Croatia - Flag Croatia

Incoterms:DDP
All prices include duty and customs fees on select shipping methods.

Please confirm your currency selection:

Croatian Kuna
Free shipping on most orders over 400 kn (HRK)
Payment accepted in Credit cards only

Euros
Free shipping on most orders over 50 € (EUR)
All payment options available

US Dollars
Free shipping on most orders over $60 (USD)
All payment options available

Bench Talk for Design Engineers

Bench Talk

rss

Bench Talk for Design Engineers | The Official Blog of Mouser Electronics


Solving Data-centric Challenges with Intel® Agilex™ FPGAs Adam Kimmel

(Source: Intel®)

With the constant expansion of the Internet of Things (IoT), ongoing rollout of 5G broadband networks, evolving IT strategies and roadmaps continue to include edge computing. This emphasis led Gartner to project in its article, What Edge Computing Means for Infrastructure and Operations Leaders, that enterprise edge computing will grow from 10 percent to 75 percent by 2025 through micro-data centers and cellular networks. Data-centric applications, such as 5G and edge computing need to process larger amounts of data, creating the need for advanced processing technology.

Data needs to travel across the entire IT infrastructure, from application to network core to data center and back. Because each end-user application is different, system architects need customizable, flexible data processing to optimize cost and energy efficiency at the point of data generation. Intel® Agilex™ FPGAs combine FPGA flexibility with application-specific integrated circuit (ASIC)’s efficiency and performance levels. This new solution integrates a wide range of semiconductor elements into one system.

Intel® AgilexFPGAs

Built on Intel’s 10nm SuperFin Technology, Intel® Agilex FPGAs provide flexibility and agility beyond previous generation FPGA devices, making them ideal for Industry 4.0 (Figure 1).

Figure 1: Intel® Agilex FPGA offers flexibility and agility to support edge devices, low-latency, and high-bandwidth. (Source: Intel®)

Three product variants exist within the Intel® Agilex™ portfolio: Flexible (F-series), Interface centric (I-series), and Memory centric (M-series). Intel® Agilex™ F-Series FPGAs feature transceiver rates up to 58Gbps, PCIe Gen4 x16, 400Gb Ethernet hard IP, and DDR4 External Memory Support using hard memory controllers. Intel® Agilex™ I-Series FPGAs increase the maximum transceiver rate to 116Gbps, and also include PCIe Gen5 x16 and Compute Express Link (CXL) hard IP. Intel® Agilex™ M-Series FPGAs add support for in-package HBM2 memory, DDR5 External Memory Support and Intel® Optane Persistent Memory support.

Architecture and Design Philosophy

The innovation in the Intel® Agilex FPGA is in the integration of flexibility and agility. Engineers combined a general-purpose, high-performance FPGA core fabric with function-specific chiplets in a single package to create a flexible device architecture. The traditional FPGA fabric makes the product flexible enough to extend to multiple applications. This integration makes the Intel® Agilex flexible to support a wide array of applications.

Although earlier commercial FPGAs contained a single monolithic die or multiple instances of the singular die, Intel® Agilex leverages advanced packaging technology to enable various unique silicon dies within the package.

This enables Intel® Agilex™ FPGAs to excel in four important areas: design re-use, transceivers, interfaces and compute/AI. Intel® Agilex™ FPGAs utilize the second generation Hyperflex core architecture, allowing design and IP reuse from previous generation Intel® Stratix® 10 FPGAs, while offering up to 40 percent higher performance or up to 40 percent lower power dissipation. Both products use the same Quad-core Arm Cortex A53 hard processor preserving investments made in both software, and peripheral design. Intel® Agilex™ FPGAs offer transceiver rates up to 116Gbps PAM4 and 58Gbps NRZ, supported by PCIe, CXL and 400Gb Ethernet hard IP to allow efficient implementation of a wide variety of serial protocols and interfaces to optical modules, processors and other devices. For compute intensive DSP applications, Intel® Agilex™ FPGAs feature a Variable Precision DSP block with hard fixed-point and hard floating-point modes of operation, including FP32, FP16, INT18/19, INT9 and BFLOAT16 number formats that support a variety of applications including AI training and inference.

New Capabilities Unlock New Applications

The features described above illustrate the benefits and capabilities of Intel® Agilex™ FPGAs. Their flexible yet agile design makes Intel® Agilex™ FPGAs ideal for acceleration applications in the data center with both PCIe and CXL processor attachment, new DSP functions for AI, high-performance external memory interfaces such as DDR4, DDR5 and Intel® Optane™ persistent memory, and in-package HBM2 memory. Other applications for Intel® Agilex™ FPGAs include cloud infrastructure, network function virtualization (NFV), automotive, aerospace, manufacturing, and military.

Conclusion

Intel® Agilex™ FPGAs offer a disruptive approach to FPGA architecture, integrating logical design flexibility with customizable agility. Intel® Agilex™ FPGAs deliver on the promise of the Intel acquisition of Altera with impressive advancements in silicon and solutions, enabling a broader market, while staying true to the well-established FPGA user community. This transformative device’s efficiency and performance gains will dramatically improve the user experience and can elevate your brand and product portfolio to a position of strength in new and emerging markets.



« Back


Adam KimmelAdam Kimmel has nearly 20 years as a practicing engineer, R&D manager, and engineering content writer. He creates white papers, website copy, case studies, and blog posts in vertical markets including automotive, industrial/manufacturing, technology, and electronics. Adam has degrees in chemical and mechanical engineering and is the founder and principal at ASK Consulting Solutions, LLC, an engineering and technology content writing firm.


All Authors

Show More Show More
View Blogs by Date