It is natural to wonder what AMD would do with its latest acquisition of Xilinx. The deal had been to make the acquisition for $35 billion, being the proposed amount of October 2020. However, an appreciation of the AMD share price in the last couple of months led to an increase in the bid, up to the record price of $49 billion.
AMD is one of the leading manufacturers of semiconductors. Its acquisition of competitor Xilinx can be a major boost to its expansion into other markets where the former company (Xilinx) held sway.
In this blog post, you will find out just what AMD intends to do with the acquisition and the overall impact in the FPGA market.
Why AMD Made the Acquisition
Mergers and Acquisition (M&A) are one of the commonest ways to make new ventures in the business world. For some time now, AMD has been looking to expand its semiconductor market. Xilinx, now a sub-company of AMD, was one of the major competitors. Xilinx is popular for being one of the leading manufacturers of Field Programmable Gate Arrays (FPGAs).
In terms of market size, the acquisition helps AMD to tap into additional markets and which is currently estimated to be $135 billion by the end of the year 2023. Formerly, AMD’s market share was pegged at $79 billion, thus we should expect up to $50 billion worth of market shares to flow into AMD’s coffers from the Xilinx acquisition.
AMD’s FPGA Offerings
AMD, before its Xilinx acquisition, has been making important moves to develop different kinds of FPGAs for several applications.
According to HardwarEsfera, the company (AMD) had sometime in 2021 began working on a chip that allows for the combined functions of a Field Programmable Gate Array (FPGA) and a Central Processing Unit (CPU).
For emphasis, both the CPU and the FPGA are two of the popular architectures for building different kinds of configurable devices. The combined functions of the two goes a long way to facilitate the integration of the different features to make the applications work much better.
From the snippets of the filed patent titled, “Method and Apparatus for Effective Programmable Instructions in Computer Systems,” the company stated the resolution to bolster configurable device designs.
The combination of CPU and FPGA in one package helps to accelerate Machine Learning (ML) for different data format operations and sharable records via Int EUs and FP.
FPGA AI Engine Fusion with EPYC Processors
AMD has also made plans to fuse the FPGA AI Engine onto the EPYC Processors. According to Tom’s Hardware, the fusion is engineered to arm the different chips from AMD with the Field Programmable Gate Array (FPGA) functionalities.
The fusion will be a combination of the FPGA-powered AI inference engine with the wide range of Central Processing Units (CPUs) by AMD. The process also supports the connection of AI accelerators to the AMD Processors, especially by using the 3D chip sophisticated stacking tech.
How it Works
Although the specifics of the fusion aren’t in the public domain for now, feelers are already dropping on what is to be expected.
First, the AMD FPGA AI Engine fusion with EPYC Processors is designed to accommodate multiple configurations, especially in the area of supporting multiple packaging options.
Second, the expected year of release for these new chips is sometime in 2023.
Third and most important, the fusion will still maintain some level of independence. For example, the Inference Engine by Xilinx will be utilized for inference purposes only. On the other hand, the AMD CPUs and GPUs will be used for addressing some of the complex workloads, associated with Artificial Intelligence (AI). To put it into perspective, the Graphic Processing Units (GPUs) will enable the training of the AI models via a compute-intensive task process.
Napatech and AMD: A Combined Effort for the Future of Datacenter Computations
In September of 2022, the duo of Napatech and AMD have solidified their plans to facilitate the deployment of SmartNICs solutions to AMD customers. The report by Yahoo Finance! shows that the two companies are on the pathway of making several datacenter services available to AMD customers all over the world.
What are SmartNICs?
SmartNICs stand for Smart Network Interface Cards. Napatech manufactures these cards for the primary uses in the Infrastructure Processing Unit (IPU) and Data Processing Unit (DPU).
The SmartNICs are primarily used for facilitating the services offered by the DPU and IPU in the following markets:
- Financial applications
- Cloud computing
The collaboration between AMD and Napatech allow the latter to extend its SmartNICs to the existing AMD customers. Experts from both companies can also come up with new ideas on how to provide additional end-to-end solutions for the processor subsystem and network infrastructure solutions from both companies.
For the AMD customers, they will benefit from the simplified process of installing a SmartNICs and using the same to load a driver that grants them access to both open and commercial applications.
AMD’s FPGA Venture and the Future
AMD made a great investment in acquiring Xilinx, the leading Field Programmable Gate Array (FPGA) that has been in the business since the 1980s. However, the competition just grew with the acquisition.
Other industry players like Intel are still possessing a lot of market shares that AMD would most likely be interested to have. AMD’s FPGA-inclined acquisition of Xilinx wouldn’t be the first in the FPGA market. Intel made a similar move in 2015 when it acquired Altera, making it one of the leading manufacturers of FPGAs at the time.
With AMD coming into the fold, it becomes a more competitive market for the trio of AMD, Intel and NVIDIA.
In its comparison, Seeking Alpha opines that Intel holds the trump card for several reasons. The company’s Agilex FPGA promises up to 2x performance than the Xilinx Versal, as well as offering up to 40% improved performance than AMD-Xilinx FPGAs.
In an industry driven by speed, it makes a good stance that Intel may have an upper hand in the FPGA manufacturing.
As AMD puts it, the plans for the Xilinx acquisition are for the “long term.” So far, the company is doing a great job and more innovations need to be in place to beat the existing competitions.