“Yes, but after you do, it’ll be blazing fast in the twenty minutes before you’ll have to recompile due to software updates.”
“So, you’re saying that it would have been really efficient if all my eight cores weren’t constantly pegged building software?”
“…Uh, yes.”
“And that optimizing software for the native architecture could have enabled power savings, if those machines weren’t all burning oceans of kilowatt hours building the same software over and over again redundantly?”
I wonder what CPU performance you would need so that it is compiling all the time given: average package release rate and average compile time of the most commonly installed ones.
“Yes, but after you do, it’ll be blazing fast in the twenty minutes before you’ll have to recompile due to software updates.”
“So, you’re saying that it would have been really efficient if all my eight cores weren’t constantly pegged building software?”
“…Uh, yes.”
“And that optimizing software for the native architecture could have enabled power savings, if those machines weren’t all burning oceans of kilowatt hours building the same software over and over again redundantly?”
“Ah…”
Blazing fast, as in, a full 1 to 3% faster than a generic build.
Some non-Gentoo distros ship CPU-optimized builds anyway, so even that is quite questionable.
I wonder what CPU performance you would need so that it is compiling all the time given: average package release rate and average compile time of the most commonly installed ones.
Well, it’d be highly contingent on what set of packages one has installed, but it’s an interesting question.