Re: [Gnu-arch-users] [PATCH] arch speedups on big trees

[Chris Mason]

On Fri, 2003-12-19 at 15:56, Tom Lord wrote:
>     > From: Andrew Suffield <address@hidden>
> 
> Thanks and "here, try this":
> 
>     > On a smallish tree with 740 files, here's the top of strace -c when I
>     > apply about 20 changesets in one run:
> 
Please define 'apply'.  I'm guessing tla update?  The tla replay times
are so much lower because all the inode signature stuff is skipped.

> Combined with the resulting kernel-cache efficiencies -- your
> intuition that (another) order of magnitude savings is quite
> achievable seems right to me too.
> 
Agreed, the problem is largely solvable without major surgery.  We
should be able to make commit fast when the file list is specified, the
code just needs a little fine tuning for it.

> How many orders of magnitude will we have come from the early days of
> larch?  Somewhere in the 3-5 range, I think, though I haven't kept
> careful track.  :-)
> 
> 
>     > and even on a smallish tree they're taking up a fair amount of
>     > time. On a really big tree, like linux, they're going to
>     > dominate. For long series of changeset application, it's going
>     > to be *horrible*.
> 
> That's never been in dispute and even _after_ these improvements will
> _still_ be an issues -- which is why so much of what gets optimized is
> how to eliminate "long series of changeset application" for work-a-day
> developers.

4 seconds isn't bad ;-)  We should be able to get pretty closet to that
for the majority of cases.

-chris