What Is an MXM GPU?

An MXM GPU is a discrete graphics processor and its memory on a small board built to the Mobile PCI Express Module standard. NVIDIA introduced MXM in 2004 so laptop makers could fit interchangeable graphics modules, and the specification is now maintained by the MXM-SIG. The current versions, MXM 3.0 and 3.1, define the module sizes, the edge connector and the signals, so a module from one supplier can fit a carrier designed to the same type.

A standard graphics card needs a full-height slot, its own fan and room for airflow. An MXM module packs the GPU and its memory onto a small board that lies flat, so the heat can be conducted into the computer’s chassis. That suits sealed and vibration-prone installs, where a large card with a fan would be a weak point.

Laptops have largely moved to GPUs soldered to the motherboard, but MXM graphics remain the standard way to add a discrete GPU to embedded, rugged and military computers.

The module plugs into a 314-pin edge connector on the carrier board, then is screwed down at its corners. Through that connector it receives PCI Express lanes from the processor, power, and control signals, and it sends its display outputs back to the carrier, which routes them to DisplayPort, HDMI or other connectors on the case. Headless modules skip the display outputs and are used purely for compute.

The connector has kept pace with PCI Express. MXM 3.1 added PCIe 3.0, and current modules run faster links where the carrier supports them; the RTX PRO 2000 Blackwell embedded module in our range uses PCIe Gen5. The specification also fixes the mounting holes, so a heat spreader or cold plate can press evenly on the GPU and memory chips and carry their heat into the case.

MXM 3.x defines two module sizes, Type A and Type B. Some suppliers also build Type B+, the Type B outline with stronger power delivery for higher-power GPUs. A carrier built for Type B can often take a Type A module, but not the reverse.

Module sizes are from the MXM 3.x specification. Power figures are typical of current modules; each module’s datasheet sets its exact rating, and the system’s cooling has to match it. Our MXM GPU modules run from 4 GB headless modules at 35 W to 16 GB Type B+ modules above 100 W.

Our guide to what an MXM slot is covers which module fits which slot in more detail.

Both put the same kind of GPU into a computer, and an embedded module often uses the same chip as a desktop graphics card, usually at lower clocks and power. The difference is the package. A PCIe card brings its own fan, heatsink and display connectors and needs a slot plus space around it. An MXM module brings none of these, so the system designer decides how it is cooled and where its outputs go.

That makes MXM the better form factor where the enclosure is sealed, small or exposed to vibration, and a PCIe card the simpler choice in a rack or a larger box PC with airflow. Embedded MXM modules also tend to have longer availability than desktop cards, which matters for systems built and supported over several years.

MXM is also a different thing from SXM, the module NVIDIA uses for its largest data centre GPUs. An SXM module mounts on a server baseboard, draws several hundred watts and links the GPUs over NVLink; our guide to SXM vs PCIe GPUs explains it.

MXM is how many rugged computers get GPU performance on vehicles and trains. Our digital train case study and Sintrones ABOX-5221-IP66 article both use MXM options, and our MXM GPU computers show the range.

We fit MXM modules in rugged and embedded computers as standard, matched to the carrier, cooling and power supply, and validate the GPU under load before the system ships.