Specifications
Part Number
G650-06076-01
Accelerators
Two Google Edge TPU ML accelerators
Performance
4 TOPS per Edge TPU, 8 TOPS combined (2 TOPS/W)
Host Interface
M.2 E-key, one PCIe Gen2 x1 lane per Edge TPU
Form Factor
M.2 2242 / 2230 style E-key module
Supported Frameworks
TensorFlow Lite with the Edge TPU compiler
Operating Systems
Linux, Windows and Windows Server
Power
~2 W per Edge TPU typical
Model Coverage
Quantised TensorFlow Lite models compiled for Edge TPU
Vendor
Google Coral
Overview
The Coral M.2 Accelerator with Dual Edge TPU puts two Google Edge TPUs on a single M.2 E-key module, giving 8 TOPS of INT8 inference to a host that has only an M.2 slot to spare. Each Edge TPU runs over its own PCIe Gen2 x1 lane, so both accelerators can work in parallel on independent model instances.
Key Benefits
8 TOPS in an M.2 slot: two Edge TPUs on one E-key module. Very low power: roughly 2 W per accelerator makes fanless and sealed enclosures practical. Parallel inference: independent PCIe lanes let each TPU serve a different model. Mature toolchain: TensorFlow Lite with the Coral Edge TPU compiler and extensive pre-compiled model coverage.
Applications
Multi-camera video analytics nodes, industrial inspection stations, retail and access-control appliances, edge servers needing parallel model instances, robotics perception, and retrofit AI acceleration for existing fanless industrial PCs.
Request a Quote — GOOGLE CORAL M.2 ACCELERATOR WITH DUAL EDGE TPU — G650-06076-01, 8 TOPS
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →