One way to speed up IO-bound code is to have multiple threads-of-execution (threads or async-io) in order to have multiple requests in the air. Specifically SSDs benefit a lot from multiple requests in parallel as that utilizes their internal parallelism. HDDs would benefit only a little bit but RAIDed HDDs can also benefit from parallelism.
I think his point is that you can match `cat`s speed because it is IO bound and you can make a linker IO bound by using lots of threads.
One way to speed up IO-bound code is to have multiple threads-of-execution (threads or async-io) in order to have multiple requests in the air. Specifically SSDs benefit a lot from multiple requests in parallel as that utilizes their internal parallelism. HDDs would benefit only a little bit but RAIDed HDDs can also benefit from parallelism.
It used to be lore in the industry that any decent linker is IO bound, including the single-threaded ones.