AMD Completes Acquisition of Xilinx

[EverydayMuffin]

Xilinx are now the "AMD Adaptive and Embedded Computing Group".

https://www.amd.com/en/press-releases/2022-02-14-amd-completes-acquisition-xilinx

It will be interesting to see what AMD does with Xilinx's lower-end devices like CoolRunner, Spartan, Artix and Zynq-7000...

[SiliconWizard]

 

[DiTBho]

Not a good day for me. That's not a good news to hear. But we will see.
I like Xilinx's devices. I have several projects based on Spartan3 and Spartan6

[FlyingDutch]

OMG! It is like a curse ...

[woofy]

AMD used to have its own programmable logic line in the form of the mach1, 2,3,4 & 5 devices. I have used both mach 1 and 2 parts in the past.
They sold their vantis subsidiary to Lattice in 1999 for $500m.
With intel buying Altera, its not surprising AMD wants its own top end FPGA technology. Otherwise if intel ever fused x64 with FPGA, AMD would be left behind.

[Beta_vulgaris]

AMD used to have its own programmable logic line in the form of the mach1, 2,3,4 & 5 devices. I have used both mach 1 and 2 parts in the past.
They sold their vantis subsidiary to Lattice in 1999 for $500m.
With intel buying Altera, its not surprising AMD wants its own top end FPGA technology. Otherwise if intel ever fused x64 with FPGA, AMD would be left behind.

AMD also have had their Am2000/Am3000 FPGA (LCA) family, second-source of Xilinx, but aborted in 1989.

[Bassman59]

AMD used to have its own programmable logic line in the form of the mach1, 2,3,4 & 5 devices. I have used both mach 1 and 2 parts in the past.
They sold their vantis subsidiary to Lattice in 1999 for $500m.
With intel buying Altera, its not surprising AMD wants its own top end FPGA technology. Otherwise if intel ever fused x64 with FPGA, AMD would be left behind.

AMD also have had their Am2000/Am3000 FPGA (LCA) family, second-source of Xilinx, but aborted in 1989.

They did PALs and such ages ago. Did they buy MMI?

[RoGeorge]

Each such consolidation usually means an unimaginable big amount of money.  But I don't get the economics.  This is not just a startup selling all as a good exit, to get an early retirement.

What happens with that pile of money afterwards?
Did the buyer sends a barge full of cache, or what?   
And what does the seller do with a pile of money?

[nctnico]

Next thing we know both FPGA branches get split off into one company and we get Xiltera. Honestly I don't see how Altera fits with Intel and how Xilinx fits with AMD. In the past both companies have bought so many other companies to end up either killing it or selling it again.

[Bassman59]

Each such consolidation usually means an unimaginable big amount of money.  But I don't get the economics.  This is not just a startup selling all as a good exit, to get an early retirement.

What happens with that pile of money afterwards?

The AMD/Xilinx deal was all stock, no cash money transfer.

Did the buyer sends a barge full of cache, or what?   

The buyer (AMD) issued shares of its stock, which it gave to Xilinx shareholders in exchange for those shareholders' Xilinx stock. The former Xilinx shareholders now own part of AMD.

And what does the seller do with a pile of money?

Well, what would you do with a pile of money?

[SiliconWizard]

Next thing is AMD merges with Intel.

[RoGeorge]

The AMD/Xilinx deal was all stock, no cash money transfer.

In that case it would be rather a merge than an acquisition, but I won't nitpick about the terms.

\[\star\ \star\ \star\]

Well, what would you do with a pile of money?

That's such a tempting thought, way too nice not to dream about it. 

I'll probably build a top notch research facility, then invite as many great minds as possible.  Not as employees, but invite them to join in what would be like a paid sabbatical decade, or so.  In short, a Scientific Research Heaven.

[nctnico]

I'll probably build a top notch research facility, then invite as many great minds as possible.  Not as employees, but invite them to join in what would be like a paid sabbatical decade, or so.  In short, a Scientific Research Heaven.

Why only a decade? I used to work at such a place; it is called a research institute and those exist already. If only the work didn't drift towards doing software engineering... A couple of months ago I visited a EU research center. The toys they have over there! Where is the drool smiley?

[RoGeorge]

Why only a decade?

A sabbatical is usually one year, but that's probably too little to put in practice a new idea.

Ten times that, seems about right to mature something new, without overthinking it.  More than 10 years would probably mean it's not the right technological time for that idea.  And anyways, times changes, so most probably 20 or 50 years later the research theme would become irrelevant.

The time frame was more about the theme/project than about the researcher.

[KE5FX]

Not a good day for me. That's not a good news to hear. But we will see.
I like Xilinx's devices. I have several projects based on Spartan3 and Spartan6

How can AMD possibly do worse than Xilinx when it comes to what matters, meaning actually delivering parts?  Have you tried to buy any mainstream FPGAs lately?   

[asmi]

How can AMD possibly do worse than Xilinx when it comes to what matters, meaning actually delivering parts?

Delivering parts is the easy part, current issues notwithstanding. The hard one is providing efficient environment and infrastrtucture for using these parts in the actual end products. Xilinx is the best on this front at this point, but we will see if it's going to stay that way in the future.

[KE5FX]

Delivering parts is the easy part, current issues notwithstanding. The hard one is providing efficient environment and infrastrtucture for using these parts in the actual end products.

Not sure what that means.

[SiliconWizard]

Not a good day for me. That's not a good news to hear. But we will see.
I like Xilinx's devices. I have several projects based on Spartan3 and Spartan6

How can AMD possibly do worse than Xilinx when it comes to what matters, meaning actually delivering parts?  Have you tried to buy any mainstream FPGAs lately?   

Can you explain how the current shortage has anything to do with anything here?
Is Xilinx doing any worse than others? I don't get it.

At least with Intel, they have their own foundries, so I could understand why one could think that it could improve the situation (being less dependent on external foundries), but AMD does not AFAIK, so I can't see how that would change anything. Oh, and by the time the acquisition would have any real impact on day-to-day operations, we can hope that the shortage situation would be over anyway (although I admit the more time passes and the less I know when it will possibly end.)

But feel free to give us more details.

[KE5FX]

Can you explain how the current shortage has anything to do with anything here?
Is Xilinx doing any worse than others? I don't get it.

At least with Intel, they have their own foundries, so I could understand why one could think that it could improve the situation (being less dependent on external foundries), but AMD does not AFAIK, so I can't see how that would change anything. Oh, and by the time the acquisition would have any real impact on day-to-day operations, we can hope that the shortage situation would be over anyway (although I admit the more time passes and the less I know when it will possibly end.)

But feel free to give us more details.

Tons of mainstream parts from both Xilinx and Intel/Altera are unavailable from authorized distributors.  Tons of parts other than FPGAs from other vendors are available.  That's about the size of it. 

If you're actively in the market for semiconductors right now, you'd find it unnecessary to ask that question.  If you aren't, count yourself lucky, and consider yourself envied.

[ali_asadzadeh]

This would open a free and easy space for Chinese companies to destroy them both! With such aggressive pricing and High tech from chines companies it would be a matter of 10 years or less till we forget about Xilinx and Altera.

[AndyC_772]

Unfortunately both companies are now in the hands of firms who are really only interested in the big devices, that they can incorporate into high performance servers and HPC clusters. That's where the demand, and the money, is.

That means companies that rely on the smaller parts, to implement critical features that simply can't be done any other way than with an FPGA, are in serious trouble - and it's not going to get better with time. I'm genuinely concerned that the Altera parts I've designed in over, say, the last 10 years, may never be manufactured again.

My to-do list for this week includes unpacking and getting started with the Efinix dev board that's sitting on my desk.

[DiTBho]

my order request is still in pending state:
Tue Nov 17, 2020:
"possible to order, delivery time on request
expected to be available on 01-Sep-2021"

Wed Feb 16, 2022
"possible to order, delivery time on request
expected to be available on 01-Nov-2022"

WTF?!?, but also LOL   

[Ed.Kloonk]

Next thing is AMD merges with Intel.

Those two are the classic abusive relationship. They take from each other. They sue each other.

Then they both complain that the neighbors house (ARM) is regarded as the best looking house in the street.

Imagine just how great their house would be if they just fucking worked together.

[SiliconWizard]

Can you explain how the current shortage has anything to do with anything here?
Is Xilinx doing any worse than others? I don't get it.

At least with Intel, they have their own foundries, so I could understand why one could think that it could improve the situation (being less dependent on external foundries), but AMD does not AFAIK, so I can't see how that would change anything. Oh, and by the time the acquisition would have any real impact on day-to-day operations, we can hope that the shortage situation would be over anyway (although I admit the more time passes and the less I know when it will possibly end.)

But feel free to give us more details.

Tons of mainstream parts from both Xilinx and Intel/Altera are unavailable from authorized distributors.  Tons of parts other than FPGAs from other vendors are available.  That's about the size of it. 

Tons of parts are unavailable in general. The size of it is, it's everywhere.

If you're actively in the market for semiconductors right now, you'd find it unnecessary to ask that question.

I am, as many of us here are I guess. And shortage is everywhere, well beyond FPGAs. I even have better luck getting ahold of Lattice FPGAs at the moment than getting STM MCUs.
And you never answered the question: what difference would AMD acquisition ever make to the shortage problem?

If you aren't, count yourself lucky, and consider yourself envied.

Count yourself lucky if a given part you need right now is available at all.

[FlyingDutch]

Hi,

but Gowin FPGAs are constantly available .

Best regards