gc-cons-threshold startup time speedup

[CarlQLange]

Hey,

I guess this is more a system-specific thing, in that it probably shouldn't be set by default, but setting (setq gc-cons-threshold 100000000) as the first line in init.el took my startup time from 5.5s to 3.5, which is a pretty decent improvement. (Yes, I use emacs as a persistent server, emacsclient etc, but still, I like short startup times)

I guess just a thought? I'm sure everyone's seen the thread on reddit I got the idea from.

Cheers

[bmag]

There already is a similar line in https://github.com/syl20bnr/spacemacs/blob/develop/layers/%2Bdistribution/spacemacs-core/config.el#L229, but I guess it's executed a bit too late to help startup time? Moving that line to init.el didn't help much in my case (from ~3.7s to ~3.3s)

[CarlQLange]

Didn't help much? Maybe it's the old game dev in me coming out, but 400ms is a super long time! :)
In any case, I think you're right - that should be moved to very early in init.el.

I also experimented with compiling all the elisp in .emacs.d/, and while i did come out of that with a 200ms improvement I don't think it's worth the hassle.

[CarlQLange]

Also, man the Spacemacs architecture has changed since I last contributed in like March... +distributions? Are there gonna be others?

[TheBB]

See #2937 for that.

[syl20bnr]

I made the change, thank you :-)

[person808]

It may be good to restore it to be closer to the default after init finishes. See this comment

[syl20bnr]

Indeed, thank you ! PR welcome :-)

[synic]

I guess you all saw the same reddit thread I did. After I read it, I went searching for the proper value to set it to, and found this:

http://lists.gnu.org/archive/html/help-gnu-emacs/2007-09/msg00241.html

From the email:

There used to be some benefit to setting it higher for large Emacs sessions
(where GC can take a long time) since it would reduce the frequency of those
GCs. In Emacs-22, this is not needed any more because of the new
`gc-cons-percentage' variable which automatically causes the limit to grow
as the Emacs process grows, thus preserving a more or less constant
GC overhead.

It makes me wonder if manually setting gc-cons-threshold will end up disabling this other feature, which was created to solve this issue.

[syl20bnr]

Good question I think it does not, or it would be surprising it does. Setting this variable should set the initial amount but not prevent the % feature to happen.
It has to be verified by monitoring the amount of GCed memory. This is the kind of metric that would be helpful to plot with dotspacemacs-configuration-layers set to all.