Hacker News Reader

Comment by dboreham | Hacker News Reader

FPGAs have never made sense. They're way too expensive to use in volume. There's no practical use case for "cool, I can reprogram the chip in the field to implement different functionality". Nobody has figured out how to usefully integrate them with a CPU to make a low-volume SOC. CPUs became so fast that most applications don't need customer hardware. Regular gate arrays are cheaper and faster above minimal volume.

They seem to only have been useful for prototyping and military applications (low volume and infinite budget).

View on HN

I see them used in pro/prosumer audio equipment, synthesizers, and effects, which is relatively low volume and medium-to-high budget. FPGAs (and CPLDs, µC+AFE, etc) are great for these applications because they have great capabilities you might otherwise need a pile of discrete components or a custom chip for, but it doesn’t make sense to design fully custom silicon if you’re only ever going to sell about 50-500 of something.

So sure, prototyping and military, but there are other uses as well. But none of them are super high-volume because once you’re selling millions of something you should be designing your own chips.

Consumer application and FPGAs are an oxymoron in itself. FPGAs are used in applications requiring special interfaces, special computing units or other custom requirements. If there is enough demand, SoCs are developed for these applications, but this is only useful in mid to high volume production. Areas like the ones you gave and many more are making heavy use of FPGAs. I work in medical for example. We are using custom designed chips for special detection purposes. But when it comes to data processing and interfacing with computers, we use FPGAs.

> CPUs became so fast that most applications don't need customer hardware.

When complicated realtime signal processing is to be done, FPGAs shine - in particular if there exists no DSP that is competitive for the task.

This. I work at a research particle accelerator. All the hard real-time, super low-level signal processing and process control stuff is done on FPGAs, there is no other way around.

You somehow managed to write a post where every single sentence is absolutely wrong. FPGAs clearly make sense for prototyping ASICs. That alone makes FPGAs make sense even if it is a tiny niche market. After all, the budget of ASIC companies is big. A few hundred FPGAs for developers are a drop in the bucket compared to the cost of an ASIC.

Too expensive in volume only applies to Xilinx and Altera and with every node shrink, the amount of designs that fit on an FPGA grow while the non recurrent development costs grow for ASICs. Due to this, the maximum volume at which FPGAs remain cost competitive keeps growing with every generation.

Smart NICs make extensive use of reconfiguration because things such as protocols are not set in stone. They can change all the time. It is also possible to build designs that make extensive use of partial reconfiguration.

MPSoCs have been a thing for a long time. If you want to get into those Google Kria 260 and you will be pleasantly surprised.

When I take a look at Effinix FPGAs those are designed specifically for vision applications and massive amounts of I/O. A CPU would struggle with multiple camera streams and consume too much power.

Yeah ok but an 100k LUT FPGA Chip from Effinix costs what? 25€? You're going to need a really high volume to get an overall cost saving. 40000 FPGAs is only a million dollars. The masks of an 22nm ASIC cost 1.5 million dollars without the rest of the development costs.

And finally your last sentence contradicts the first. Next time stick with a story.

The problem is that FPGA companies are really CAD tool companies who see their chips as copy-protection/payment-assurance schemes for their software.

Unfortunately their CAD tools suck, but that's beside the point.

Reply on news.ycombinator.com