the end-to-end path becomes the practical bottleneck.

MB/s is not the same question as GB versus GiB

Bits-versus-bytes and decimal-versus-binary are two separate unit problems. First decide whether the quantity is measured in bits or bytes. Then check whether the prefix represents a decimal SI multiple or an explicitly binary prefix. For the storage-capacity side of that distinction, see why a 1 TB drive can appear as about 931 GiB.

A reliable comparison workflow

  1. Copy the unit exactly, including capitalization and “per second.”
  2. Convert bits to bytes or bytes to bits before comparing two rates.
  3. Check whether the source is using decimal or binary prefixes.
  4. Treat interface/link rates as ceilings, not guaranteed payload throughput.
  5. Identify the likely bottleneck across the full path, not just the fastest advertised component.

Primary sources

The converter shows unit-equivalent raw rates. It does not estimate protocol efficiency or real-world throughput.