GPUs are used in embedded applications demanding high-resolution, multi-display capability, parallel computing, and AI enablement. ADLINK sees applications such as clinical diagnostic imaging devices, rugged military laptops and imaging radar for aerospace and defense, and gaming slot machines as presenting considerable potential to benefit from embedded graphics.

Top-down applications that can benefit from embedded graphics further include:

• High-resolution, multi-display setup: Applications include air traffic control, electronic chart displays and information systems (ECDIS), video walls, digital signage, gaming, and healthcare environments
• Parallel computing: High-performance application processing including radar/ sonar systems in aerospace and defense, ultrasound imaging in healthcare, and accelerated multi-access edge computing (AMEC) in telecoms
• AI engines: System training and inferencing in smart manufacturing, smart city, telecom, aerospace and defense, and transportation

Longevity support: Product availability is 18 months for consumer-grade NVIDIA® GeForce® GPUs and 3 years for NVIDIA Quadro GPUs while NVIDIA Quadro Embedded GPUs support 5+ year availability in line with embedded application requirements. ADLINK MXMs support 5-year availability. PEG cards support 3-year availability which can be extended to additional 5 years under an extended life program.

Lower development costs: Longevity support translates into reduced development time, effort, and costs associated with graphics entering EOL. Development costs can reach six figures(i) for safety-critical applications that undergo time-consuming verification and testing processes. As consumer-grade graphics could reach EOL three times through the lifecycle of an embedded graphics, it is obvious that developing with embedded graphics would yield a superior return on investment (ROI).

Increased system responsiveness: NVIDIA GPUDirect RDMA in NVIDIA Quadro Embedded GPUs can boost data throughput by approximately 80% and lower latency by 60%(ii) . A direct path for data exchange is established between the GPU and third-party devices(iii), eliminating the need to copy data to CPU memory before reaching the GPU, such that data throughput and system responsiveness are significantly increased for GPU-accelerated applications.

i Product Development Cost and Timelines, https://www.acornpd.com/blog/product-development-cost-and-timelines
ii The Software and workloads used in performance tests may have been optimized for performance only on ADLINK platforms. Performance tests are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. Contact ADLINK for more complete information about performance and benchmark results.
iii NVIDIA GPUDirect™ Technology, http://developer.download.nvidia.com/devzone/devcenter/cuda/docs/GPUDirect_Technology_Overview.pdf.

For graphics-intensive applications involved with high-resolution, multi-display setup, please refer to performance benchmark testing results in Figure 1. For compute-intensive applications involved with highly-parallel workloads or AI training and inferencing, please refer to performance benchmark testing results in Figure 2.

Figure 1: Passmark G3D Benchmark

Figure 2: DirectCompute Benchmark

High-resolution, multi-display setup, parallel computing, and AI engines are three common use of graphics. Other graphics products supporting long life availability and more than four display outputs have an advantage over ADLINK Embedded Graphics products in the first use case scenario. As for highly-parallel computing and AI applications, ADLINK Embedded Graphics products can reduce CPU overhead, boost data throughput by approximately 80%, lower latency by 60%, and increase system responsiveness.

ADLINK’s Mobile Express modules (MXMs), with high performance per watt and extended operating temperature options, are an ideal fit for mobile, SWaP-constrained, passive-cooling applications such as portable ultrasound, airborne radar, and aerial infrared imaging.

PCI Express Graphics (PEG) cards connect via a common interface and are easy to integrate and use in many embedded market segments, such as healthcare (magnetic resonance imaging (MRI) and computed tomography (CT)), industrial automation (AOI inspection), and telecom multi-access edge computing.

ADLINK is the top choice for embedded customers. Among only four NVIDIA Quadro Embedded Partners in the world, ADLINK is the only one with market-proven experience offering diverse product portfolios that address embedded application needs. ADLINK’s offerings include embedded graphics in MXM and PEG form factors and MXM- and PEG-compatible platforms and modules to accelerate and facilitate GPU adoption in embedded applications.

ADLINK is the top choice for embedded customers. Among only four NVIDIA Quadro Embedded Partners in the world, ADLINK is the only one with market-proven experience offering diverse product portfolios that address embedded application needs.

ADLINK, as an NVIDIA Quadro Embedded Partner, can provide technical support for GPU firmware customization, an important consideration for embedded customers.

ADLINK’s offerings include:

• Embedded graphics in MXM and PEG form factors and MXM- and PEG-compatible platforms and modules to accelerate and facilitate GPU adoption in embedded applications.
• GPU onboard solutions in different verticals (e.g. gaming and medical ) and in various form factors (e.g. PC/104, VPX…) to fulfil ODM projects

ADLINK and our distributors offer MXM modules powered by NVIDIA Quadro Embedded GPUs with 5-year availability and technical support that cannot be offered by online vendors. Online NVIDIA GPU-powered MXMs are highly likely gray market goods without warranty or longevity support. Additionally, Quadro-powered MXMs are for embedded usage, requiring VBIOS, system BIOS and cookie support to drive.

Product availability is 18 months for consumer-grade NVIDIA® GeForce® GPUs and 3 years for NVIDIA Quadro GPUs while NVIDIA Quadro Embedded GPUs support 5+ year availability in line with embedded application requirements. ADLINK MXMs support 5-year availability. PEG cards support 3-year availability which can be extended to additional 5 years under an extended life program. See next slide for details.

Some Jetson platforms are available through 2025. The long life support is in line with embedded application requirements.