Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

1) FPGAs are not easy to develop for, and totally different from the skillset of a Photoshop developer. It is not as easy as dreaming up a design and pressing a button; if you are anywhere close to pushing the envelope it takes expertise & time.

2) GPUs, being purpose-built and mass-marketed, are both much cheaper and much faster. Think of the Ford Fiesta ST vs. a Jeep Wrangler. The Jeep is simple & more reconfigurable, yet is slower and more expensive- for exactly those reasons!

3) GPUs and CPUs compliment eachother well. The only gap is heavily parallel branching code- GPUs are bad at if-statements, CPUs are bad at heavily parallel. But a branch predictor is the key to if-statements, and branch predictors are hard.

FPGAs are fundamentally a prototyping tool. Can you think of examples of prototyping tools that eventually broke into the main market, replacing the incumbent product?



>FPGAs are fundamentally a prototyping tool. Can you think of examples of prototyping tools that eventually broke into the main market, replacing the incumbent product?

Python


> FPGAs are fundamentally a prototyping tool.

FPGAs can be used for prototyping ASICs, but that's by far not their major use-case.


What is then? I know they are used alongside PLDs, but I thought PLDs still dominated that use-case.


They are essentially low-volume ASIC replacements. Some applications:

- digital signal processing (DSP)

- network equipment (modern high-end FPGAs are able to function as 100G Ethernet switches and routers, just connect them to some PHYs and fast DRAM)

- systems on chip (SoC): processor, fast DSP, all kinds of IOs, memory controller – all on a single chip

- realtime video stream processing

For more examples, see http://www.xilinx.com/applications/index.htm




Consider applying for YC's Fall 2026 batch! Applications are open till July 27.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: