aaron spink (aaronspink.delete@this.notearthlink.net) on August 15, 2019 5:16 pm wrote:
> Doug S (foo.delete@this.bar.bar) on August 13, 2019 11:15 pm wrote:
> > aaron spink (aaronspink.delete@this.notearthlink.net) on August 13, 2019 7:23 pm wrote:
> > > LOL. Yeah, most projects I work on have linking times that
> > > are on the order of 150x longer than your compile
> > > (and that's using the good linker, with other versions of the linker, linking alone can take an hour).
> >
> >
> > It can't actually take that long to LINK, can it? The only time I've seen such long "linking" times
> > is with a compiler doing cross module optimization, so much of the optimization is done at link time.
> >
>
> For an actual large scale project? No, that's normal. That particular project was an RTL level model
> of a full 8C SOC. You are looking at 10s of millions of lines of actual complex and dense source code
> that generates object file sets in the 100s of megabytes range. Honestly, 15 minutes to link is doing
> pretty good. It was honestly by far the best RTL dev environment I've ever seen or heard about with
> QOL, avg compile times, and runtime performance that put current environments to shame.
>
> And as I said, that was with a good rev of the linker. Every couple of months a different rev would come out
> that would take upwards of 4+x as long which would force us to revert back to a stable fast version.
>

Does that operate on C code or VHDL or what? Just curious if really fast c compiler/linker with good optimizations would solve that problem.