Embedded MXM GPU Modules based on NVIDIA Turing™ Architecture

Product catalog summary

Overview
The document provides detailed specifications and features of ADLINK Technology's Mobile PCI Express Modules equipped with NVIDIA Embedded GPUs based on the Turing™ architecture. These modules are designed for high-performance computing and graphics applications.

Key Features

Utilizes NVIDIA Turing™ architecture for embedded graphics.
Available in standard MXM 3.1 Type A/B/B+ form factors.
Offers up to 3072 CUDA cores and up to 9.4 TFLOPS peak FP32 performance.
Supports up to 16GB GDDR6 memory with a 256-bit interface and up to 448GB/s memory bandwidth.
Capable of supporting up to 4 DisplayPort 1.4a displays with a total graphics power (TGP) of 110W.
Guaranteed 5-year availability.

Product Variants

EGX-MXM-T1000: Features NVIDIA Quadro® T1000, MXM 3.1 type A, with 4GB GDDR6 memory and 896 CUDA cores.
EGX-MXM-RTX3000: Features NVIDIA Quadro® RTX3000, MXM 3.1 type B, with 6GB GDDR6 memory and 1920 CUDA cores.
EGX-MXM-RTX5000: Features NVIDIA Quadro® RTX5000, MXM 3.1 type B+, with 16GB GDDR6 memory and 3072 CUDA cores.

Specifications

Graphics Core: Quadro® T1000, RTX3000, RTX5000.
Memory: Ranges from 4GB to 16GB GDDR6 with varying bandwidths.
GPGPU Computing: CUDA cores range from 896 to 3072, with additional Tensor and RT cores in RTX models.
Compute API: Supports CUDA Toolkit 8.0+, CUDA Compute version 6.1+, and OpenCL™ 1.2.
Graphics API: Compatible with DirectX® 12, OpenGL 4.6, Vulkan 1.0.
Display Outputs: 4x DisplayPort 1.4a, supporting 4K at 120Hz or 8K at 60Hz.
Interface: MXM 3.1, PCI Express Gen3 x16.
Dimensions: Vary by model, with a consistent height of 4.8 mm.
Operating Temperature: Standard and extended ranges available.
Power Consumption: Ranges from 50W to 110W TGP.
OS Support: Compatible with Windows 11, 10, and Linux drivers, 64-bit.

Conclusion
ADLINK's MXM modules provide robust solutions for embedded graphics applications, offering a range of performance levels and features to meet diverse computing needs.

Catalog excerpts

Embedded MXM GPU Modules based on NVIDIA Turing™ Architecture-1

Embedded MXM Modules based on NVIDIA Turing™ Architecture Mobile PCI Express Modules with NVIDIA Embedded GPUs • NVIDIA embedded graphics based on Turing™ architecture • Standard MXM 3.1 Type A/B/B+ form factor • Up to 3072 CUDA cores • Up to 9.4 TFLOPS peak FP32 performance • Up to 16GB GDDR6 memory, 256-bit • Up to 448GB/s maximal memory bandwidth • Support up to 4 DP 1.4a displays, 110W TGP • 5-year availability Ordering Information EGX-MXM-T1000 NVIDIA® Quadro® T1000 Embedded Graphics, MXM 3.1 type A, 82 x 70mm, PCIe x16 Gen3 EGX-MXM-RTX3000 NVIDIA® Quadro® RTX3000 Embedded Graphics, MXM 3.1 type B, 82 x 105mm, PCIe x16 Gen3 EGX-MXM-RTX5000 NVIDIA® Quadro® RTX5000 Embedded Graphics, MXM 3.1 type B+, 82 x 110mm, PCIe

Open the catalog to page 1

Embedded MXM GPU Modules based on NVIDIA Turing™ Architecture-2

Model Name Graphic Core GPU Memory GPGPU Computing CUDA Cores Tensor Cores RT Cores Compute API Graphics API 896 CUDA cores, 2.6 TFLOPS peak FP32 performance - 1920 CUDA cores, 3072 CUDA cores, 5.3 TFLOPS 9.4 TFLOPS peak FP32 performance peak FP32 performance 240 Tensor Cores 384 Tensor Cores 30 RT Cores 48 RT Cores CUDA Toolkit 8.0 and above, CUDA Compute version 6.1 and above, OpenCL™ 1.2 DirectX® 12, OpenGL 4.6, Vulkan 1.0 API Display 4x DisplayPort 1.4a digital video outputs 4K at 120Hz or 8K at 60Hz MXM 3.1, PCI Express Gen3 x16 support Display Outputs Interface Mechanicals Dimensions Form...

Open the catalog to page 2

All ADLINK TECHNOLOGY catalogs and technical brochures

*Prices are pre-tax. They exclude delivery charges and customs duties and do not include additional charges for installation or activation options. Prices are indicative only and may vary by country, with changes to the cost of raw materials and exchange rates.

Brand list