cgc wrote on Sun, 02 January 2011 00:13 |
Hehe, the article is about FPGAs which have been around for many years. Lots of audio and video devices use them already. They would have to be programmed to mimic an x86 chip in order to run a modern OS and software. At that point it would be doubtful that they would run any faster than current chips.
|
There's not really any doubt there, they would be slower, much slower.
FPGAs aren't particularly fast, they're always much slower than the current crop of full custom silicon, it's just that you can configure them to be particularly specialized, so they do a specific job very well (a custom chip with the same architecture would still be faster, but you may well do the job faster than a generic processor running a program to perform the same task).
So you don't use them to run a standard OS and software, you use them as accelerators for specific tasks.
I wonder if there's actually anything new in the project that's being reported on in this article except perhaps that the latest models of FPGA allow the implementation of certain architectures that in the past would have required a full custom implementation.