Skip to Content
Björn Tech
Aetina Product Group

GPU and AI Accelerator Modules (CoreEdge MXM)

Product Series 05

GPU and AI Accelerator Modules (CoreEdge MXM)

Embedded GPU modules in the MXM 3.1 form factor, an MXM module with four Hailo-8 processors and an MXM expansion kit.

SectorsMedicalDefence IndustryAutonomous Systems & Robotics

MXM Modules

MX500B-QA · MX2000B-VA · MX4000B-WA · MX5000B-XA

CoreEdge MXM · NVIDIA RTX PRO Blackwell

MX500B-QA · MX2000B-VA · MX4000B-WA · MX5000B-XA

  • NVIDIA® RTX PRO™ 500 Blackwell Embedded GPU (MX500B-QA): 9.19 TFLOPS peak FP32 performance
  • NVIDIA® RTX PRO™ 2000 Blackwell Embedded GPU (MX2000B-VA): 13.78 TFLOPS peak FP32 performance
  • NVIDIA RTX PRO 4000 Blackwell Embedded GPU (MX4000B-WA)
  • NVIDIA RTX PRO 5000 Blackwell Embedded GPU (MX5000B-XA): 40.62 TFLOPS peak FP32 performance
MX2000B-VA · View ProductMX4000B-WA · View ProductMX5000B-XA · View ProductMX500B-QA · View Product
MX2000A-VA · MX3500A-SA · MX5000A-WA

CoreEdge MXM · NVIDIA RTX Ada Generation

MX2000A-VA · MX3500A-SA · MX5000A-WA

  • NVIDIA RTX 2000 Ada Generation Embedded GPU (MX2000A-VA, MXM 3.1 Type A): 3072 CUDA cores, 24 RT cores and 96 Tensor cores, 8GB GDDR6 memory, 12.99 TFLOPS peak FP32 performance
  • NVIDIA RTX 3500 Ada Generation Embedded GPU (MX3500A-SA): 23.04 TFLOPS peak FP32 performance
  • NVIDIA RTX 5000 Ada Generation Embedded GPU (MX5000A-WA): 9728 CUDA cores, 76 RT cores and 304 Tensor cores, 16GB GDDR6 memory, 41.15 TFLOPS peak FP32 performance
  • PCIe Gen 4 x16 interface
  • Error Correction Code (ECC) support (MX3500A-SA, MX5000A-WA)
MX2000A-VA · View ProductMX3500A-SA · View ProductMX5000A-WA · View Product
M3A500-PP · M3A1000-PP · M3A2000-VY · M3A4500-WP

CoreEdge MXM · NVIDIA RTX Ampere

M3A500-PP · M3A1000-PP · M3A2000-VY · M3A4500-WP

  • NVIDIA RTX A500 Embedded GPU based on Ampere architecture (M3A500-PP): 2048 CUDA cores, 16 RT cores and 64 Tensor cores, 4GB GDDR6 memory, PCIe Gen 4 x4 interface, 7.28 TFLOPS peak FP32 performance
  • NVIDIA RTX A1000 Embedded GPU based on Ampere architecture (M3A1000-PP): 2048 CUDA cores, 16 RT cores and 64 Tensor cores, 4GB GDDR6 memory, PCIe Gen 4 x8 interface
  • NVIDIA RTX A2000 Embedded GPU based on Ampere architecture (M3A2000-VY): 2560 CUDA cores, 20 RT cores and 80 Tensor cores, 8GB GDDR6 memory, 8.25 TFLOPS peak FP32 performance
  • NVIDIA RTX A4500 Embedded GPU based on Ampere architecture (M3A4500-WP)
  • Error Correction Code (ECC) support (M3A2000-VY, M3A4500-WP)
M3A1000-PP · View ProductM3A2000-VY · View ProductM3A4500-WP · View ProductM3A500-PP · View Product
M3T1000-PN · M3T3000-QN · M3T5000-WN

CoreEdge MXM · NVIDIA Quadro Turing

M3T1000-PN · M3T3000-QN · M3T5000-WN

  • NVIDIA Quadro T1000 embedded graphics based on NVIDIA Turing architecture (M3T1000-PN): 896 CUDA cores, 4GB GDDR6 memory, 2.6 TFLOPS peak FP32 performance
  • NVIDIA Quadro RTX3000 embedded graphics based on NVIDIA Turing architecture (M3T3000-QN): 1920 CUDA cores, 30 RT cores and 240 Tensor cores, 6GB GDDR6 memory, 5.3 TFLOPS
  • NVIDIA Quadro RTX5000 embedded graphics based on NVIDIA Turing architecture (M3T5000-WN): 3072 CUDA cores, 48 RT cores and 384 Tensor cores, 16GB GDDR6 memory, 9.4 TFLOPS
  • Support up to 4 DisplayPort 1.4 displays
  • Support CUDA Compute version 7.5, OpenCL 1.2, OpenGL 4.6, DirectX 12 and Vulkan 1.1 API
  • 5 year life cycle availability
M3T1000-PN · View ProductM3T3000-QN · View ProductM3T5000-WN · View Product
MXA380E-QA

CoreEdge MXM · Intel Arc

MXA380E-QA

  • Intel Arc A380E GPU for the edge
  • 128 Intel® XMX Engines Cores, 8 Xe-cores, 8 Ray Tracing and 128 Xe Vector Engines, 6GB GDDR6 memory
  • PCIe Gen4.0 x8 signal
  • Five-year long-term product availability support
View Product
AI-MXM-H84W

CoreEdge MXM · Hailo-8

AI-MXM-H84W

  • 4 x HAILO-8 Industrial version
  • 104 TOPS INT8
  • Support -40°C~+85°C
  • TDP 25W (Typical)
  • MXM TYPE B
View Product
Compare Models (14)
ModelGPUCUDA coresMemoryFP32 (TFLOPS)TGPForm factor
MX500B-QARTX PRO 5001.7926 GB GDDR79,1960 WMXM Type A
MX2000B-VARTX PRO 40007.68016 GB GDDR733,75115 WMXM Type B
MX4000B-WARTX PRO 500010.49624 GB GDDR740,62115 WMXM Type B
MX5000B-XA——————
MX2000A-VANVIDIA RTX™ 3500 Ada Generation Embedded GPU512012GB GDDR6—115 WMXM graphics module version 3.1, Type B
MX3500A-SANVIDIA RTX™ 5000 Ada Generation Embedded GPU972816GB GDDR6—115 WMXM graphics module version 3.1, Type B
MX5000A-WA——————
M3A500-PPNVIDIA RTX™ A50020484GB GDDR6—35W/ 45WMXM graphics module version 3.1, Type A
M3A1000-PPNVIDIA RTX™ A200025608GB GDDR6—60WMXM graphics module version 3.1, Type A
M3A2000-VYNVIDIA RTX™ A4500588816GB GDDR6—125WMXM graphics module version 3.1, Type B
M3A4500-WP——————
M3T1000-PNNVIDIA Quadro RTX300019206GB GDDR6—80 WMXM graphics module version 3.1, Type B
M3T3000-QNNVIDIA Quadro RTX5000307216GB GDDR6—110 WMXM graphics module version 3.1, Type B+
M3T5000-WN——————

Values from the specification tables on the product pages; see each product page for details.

Compare Models (14)
ModelGPUCUDA coresMemoryFP32 (TFLOPS)TGPForm factor
MX500B-QARTX PRO 5001.7926 GB GDDR79,1960 WMXM Type A
MX2000B-VARTX PRO 40007.68016 GB GDDR733,75115 WMXM Type B
MX4000B-WARTX PRO 500010.49624 GB GDDR740,62115 WMXM Type B
MX5000B-XA——————
MX2000A-VANVIDIA RTX™ 3500 Ada Generation Embedded GPU512012GB GDDR6—115 WMXM graphics module version 3.1, Type B
MX3500A-SANVIDIA RTX™ 5000 Ada Generation Embedded GPU972816GB GDDR6—115 WMXM graphics module version 3.1, Type B
MX5000A-WA——————
M3A500-PPNVIDIA RTX™ A50020484GB GDDR6—35W/ 45WMXM graphics module version 3.1, Type A
M3A1000-PPNVIDIA RTX™ A200025608GB GDDR6—60WMXM graphics module version 3.1, Type A
M3A2000-VYNVIDIA RTX™ A4500588816GB GDDR6—125WMXM graphics module version 3.1, Type B
M3A4500-WP——————
M3T1000-PNNVIDIA Quadro RTX300019206GB GDDR6—80 WMXM graphics module version 3.1, Type B
M3T3000-QNNVIDIA Quadro RTX5000307216GB GDDR6—110 WMXM graphics module version 3.1, Type B+
M3T5000-WN——————

Values from the specification tables on the product pages; see each product page for details.

Expansion Kit

DEV-MXM-4H-A1

MXM expansion kit

DEV-MXM-4H-A1

  • Built based on PCI Express x16 Gen4, which converts the MXM slot to a PCI Express slot
  • Lets developers use the original PCIe system to test the compute performance of the MXM module in the early stage of a vision computing project
View Product

Why Björn Tech?

Integration in a Single System

We plan Aetina AI systems together with POLYRACK enclosures, our thermal management partners and Innodisk storage, memory and cameras as one project.

The Right Platform for the Application

A palm-sized fanless Mini system, a GPU-expansion workstation or a 2U server: we choose the platform with you based on what the application needs.

A Local Point of Contact

As the authorised distributor in Türkiye we are your single point of contact for quotes, product selection and project support.

Talk to our engineering team about your project.

Contact Us