diff options
author | Christian Grothoff <christian@grothoff.org> | 2016-05-02 04:13:06 +0000 |
---|---|---|
committer | Christian Grothoff <christian@grothoff.org> | 2016-05-02 04:13:06 +0000 |
commit | 84ec6a79caafa209fef4ac8e1b87769863a79346 (patch) | |
tree | 525383b1cf0b7c9a227479d412fc16a773ebbb69 /doc | |
parent | cc803ec8c257bea820e20927b38b5a5ae1fce7c6 (diff) | |
download | libmicrohttpd-84ec6a79caafa209fef4ac8e1b87769863a79346.tar.gz libmicrohttpd-84ec6a79caafa209fef4ac8e1b87769863a79346.zip |
adding logic to do 'fast' calls to read/write on a single connection without going back to the main event loop to help FE with his performance issue
Diffstat (limited to 'doc')
-rw-r--r-- | doc/libmicrohttpd.texi | 19 |
1 files changed, 19 insertions, 0 deletions
diff --git a/doc/libmicrohttpd.texi b/doc/libmicrohttpd.texi index 497595cc..408f471d 100644 --- a/doc/libmicrohttpd.texi +++ b/doc/libmicrohttpd.texi | |||
@@ -520,6 +520,25 @@ call has fundamentally lower complexity (O(1) for @code{epoll()} | |||
520 | vs. O(n) for @code{select()}/@code{poll()} where n is the number of | 520 | vs. O(n) for @code{select()}/@code{poll()} where n is the number of |
521 | open connections). | 521 | open connections). |
522 | 522 | ||
523 | @item MHD_USE_EPOLL_TURBO | ||
524 | @cindex performance | ||
525 | Enable optimizations to aggressively improve performance. Note that | ||
526 | the option is a slight misnomer, as these days it also enables optimziations | ||
527 | that are unrelated to @code{MHD_USE_EPOLL_LINUX_ONLY}. Hence it is OK to | ||
528 | use this option with other event loops. | ||
529 | |||
530 | Currently, the optimizations this option enables are based on | ||
531 | opportunistic reads and writes. Bascially, MHD will simply try to | ||
532 | read or write or accept on a socket before checking that the socket is | ||
533 | ready for IO using the event loop mechanism. As the sockets are | ||
534 | non-blocking, this may fail (at a loss of performance), but generally | ||
535 | MHD does this in situations where the operation is likely to succeed, | ||
536 | in which case performance is improved. Setting the flag should generally | ||
537 | be safe (even though the code is slightly more experimental). You may | ||
538 | want to benchmark your application to see if this makes any difference | ||
539 | for you. | ||
540 | |||
541 | |||
523 | @item MHD_SUPPRESS_DATE_NO_CLOCK | 542 | @item MHD_SUPPRESS_DATE_NO_CLOCK |
524 | @cindex date | 543 | @cindex date |
525 | @cindex clock | 544 | @cindex clock |