For context, see slides 5 and 6 about the history of Hyperscan and Sensory Networks Inc.: https://openisf.files.wordpress.com/2015/11/oisf-keynote-201...
For context, see slides 5 and 6 about the history of Hyperscan and Sensory Networks Inc.: https://openisf.files.wordpress.com/2015/11/oisf-keynote-201...
Here is a real history https://twitter.com/joeerl/status/1115990630793207808
I think FPGA could do a great job of achieving worst-case performance by simulating NFAs in a very straightforward fashion (beating s/w). However, no-one does this as the headline numbers are terrible.
The thread you posted contains an assertion by someone who built an FPGA parser with no performance numbers to back it up; the performance comparison may well reflect relative skills with software and hardware as opposed to the limits of software. Interestingly, one of the first responses refers to another project of mine. :-) (simdjson). Unfortunately, given the subsequent posts on the thread, it does not look like the author can provide further details on the system, which is sad.
It often can, because it can clock 50x higher.
And if you want a fair comparison, customize both to the job.