Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AMD Receives Approval for Acquisition of Xilinx

amd.com

Post reply on HN
Print view View original

51–60 of 69 posts

Re: AMD Receives Approval for Acquisition of Xilinx

#51

Post by throwmeawaysoon »

vzidex wrote: post #49
hcrisp wrote: post #39

Earlier quoted context omitted.

You don't seem bullish on the prospects of using Vitis [0] to deploy a machine learning model to a Xilinx FPGA? [0] https://www.xilinx.com/products/design-tools/vitis/vitis-pla...

Disclaimer: I work in this space (not at Xilinx), comments are strictly my own opinions and do not reflect any positions of my employer, etc. Broadly speaking, FPGA-based ML model accelerators are in an interesting space right now, where they aren't particularly compelling from a performance (or perf / Watt, perf / $, etc.) perspective. If you just need performance, then a GPU or ASIC-based accelerator will serve you…

>come up with a "better" (performant, cheaper, easier to use, etc.) solution than GPUs for ML applications

you probably are aware but Xilinx themselves is attempting this with their versal aie boards which (in spirit) similar to GPUs, in that they group together a programmable fabric of programmable SIMD type compute cores.

https://www.xilinx.com/support/documentation/architecture-ma...

i have not played with one but i've been told (by a xilinx person, so grain of salt) the flow from high-level representation to that arch is more open

https://github.com/Xilinx/mlir-aie

Re: AMD Receives Approval for Acquisition of Xilinx

#52

Post by marcofatica »

ZetaZero wrote: post #22
chomp wrote: post #11

Earlier quoted context omitted.

> AMD’s acquisition of ATI is generally viewed as a strategic error. Historically it was (when AMD was drowning in debt from the acquisition and had no competitive processors on the timeline), but is this still true? For instance, all of the major game consoles (ps4/ps5/xbone/xbox series x/switch) use AMD boards, and as I recall, those deals lifted the company's financials quite well.

It's complicated. The margins on console chips are lower than PC margins. When AMD had spare capacity (GloFo), that was great. But with everyone at TSMC and AMD demand high, those wafers could be used for more profitable chips. But the console deals definitely helped AMD in the dire times. IMO, the strategic error was AMD paying a bunch of cash for ATI, instead of an all stock deal. In 2006, with their stock near the…

AMD still uses GloFo

Re: AMD Receives Approval for Acquisition of Xilinx

#53

Post by opportune »

vzidex wrote: post #16
gjsman-1000 wrote: post #2

If I have one wish for AMD, I wish that they would make FPGAs (and Xilinx) a more open and diverse platform like the PC. Not that the PC is perfect (there's still closed-source firmware), but any improvement to the current state of FPGAs would be welcomed.

You've struck on the fundamental problem that the FPGA industry has been trying to solve for 30+ years - how to get an FPGA into the hands of every developer, like how GPUs have propagated to be essential tools. Nobody has come up with a good answer yet. Developing for an FPGA still requires domain-specific knowledge, and because place & route (the "compile" for an FPGA) is a couple of intertwined NP-hard problems de…

Why not ship integrated FPGA in CPUs?

Being able to offload a repeated, complex MIMD computation to an FPGA treated like an instruction could be a huge win for scientific computing and any large, steady workload that is expensive enough for companies to invest in optimizing for the FPGA. If this became commonplace and relatively inexpensive then large corporations would likely fund improvements into compilers to make the developer experience simpler and faster.

Re: AMD Receives Approval for Acquisition of Xilinx

#54

Post by opportune »

gjsman-1000 wrote: post #6
brian_herman wrote: post #5

Earlier quoted context omitted.

Yeah a co processor with fpga on it or even an expansion card I can buy at micro center for 100 bucks would be great!

Or even better, a useful FPGA that is inside the Ryzen processor itself. It might be very small, but if it was standardized, had an easier way to program it, and programs could load acceleration routines into it, that would be so cool.

This is what I’ve been thinking about for ages. It could be a huge accomplishment and greatly improve efficiency for scientific computing, data analytics and OLAP, cloud gaming, general backend development, etc

Re: AMD Receives Approval for Acquisition of Xilinx

#55

Post by vzidex »

throwmeawaysoon wrote: post #51
vzidex wrote: post #49

Earlier quoted context omitted.

Disclaimer: I work in this space (not at Xilinx), comments are strictly my own opinions and do not reflect any positions of my employer, etc. Broadly speaking, FPGA-based ML model accelerators are in an interesting space right now, where they aren't particularly compelling from a performance (or perf / Watt, perf / $, etc.) perspective. If you just need performance, then a GPU or ASIC-based accelerator will serve you…

>come up with a "better" (performant, cheaper, easier to use, etc.) solution than GPUs for ML applications you probably are aware but Xilinx themselves is attempting this with their versal aie boards which (in spirit) similar to GPUs, in that they group together a programmable fabric of programmable SIMD type compute cores. https://www.xilinx.com/support/documentation/architecture-ma... i have not played with one but…

Fascinating, thank you! Admittedly I don't keep the closest tabs on what Xilinx is doing.

Re: AMD Receives Approval for Acquisition of Xilinx

#56

Post by mindslight »

vzidex wrote: post #16
gjsman-1000 wrote: post #2

If I have one wish for AMD, I wish that they would make FPGAs (and Xilinx) a more open and diverse platform like the PC. Not that the PC is perfect (there's still closed-source firmware), but any improvement to the current state of FPGAs would be welcomed.

You've struck on the fundamental problem that the FPGA industry has been trying to solve for 30+ years - how to get an FPGA into the hands of every developer, like how GPUs have propagated to be essential tools. Nobody has come up with a good answer yet. Developing for an FPGA still requires domain-specific knowledge, and because place & route (the "compile" for an FPGA) is a couple of intertwined NP-hard problems de…

If the FPGA industry thinks it has been trying to do this for decades, then it has been going about it seriously wrong! Keeping your systems as black boxes, with unit prices and development prices that make them prohibitive for anything but high margin device, effectively guarantees they'll never become popular consumer commodities.

With how open development works, the straightforward minimal investment is to publicly document some devices' bitstream formats and bootstrap the ecosystem by releasing some reliable Libre place and route software. The software doesn't even have to contain all of the trade secret heuristics, it just has to work with (./configure && make && make install) and be functionally adequate enough that individual developers can scratch their own itches.

Re: AMD Receives Approval for Acquisition of Xilinx

#57

Post by sangnoir »

systemvoltage wrote: post #10
gjsman-1000 wrote: post #4

Earlier quoted context omitted.

To be fully honest, I don't understand why companies do so much business in China. Yes, there's money, but there's so much danger and risk involved it increasingly looks like betting on Bitcoin for your retirement. An example of this would be what happened to the ARM China division, which has gone completely rogue and the Chinese government couldn't care less. There's also the fact that, Intel and AMD for making the…

Companies aren't to blame. Our lawmakers are. They need to pass laws that protect US interests but for last 3 decades, they've sold the American worker's soul to the CCP & private-gov enterprise in China. Apple spent $275 billion to boost Chinese manufacturing, but fuck-all to do anything in US. They could have built a whole ecosystem of US electronics manufacturers and make everything in US + Mexico, but their share…

> They need to pass laws that protect US interests but for last 3 decades, they've sold the American worker's soul to the CCP & private-gov enterprise in China.

Companies are not blameless - it's a 2-step process. Lawmakers sold the American worker's soul to capital (companies), and the companies sold them on to China, with a fat markup.

The motivations are clear: politicians get donations and cushy jobs, the company executives get very rich due to increased short-term profits and increased valuations on the hope of establishing a foothold in a humongous growth market, China gets tech transfer, exports and a skilled workforce, and American workers get nothing, save perhaps a tiny sliver of company ownership through a 401k, if they are lucky to have one.

Re: AMD Receives Approval for Acquisition of Xilinx

#58

Post by jlokier »

dragontamer wrote: post #48
intricatedetail wrote: post #41

I hope they won't kill their CPLD line that has been obsoleted several times and then pushed back. It's a lifesaver for designing digital circuits if you don't want to use a ton of 74 chips.

It was my understanding Lattice's FPGAs / CPLDs are more tailored for the smaller (possibly hobbyist scale) EEs out there. Bonus points: Lattice has turned a blind eye at the open source community, meaning Lattice's FPGAs / CPLDs have open-source toolsets. Kinda sad that "blind eye" is the best we can hope for from these companies, but that's how it is for now...

Not all these companies.

QuickLogic openly embraces open source tooling.for their FPGAs.

Re: AMD Receives Approval for Acquisition of Xilinx

#59

Post by pjmlp »

gjsman-1000 wrote: post #2

If I have one wish for AMD, I wish that they would make FPGAs (and Xilinx) a more open and diverse platform like the PC. Not that the PC is perfect (there's still closed-source firmware), but any improvement to the current state of FPGAs would be welcomed.

PC are only open due to IBM's failure to prevent Compaq's endeveours.

If anything, we have seen the whole industry moving back to those days as means to get out of razor thin margins, specially now that desktops are a very niche market for most consumers.

Re: AMD Receives Approval for Acquisition of Xilinx

#60

Post by pjmlp »

opportune wrote: post #53
vzidex wrote: post #16

Earlier quoted context omitted.

You've struck on the fundamental problem that the FPGA industry has been trying to solve for 30+ years - how to get an FPGA into the hands of every developer, like how GPUs have propagated to be essential tools. Nobody has come up with a good answer yet. Developing for an FPGA still requires domain-specific knowledge, and because place & route (the "compile" for an FPGA) is a couple of intertwined NP-hard problems de…

Why not ship integrated FPGA in CPUs? Being able to offload a repeated, complex MIMD computation to an FPGA treated like an instruction could be a huge win for scientific computing and any large, steady workload that is expensive enough for companies to invest in optimizing for the FPGA. If this became commonplace and relatively inexpensive then large corporations would likely fund improvements into compilers to make…

There are such CPUs, and the uptake has been minimal, because as proven by GPGPUs not every developer is capable of actually use them.

Your example could be as easily done in a GPGPU.

Post reply on HN