new and original XC7Z020-1CLG400C IC SOC CORTEX-A9 667MHZ 400BGA ic chips integrated circuits electronics components one spot buy

Product Attributes

Accelerating applications with embedded AI

To make adaptive computing more accessible to more users, Xilinx is also committed to improving the ease of use of its products from a software perspective. Highlights include optimized libraries, environments more familiar to software developers, languages, and standard frameworks, including the introduction of TensorFlow capabilities. Especially for the AI developer and data scientist crowd, Xilinx has specifically built the all-in-one platforms Vitis and Vitis AI and introduced open-source neural networks.

To accelerate applications with embedded AI, it is often necessary to consider not only the technical capabilities of AI acceleration but also many other processing units. In this regard, Xilinx has strong flexibility and adaptability to achieve overall acceleration through a one-stop platform capability. It is worth mentioning that, compared to similar products, Celerity accelerates not only AI neural networks, but also multiple AI, and even non-AI businesses, enabling customers to have fully accelerated adaptive computing capabilities.

Celerity is introducing the AI engine under the 7nm Versal architecture, a coarse-grained reconfigurable architecture, a set of programmable logic processing units that support more advanced programming models, called CGRA (Coarse-Grained Reconfigurable Array), which can combine single instruction/multiple data (SIMD) and very long instruction word ( VLIW) into an optimal environment. Simply understood, the 7nm Versal family allows for higher AI inference performance, outperforming traditional CPUs and GPUs by many times in terms of performance per power consumption.

Now, the latest generation of AIE is a 7nm process node, introduced mainly for wireless and aerospace DSP processing, with MLPERF beyond T4. Xilinx hopes to introduce more dedicated data types to serve machine learning, in addition to enabling 2-3x improvement in his base performance.

The data centre ecosystem continues to grow

In the data center market, Xilinx has achieved twice the revenue growth in three years. Again, the revenue growth includes not only chips but also compute, storage, and acceleration cards, among others. The SN1000 SmartNIC, in particular, offers significant performance improvements, including the ability to offload on the CPU, allowing the CPU to do some of the more important processing work, but also enabling some processing to be done closer to the network, including security, compression, and decompression.

To date, Xilinx has developed a unique eco-force in the data center market. There are now over 50 certified servers with close working relationships with Xilinx, including Lenovo, Dell, Wave, HP, and other industry leaders. More than 20,000 trained developers, over 1,000 members with accelerator programs, and over 200 publicly released applications have joined the Celeris eco-army. In the future, developers will also be able to use, purchase and develop Celeris-based applications more efficiently through the new Celeris App Store.

Xilinx’ ability to grow rapidly in the data center market is driven by cloud computing. FPGAs are a key application in cloud infrastructure and workload support, and Celeris has just the right services for this. For example, Amazon AWS’ AQUA, enables the acceleration of Redshift databases. With Xilinx’ technology and products, AWS can help users achieve acceleration in all aspects, including scanning, filtering, encryption, compression, etc., enabling Redshift databases to be accelerated by more than 10 times.

All in all, Xilinx has delivered a perfect answer for users in the past three years. Whether it’s computing, acceleration or AI innovation, or deployments related to 5G, Xilinx has shown very strong growth. And with the acquisition by AMD, Xilinx will build on its original capabilities and take on a new journey.