bug#37321: 27.0.50; Excessive gc in a use case (el-search)

[Michael Heerdegen]

Eli Zaretskii <eliz@gnu.org> writes:

> Start with 4 times the default.  If that's not enough, double the
> value until you are satisfied, and tell what was the final value.

I leave it to you to interpret the result (can we expect that other use
cases have a similar optimum?):



My current setup (setq gc-cons-percentage .8)

52.331413s (7.141519s in 20 GCs)


Default: (setq gc-cons-threshold 800000)

87.626289s (41.526705s in 269 GCs)


(setq gc-cons-threshold (* 800000 2))

86.213154s (40.132441s in 267 GCs)


(setq gc-cons-threshold (* 800000 4))

87.422085s (40.967298s in 267 GCs)


(setq gc-cons-threshold (* 800000 8))

84.519948s (38.535352s in 245 GCs)


(setq gc-cons-threshold (* 800000 16))

81.596529s (34.603730s in 193 GCs)


(setq gc-cons-threshold (* 800000 32))

70.866717s (24.850518s in 123 GCs)


(setq gc-cons-threshold (* 800000 64))

61.893515s (14.088496s in 63 GCs)


(setq gc-cons-threshold (* 800000 128))

56.277259s (8.122554s in 32 GCs)


(setq gc-cons-threshold (* 800000 256))

55.173263s (6.067774s in 17 GCs)


(setq gc-cons-threshold (* 800000 512))

56.116736s (8.825274s in 9 GCs)


(setq gc-cons-threshold (* 800000 1024))

70.832969s (19.474471s in 5 GCs)



Regards,

Michael.