AMD set to buy Xilinx for R564 billion in stock

[Hanno Labuschagne]

AMD set to buy Xilinx for R564 billion in stock

Advanced Micro Devices Inc. agreed to buy Xilinx Inc. for $35 billion in stock, taking the chipmaker into more diverse and profitable markets and adding to its data center offerings.

Xilinx investors will get 1.7234 AMD shares for each Xilinx stock they own. That values Xilinx at about $143 a share, 25% more than the closing price on Monday and 35% above the price before news of a possible deal was reported earlier in October.

The deal is a coup for AMD Chief Executive Officer Lisa Su, creating a company with a larger research-and-development budget and a broader array of products to take on Intel Corp.

[Bloomberg]

 

[Rickster]

Thats 2,088,888,888 21 piece KFC buckets.

Must suck to be starving and know this.

 

[sjm]

Thats 2,088,888,888 21 piece KFC buckets.

Must suck to be starving and know this.

It would suck to be starving and be offered a 21 piece KFC bucket.

 

[Sarg3_ZN]

Thats 2,088,888,888 21 piece KFC buckets.

Must suck to be starving and know this.

Now you have piqued the interest of tenderpreneurs all over SA especially the ones planning a career in politics

 

[newby_investor]

I actually have a little bit of both AMD and Xilinx. Guess my holdings are being consolidated.

Still, I am wondering what AMD is hoping to gain out of this, other than the ability to compete more directly with Intel which now owns Altera.

 

[DailyTech]

FPGAs , yeah

 

[cguy]

I actually have a little bit of both AMD and Xilinx. Guess my holdings are being consolidated.

Still, I am wondering what AMD is hoping to gain out of this, other than the ability to compete more directly with Intel which now owns Altera.

Standard CPU architectures are going to go extinct over the next decade or so for anything related to HPC. As time goes by, the HPC of today becomes the norm a decade or two later, so this is likely a play to stay relevant long term.

 

[newby_investor]

Standard CPU architectures are going to go extinct over the next decade or so for anything related to HPC. As time goes by, the HPC of today becomes the norm a decade or two later, so this is likely a play to stay relevant long term.

I've used FPGAs, both Altera and Xilinx.

The bigger they get, the more of a pain they get to work with. And Xilinx aren't exactly helpful to prospective developers (maybe if you dangle the prospect of billions of dollars in front of them). In my experience, GPUs are much better for HPC.

In certain situations an FPGA works better, such as connected directly to an ADC like in Xilinx's RFSoCs, but for general purpose compute... Well maybe things will change as time progresses. I hope AMD have the wherewithal to manage Xilinx properly.

 

[cguy]

I've used FPGAs, both Altera and Xilinx.

The bigger they get, the more of a pain they get to work with. And Xilinx aren't exactly helpful to prospective developers (maybe if you dangle the prospect of billions of dollars in front of them). In my experience, GPUs are much better for HPC.

In certain situations an FPGA works better, such as connected directly to an ADC like in Xilinx's RFSoCs, but for general purpose compute... Well maybe things will change as time progresses. I hope AMD have the wherewithal to manage Xilinx properly.

We use both pure FPGAs and GPUs for HPC. They solve different cross sections of the HPC space. GPUs are nearly useless, when the problem can't be cast as mostly SIMD. Xilinx's Versal is trying to solve for situations where neither of them are quite right by themselves - it's actually more along this line than the pure FPGAs I was referring to. Xilinx is no longer just an FPGA company.

 

[newby_investor]

We use both pure FPGAs and GPUs for HPC. They solve different cross sections of the HPC space. GPUs are nearly useless, when the problem can't be cast as mostly SIMD. Xilinx's Versal is trying to solve for situations where neither of them are quite right by themselves - it's actually more along this line than the pure FPGAs I was referring to. Xilinx is no longer just an FPGA company.

I haven't used Versal ones. We used Virtex-5 through Virtex-7 FPGAs which did the job when GPUs couldn't, though granted our application is pretty solidly SIMD (for the most part).

We started looking at Alveos which seemed promising, but in the end they were practically pretty difficult to work with, and the support we received wasn't great. We'd basically need to reverse-engineer them in order to actually use the amazing capability that they supposedly have, because Xilinx certainly haven't tried very hard to make development very easy.

I've seen marketing material for the Versal products but I haven't talked to anyone who has used one successfully. It sounded really cool if it could do what Xilinx say it could, but I remain skeptical until I can actually try it out and see it working.

 

[dualmeister]

Xilinx's RFSoCs,

Best selling product

 

[konfab]

I actually have a little bit of both AMD and Xilinx. Guess my holdings are being consolidated.

Still, I am wondering what AMD is hoping to gain out of this, other than the ability to compete more directly with Intel which now owns Altera.

They are wanting to take on Apple and Nvidia as well methinks.
The most obvious market is inference for CNNs.

 

[cguy]

I haven't used Versal ones. We used Virtex-5 through Virtex-7 FPGAs which did the job when GPUs couldn't, though granted our application is pretty solidly SIMD (for the most part).

We started looking at Alveos which seemed promising, but in the end they were practically pretty difficult to work with, and the support we received wasn't great. We'd basically need to reverse-engineer them in order to actually use the amazing capability that they supposedly have, because Xilinx certainly haven't tried very hard to make development very easy.

I've seen marketing material for the Versal products but I haven't talked to anyone who has used one successfully. It sounded really cool if it could do what Xilinx say it could, but I remain skeptical until I can actually try it out and see it working.

I don’t think I’ve ever met anyone who hasn’t thought the FPGA toolchains are anything but a dumpster fire. . It’s one of the reasons the Versal is relatively attractive - lots of CPUs (once all set up at least - and interfacing with the rest of the chip is still a nightmare).

 

[newby_investor]

They are wanting to take on Apple and Nvidia as well methinks.
The most obvious market is inference for CNNs.

Well... Nvidia now pretty much has Apple by the sensitives with the Arm acquisition. Interesting to see how that will play out.

 

[konfab]

Well... Nvidia now pretty much has Apple by the sensitives with the Arm acquisition. Interesting to see how that will play out.

ARM's entire value is in its licence. If Nvidia does anything stupid, Intel will be there to lap up Apple.

 

[dualmeister]

NVIDIA bought Mellanox. Now NVIDIA Networking.

 

[DailyTech]

agree