https://bz.apache.org/bugzilla/show_bug.cgi?id=63916
--- Comment #34 from Michael Osipov <micha...@apache.org> --- (In reply to Christopher Schultz from comment #33) > (In reply to Rainer Jung from comment #23) > > Plus: performance behavior plus provider string would also be interesting > > for MacOS. > > Some info from MacOS: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [146988] > Writing 10MB took [35] milliseconds > > With 16k buffer: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller 16384 > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [16384] > Writing 10MB took [42] milliseconds > > With 8k buffer: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller 8192 > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [8192] > Writing 10MB took [63] milliseconds > > With 4k buffer: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller 4096 > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [4096] > Writing 10MB took [91] milliseconds > > With 1k buffer: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller 1024 > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [1024] > Writing 10MB took [248] milliseconds > > With 32 byte buffer: > > $ java -showversion org/apache/tomcat/Bug63916NioPoller 32 > openjdk version "13.0.1" 2019-10-15 > OpenJDK Runtime Environment (build 13.0.1+9) > OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing) > Selector Provider: sun.nio.ch.KQueueSelectorProvider@1ac1a5f6 > Default send buffer size is [146988] > Using send buffer size [32] > Writing 10MB took [5279] milliseconds > > More data (buffer size, duration): > 16,10383 > 32,5206 > 64,2695 > 128,1413 > 256,744 > 512,435 > 1024,238 > 2048,149 > 4096,85 > 8192,61 > 16384,40 > 146988,35 > > Graphed: https://jsfiddle.net/etp8sa5f/ > > Interestingly, it's almost a perfectly straight trend line when graphed on a > logarithmic scale. I think that means there aren't any inconsistencies on > this platform: buffer size is hugely important but there aren't any > particularly bad buffer sizes (other than "smaller is worse"). Awesome, can you retry with Java 8? But the results are quite surprising because I have expected similar results as with FreeBSD due to the BSD userland and kernel parts in macOS. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
