MemryX

schematics

none

toolchain

partial (open runtime (MxAccl) + utils + driver mirror, closed NeuralCompiler)

datasheets

public

blobs

required (proprietary silicon)

retail

open_market (Amazon/Mouser ~$149)

form_factor

module (M.2 2280 module carrying four MX3 chips)

Proprietary dataflow silicon with a public datasheet and several open-source runtime and driver repositories, though the compiler core and chip firmware are closed.

• https://developer.memryx.com/specs/M.2_datasheet.html recorded 2026-08-13

  openly published MX3 M.2 module datasheet - "MemryX MX3 (x4)" AI processor, PCIe Gen 3 2-lane interface, M.2-2280-D5-M form factor, operating range and CE/FCC Class A and RoHS compliance, plus pinout, placement, power-rail, clock and strap-pin tables; no board design files are offered

• https://github.com/memryx/MxAccl recorded 2026-08-13

  "open-source code for both the MemryX C++ runtime library and the acclBench benchmarking tool" under MPL-2.0, with the SDK it depends on supplying "precompiled libraries, drivers, and tools"

• https://www.tomshardware.com/tech-industry/artificial-intelligence/memryx-launches-usd149-mx3-m-2-ai-accelerator-module-capable-of-24-tops-compute-power recorded 2026-08-13

  "The MX3 M.2 module is available for $149 USD, making it an affordable option for developers and organizations seeking to add AI processing capabilities to edge devices"

The module retails at $149, and independent reviews, a 2025 industry award and an active GitHub organization point to growing but emerging traction.

• https://github.com/memryx/MxAccl recorded 2026-08-13

  live public repository for the MemryX C++ runtime and acclBench benchmarking tool

• https://www.tomshardware.com/tech-industry/artificial-intelligence/memryx-launches-usd149-mx3-m-2-ai-accelerator-module-capable-of-24-tops-compute-power recorded 2026-10-08

  "The MX3 M.2 module is available for $149 USD"; a technology start-up originating from the University of Michigan; no distributor is named

Four MX3 chips on the M.2 module combine for up to 24 TOPS, with support for 4, 8 and 16-bit weights plus BFloat16. On-chip parameter capacity limits how large a model can run, and the figures come from launch coverage rather than a vendor specification table.

• https://developer.memryx.com/specs/M.2_datasheet.html recorded 2026-09-29

  MemryX MX3 M.2 datasheet listing "MemryX MX3 (x4)" and stating "The 4 MX3 chips appear as one logical unit to the NeuralCompiler and runtime software stack"; no TOPS or parameter-capacity row.

• https://www.tomshardware.com/tech-industry/artificial-intelligence/memryx-launches-usd149-mx3-m-2-ai-accelerator-module-capable-of-24-tops-compute-power recorded 2026-09-29

  Launch coverage of the MX3 M.2 module stating "Each MX3 chip is said to deliver 6 TOPS (Tera Operations Per Second), offering a total of up to 24 TOPS of compute power", support for "4-bit, 8-bit, 16-bit weights, and BFloat16", and "a total of up to 42 million parameters" across the four chips.