On Thu, Aug 15, 2019, 05:19 sandao <yirenwe...@gmail.com> wrote: > In my opinion, the slice which "as" point to, is all allocated on the > heap in two case . so, I think the mark time should not be so different. > Can you tell me more details for you answer(when as is allocated on the > stack, gc has a smaller task.)? > > 在 2019年8月15日星期四 UTC+8上午11:02:46,Gergely Födémesi写道: >> >> .On Wed, Aug 14, 2019, 17:56 sandao <yiren...@gmail.com> wrote: >> >>> type A struct { >>> a int >>> } >>> >>> func gen() []*A { >>> r := make([]*A, 1000000) >>> for i := 0; i < len(r); i++ { >>> r[i] = &A{i} >>> } >>> return r >>> } >>> >>> //var as []*A = gen() >>> >>> func main() { >>> go func() { >>> err := http.ListenAndServe(":8034", nil) >>> if err != nil { >>> //fmt.Printf("ListenAndServe:%s\n", err) >>> } >>> }() >>> var as []*A = gen() >>> fmt.Println(as[0]) >>> time.Sleep(100 * time.Second) >>> } >>> >>> when debug gc: >>> 1) if the "var as []*A = gen()" is in main , the debug info is : gc >>> 4 @2.757s 0%: 0.049+0.59+0.025 ms clock, 2.3+0/0.54/0.28+1.2 ms cpu, >>> 16->16->0 MB, 31 MB goal, 48 P (forced) >>> 2) if the "var as []*A = gen()" is not in main, the debug info >>> is: gc 9 @7.172s 0%: 0.077+25+0.020 ms clock, 3.7+0/24/0.73+0.99 ms >>> cpu, 16->16->15 MB, 31 MB goal, 48 P (forced) >>> >>> Anybody konw why the mark time (red numbers) is so different? >>> >> >> when as is allocated on the stack, gc has a smaller task. >> > see for yourself what the compiler says about that decision: https://golang.org/cmd/compile/ you'll need -m
you can also use pprof to check your assumptions. (there isn't any computer near me currently) -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CA%2BctqrpTonR2GO8woPdCkr2Ta7BA262h4uSrqkUYvBFREGqQPw%40mail.gmail.com.