Hello golang experts,
I am curious how does goroutine scheduler picks what goroutine to run,
among several runnable. Does it optimize for fairness in any way?
I ran a quick experiment and found out that goroutines that run for longer
intervals between yield points receive proportionally larger CPU share.
In the following code:
package main
import (
"fmt"
"runtime"
"time"
)
func run(n int) int {
ret := 1
for i := 0; i < n; i++ {
ret *= i
}
return ret
}
func foo(ch chan float64, n, m int) {
f := 0
var run_time, wait_time, start_time, end_time int64
for i := 0; i < m; i++ {
start_time = time.Now().UnixNano()
if end_time != 0 {
wait_time += start_time - end_time;
}
f += run(n)
end_time = time.Now().UnixNano();
run_time += end_time - start_time;
runtime.Gosched()
}
ch <- float64(run_time) / float64(run_time + wait_time)
}
func main() {
ch := make(chan float64)
go foo(ch, 100000, 100)
go foo(ch, 1000000, 100)
v := <-ch
fmt.Printf("first: %f\n", v)
v = <-ch
fmt.Printf("second: %f\n", v)
}
Here, one goroutine gets 10x cpu time compared to the other.
$ GOMAXPROCS=1 ./ss
first: 0.911957
second: 0.091656
Does this test expose the scheduler's cpu policy correctly, or it is
biased? What is the best reading about scheduler's policies?
Thank you!
Dmitry.
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
