AMD Reportedly In Talks To Acquire Xilinx

[Peter Fodrek]

Magnum was not enough ))

AMD Chip Codenamed “Magnum” Spotted on Shipping Manifest
5 years ago
A strange entry has appeared on a shipping manifest at Zauba, with AMD’s codename “Magnum” relating to an FPGA prototype board, as well as DTV. This is certainly a strange entry, but what does it mean?

https://www.eteknix.com/amd-chip-codenamed-magnum-spotted-shipping-manifest/

AMD 'Magnum' FPGA chip turns up, possible NVIDIA Shield competitor?
AMD's new codename 'Magnum' shows up, could power a 4K-capable set-top box.
Published Sun, Oct 25 2015 8:35

Read more: https://www.tweaktown.com/news/48153/amd-magnum-fpga-chip-turns-up-possible-nvidia-shield-competitor/index.html


AND AMD patent as well

https://diit.cz/clanek/amd-magnum-fpga

So using Xilinx FGPA as "Reconfigurable Logic Fabric" and use it for example to run Linux kernel and/or module drivers or Game Engines may AMD help in performance....

[skeevy420]

Sounds like a reasonable plan considering Xilinix has built their own ecosystem around deep learning on fpga. Imagine a GPU with a fast fpga included. If the XLA compiler infrastructure could generate code for those FPGAs, one could possibly run parts of the network on GP and other parts on the FPGA and share the same memory. Our configure the FPGA to provide tensor cores. Many things would be possible.

That sounds like RDNA3 & CDNA2

[Adarion]

I was surprised they had the money, but then, yes, "financial engineering". But in all it might actually lead to very interesting outcomes. I hope they'll still support their driver and FOSS side, too, and not hurt themselves lifting and fighting all these little battle on many frontiers.

[ms178]

Xilinx and AMD would suit together quite well, for better FPGA integration on the CPU this aquisition screems to happen sooner rather than later - just look what Intel and Nvidia acquired in the recent past and you will notice that their eyes are also beyond the classical PC ecosystem. Also AMD needs this if they want to get a bigger player in HPC as they are currently relying on partners to complement their technology. But integration of these specialized chips is the next big thing and AMD lacks the manpower, software and hardware expertise in this area to build these technologies from scratch. Next to Xilinx FPGAs, they recently went after SmartNICs/DPUs and I can see that all of these technologies are areas where AMD is lacking but have a huge impact on the data center (and these could even benefit the consoles and consumer PCs of the future).

But I am far from the first one to come to this conclusion, here is a quote from The Next Platform which thought of this in January 2020:"Xilinx is the only standalone company of any size that makes reconfigurable logic devices. Give that, the natural buyer for Xilinx would be AMD, Intel’s arch-nemesis. AMD, of course, already has a highly competitive lineup of CPUs and GPUs to challenge its much larger rival, and the addition of an FPGA portfolio would open a third front. It would also provide AMD entry into a whole array of new application markets where FPGAs operate: ASIC prototyping, IoT, embedded aerospace/automotive, 5G communications, AI inference, database acceleration, and computational storage, to name a few." -- source: https://www.nextplatform.com/2020/01...he-new-normal/

They even thought about a merger of Xilinx, AMD and Broadcom for their synergies in HPC and AI.

So people who might not get it at first, AMD needs to grow from a GPU/CPU company to into a bigger ecosystem player to provide integrated solutions for the data center or they will get crushed by Intel or Nvidia sooner or later.

[scottishduck]

Anyone who might be hoping this means Xilinx stuff being open sourced is going to be sorely disappointed. They provide a lot of black box silicon and there’s no way their clients would allow it.

[CommunityMember]

On the other side, imagine AMD producing ARM processors

AMD is one of the handful of publicly named companies with a 64-bit ARM architectural license, and they already have dabbled in ARM processors (targeted mostly at the server space), so if x86 runs out of mind/market share they could move towards ARM, but it would be a likely be a brutal market with lower profits, so as long as the hyperscalers are still buying x86 by the gazillions they likely will continue to pump them out.

[ThoreauHD]

This could be good for Xilinx on the manufacturing side. Their highest end products are at 16nm. Meanwhile, AMD's stock price just dropped as Xilinx shoots up.

[cl333r]

This could be good for Xilinx on the manufacturing side. Meanwhile, AMD's stock price just dropped as Xilinx shoots up.

I guess because AMD has to pay $30 freaking billions, especially that it's been laying off lots of staff lately.

[CommunityMember]

This could be good for Xilinx on the manufacturing side. Their highest end products are at 16nm.

Actually, Xilinx is using TSMC's 7nm process (same as AMD) for some new(er) products (the versal premium). As always, it is a matter of choosing the appropriate lithography for the product, and not everything needs to be targeted at a future 3nm process.

[zxy_thf]

As a reminder, Intel acquires Altera (another FPGA company) 5 years ago.