Skip to main content
Unix & Linux

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

Required fields*

5
  • I was surprised at the mmap result TBH. I used to think mmaping was faster than read/write, but then I saw some linux benchmarks that showed the opposite. Looks like it is very true in this case. Commented Mar 2, 2016 at 0:27
  • 4
    mmap is going to get vastly better results on linux because it'll map to huge pages these days, and TLB misses are sloooowwwwwww. Commented Mar 2, 2016 at 1:12
  • There may be some benefit to reading different parts of the file in separate threads (e.g. with an OpenMP for loop) so that some progress might be made while one thread is stalled waiting for input. But on the other hand, it might hamper I/O scheduling, so all I can recommend is to try it, and measure! Commented Mar 2, 2016 at 12:52
  • The read() version may benefit from read-ahead. Commented Mar 2, 2016 at 17:34
  • 1
    @TobySpeight Yeah, multithreading might speed it up. Also looking scanning two bytes at a time via a 2^16 look up tables provided a pretty good speed up last time I played with it. Commented Nov 22, 2019 at 14:54