You may have read the discussion, but you missed the point.
: > Creates in TMPFS are going to be orders of magnitude faster than in UFS.
:
: But the number of files created during an Emacs compilation is rather low
: compared to the total time (remember my "benchmark" is to compile Emacs on a
: Sun once on a tmpfs partition and the other time with a uifs partition but in
: both cases the /tmp files used internally by gcc were on a tmpfs so are
: irrelevant for this discussion).
Great. So how about you try that same compile on a Linux box & ext2fs
and demonstrate how much slower it is than when you use tmpfs? You keep
harping on tmpfs vs ufs, which is /not/ the same as tmpfs vs ext2fs.
For all practical purposes, Linux ext2fs performance >= Solaris tmpfs
performance.
I'm sorry to keep beating on this, but the question was "shouldn't we
be thinking about tmpfs?" and the answer is "no, we shouldn't" and the
reason is "ext2fs is just as fast, if not faster" and the fact that "tmpfs
is much faster than ufs" has absolutely no bearing on the discussion.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu