Advantech expand Industrial GPU Solution Lineup with MXM Cards and edge computer
Advantech, a leading global provider of industrial edge computers and GPU solutions, has introduced in June 2024 its latest MXM GPU cards—the SKY-MXM-2000A, SKY-MXM-3500A, and SKY-MXM-5000A—powered by NVIDIA's Ada Lovelace architecture. Alongside these advanced GPU cards, Advantech has also launched the UNO-148 V2, an ultra-compact GPU edge computer engineered to enhance AI processing capabilities at the edge.
As edge computing evolves, there is an increasing demand for high-performance, efficient, and reliable AI processing solutions. Advantech meets this need with its new NVIDIA Ada Lovelace-based MXM GPU cards, which offer powerful processing capabilities and energy efficiency for AI-driven professional workflows. The UNO-148 V2 GPU edge computer, designed for easy deployment and customisation, adds to the edge AI landscape with its compact form factor and flexible I/O design. It also includes an optional second-stack extension kit, enabling it to support MXM GPUs and handle diverse edge AI applications with modular flexibility and robust connectivity.
New MXM GPU Series Leveraging the NVIDIA Ada Lovelace Architecture
The new MXM GPU series from Advantech, powered by NVIDIA’s Ada Lovelace architecture, includes the SKY-MXM-5000A, SKY-MXM-3500A, and SKY-MXM-2000A. With compact designs, robust durability, and low power requirements, these GPUs are optimized for portable computing in tight, fanless environments.
Key features of the NVIDIA Ada Lovelace-based MXM GPUs are:
- CUDA Cores: Double the graphics and compute performance over the previous NVIDIA Ampere generation.
- 3rd-Generation RT Cores: Delivers up to twice the rendering throughput of the Ampere architecture.
- 4th-Generation Tensor Cores: Provides twice the AI computing power compared to the Ampere generation.
- Enhanced Image Processing: Achieves 2.5 times faster image processing with upgraded video encoding and decoding engines.
- Extended Product Lifecycle: Over 5 years of support, ensuring long-term stability and reliability.
- Customize Thermal Solutions: Supports stable performance in temperatures ranging from 0°C to 55°C, ideal for harsh industrial settings.
These advancements accelerate the development of AI applications across industrial manufacturing, healthcare, smart cities, and gaming, enabling machine automation, machine vision, medical imaging, and driver assistance systems.
Why choose MXM GPU cards for Edge AI Applications?
Video link to YouTube here.
UNO-148 V2 with MXM GPU Integration for Enhanced AI Capabilities
Meanwhile, the latest addition to Advantech’s Edge Controller series, the UNO-148 V2, is introduced alongside the UNO-MXM-CB01 carrier stack to address diverse AI-driven industrial applications. Equipped with the newest Intel® Core™ processors (13th Gen), the UNO-148 V2 delivers enhanced computing power. Its modular design supports multiple expansion options, particularly through integration with the UNO-MXM-CB01 carrier stack, allowing compatibility with Advantech’s NVIDIA-based MXM GPU cards. This setup enables the UNO-148 V2 to support a 1 x MXM 3.1 Type A form factor GPU module with a power capacity of up to 35W.
The MXM expansion module further enhances functionality by providing 4 independent display outputs, ideal for real-time data processing and visual displays in high-demand environments such as airports, MRT stations, and control rooms. With features like port isolation and RAID-supported NVMe SSD storage, the UNO-148 V2 offers durable performance for edge computing and GPU-intensive tasks, making it highly suitable for sectors like healthcare and manufacturing, especially in image capture. Designed for longevity and consistent performance, the UNO-148 V2 , combined with UNO-MXM modules, excels in AI-intensive environments, making it an ideal choice for edge computing and GPU-enhanced applications.
Selection Guide
NVIDIA MXM GPUs
Advantech's MXM series features a compact and durable design with a small form factor and low power requirements, making them ideal for confined spaces in applications such as surgical systems, gaming, and autonomous driving.
See the full selection guide with all Industrial GPU Solutions (v2024) here. If you've found a product you're interested in and it's not on our website, contact us via this contact form.
Category | NVIDIA GPU Cards | ||||||
| Product image |  |  |  |  |  |  |  |
Model | |||||||
Part Number | SKY-MXM5000A-6SDA | SKY-MXM3500A-2SDA | SKY-MXM2000A-8SDA | SKY-MXMA4500-6SDA | SKY-MXMA2000-8SDA | SKY-MXMA500-4SHA | SKY-MXMT1000-4HDB |
GPU Architecture | Ada Lovelace | Ada Lovelace | Ada Lovelace | Ampere | Ampere | Ampere | Turing |
GPU Memory | 16GB GDDR6 with ECC | 12GB GDDR6 with ECC | 8GB GDDR6 with ECC | 16GB GDDR6 with ECC | 8GB GDDR6 with ECC | 4GB GDDR6 | 4GB GDDR6 |
Memory Interface | 256-bit | 192-bit | 128-bit | 256-bit | 128-bit | 64-bit | 128-bit |
Max Clock | 9000 MHz | 9000 MHz | 8000 MHz | 8000 MHz | 7000 MHz | 7000 MHz | 6000 MHz |
Memory BW | 576 GB/s | 432 GB/s | 256 GB/s | 512 GB/s | 224 GB/s | 112 GB/s | 192 GB/s |
CUDA Cores | 9728 | 5120 | 3072 | 5888 | 2560 | 2048 | 896 |
RT Cores | 76 | 40 | 24 | 46 | 20 | 16 | - |
Tensor Cores | 304 | 160 | 96 | 184 | 80 | 64 | - |
Tensor Tflops (FP16 Dense/Sparse) | 165/329 | 92/184 | 52/104 | 70/140 | 34/66 | 25/50 | N/A |
Max FP 32 Per f | 41.15 | 23.04 | 12.8 | 17.66 | 8.64 | 6.54 | 2.7 TF |
GPU Clock | 1425 MHz | 1725 MHz | 1635 MHz | 930 MHz | 1117 MHz | 652 MHz | 1395 MHz |
Boost Clock | 2115 MHz | 2250 MHz | 2115 MHz | 1500 MHz | 1612 MHz | 1597 MHz | 1455 MHz |
Form Factor | MXM Type B+ | MXM Type B+ | MXM Type A | MXM Type B+ | MXM Type A | MXM Type A | MXM Type A |
Dimension (L x H) | 82 x 105 mm | 82 x 105 mm | 82 x 70 mm | 82 x 105 mm | 82 x 70 mm | 82 x 70 mm | 82 x 70 mm |
Interface | MXM 3.1, PCIe 4.0 x16 | MXM 3.1, PCIe 4.0 x16 | MXM 3.1, PCIe 4.0 x8 | MXM 3.1, PCIe 3.0 x16 | MXM 3.1, PCIe 3.0 x8 | MXM 3.1, PCIe 3.0 x4 | 89MXM 3.1, PCIe 3.0 x16x6 |
TGP Power | 115 W | 115 W | 60 W | 115 W | 60 W | 35 W | 50 W |
Display Output | 4 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz | 4 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz | 3 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz | 4 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz | 4 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz | Headless Design No Display Output | 4 x DP 1.4a, HDMI 2.1 4K at 120Hz or 8K at 60Hz |
NVENC | 2(8th Gen) | 2(8th Gen) | 1(8th Gen) | 1(7th Gen) | 1(7th Gen) | 1(7th Gen) | 1(6th Gen) |
NVDEC | 5(5th Gen) | 1(5th Gen) | 1(5th Gen) | 2(5th Gen) | 2(5th Gen) | 1(5th Gen) | 3(4th Gen) |
Operating temperature | 0 ~ 55°C (dependent on CPU and cooler solution) | ||||||
Storage temperature | -40 ~ 85°C (-40 ~ 185°F) | ||||||
Vibration (Non-operating) | 2G | ||||||
OS support | Windows 10/11, 64-bit; Linux Drivers, 64-bit | ||||||
✓: supported, – : not supported, △ : optional
Din-Rail Automation Controller (M2I Control Edge)
The UNO-1 series Din-Rail edge controller is built for quick and flexible configuration in edge computing and machine intelligence applications.
Category | M2I Control Edge | |||
| Product image | /UNO-127_Right-_Square20220118114345.jpg) |  | /UNO-148-V2_03--20240328182429.jpg) | |
Model | ||||
CPU | Intel Atom® x6413E processor | Intel Atom® x6413E processor, Intel Atom® x6425RE processor | Intel® Core™ processors (13th Gen) | |
Memory | 1 x DDR4 with 4/8 GB | 1 x DDR4 with 8 GB | 1 x DDR5 with 8 GB | |
Storage | 1 x eMMC (64 GB), 1 x mSATA (shared with mPCIe) | 1 x SATA M.2 SSD slot (2242 M-Key) | 2 x NVMe/SATA M.2 SSD slot (2280 M-Key) | |
I/O Interface | Displays | 1 x HDMI | 2 x DP | 2 x DP |
USB Ports | 2 x USB 3.2 | 3 x USB 3.2, 1 x USB 2.0 | 3 x USB 3.2, 1 x USB 2.0 | |
Serial Ports | 2 x RJ-45 | 2 x RS-232/422/485 | 2 x RS-232/422/485 | |
Expansion | 1 x Full-size mPCIe Slot | 1 x Full-size mPCIe Slot, 1 x M.2 E-Key 2230, 1 x SIM Card Slot | 1 x Full-size mPCIe Slot, 1 x M.2 E-Key 2230 | |
SIM Card Slot | Nano SIM card | Nano SIM card | Nano SIM card | |
Default OS | - | - | - | |
OS Support | Windows 10 LTSC, Windows 11, Ubuntu 20.04, Ubuntu Core 20 | Win 10 2019 LTSC, Win 11, Advantech Linux (Ubuntu 20.04) | Windows 10 LTSC, Windows 11, AdvantechLinux | |
Certification | CE, FCC, UL, CCC, BSM | CE, FCC, UL, CCC, BSMI | CE, FCC, UL, CCC, BSMI | |
Hardware Security | TPM 2.0 | TPM 2.0 | TPM 2.0 | |
Operating Temperature | -20 ~ 60°C | -40 ~ 70°C | -20 ~ 60°C | |
Power Consumption | 18.5W | 21W | 31W | |
Power Input | 10 ~ 30 VDC | 10 ~ 30 VDC | 10 ~ 30 VDC (TBD) | |
Mechanical | Mounting | DIN-rail | DIN-rail | DIN-rail |
Dimension (W x H x D) | 33.5 x 100 x 70 mm | 35 x 150 x 105 mm | 53 x 200 x 140 mm | |
Weight (Net) | 0.5 kg | .8 kg | 1 kg | |
| Product image |  |  |  | |
Part Number | ||||
I/O & Expansion | 2 x RS-232/422/485, 2 x RS-232, 1 x GbE, 1 x M.2 B-Key (SATA/USB 3.0) | 1 x iDoor slot (chassis only) | 1 slot for PCIe x4 | |
Dimension (W x H x D) | 30 x 100 x 70 mm | 35 x 150 x 105 mm | 57 x 200 x 140 mm | |
Description | UNO-127 2nd stack extension kit for 4 x COM & 1 x LAN | UNO-137 V2 2nd stack extension kit for iDoor | UNO Carrier Stack supports MXM GPU module aimed for image processing and AI inference | |
✓: supported, – : not supported, △ : optional