[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] Write procedure is not backwards compatible

John Boyle scripsit:

> write/fast consistently takes 4500-4700 milliseconds to print all of these
> structures out, which makes sense, because it does the same amount of work
> on each cons cell, no matter how long the tail of the list is.  Over the
> range of structures with 40 cons cells to those with 40,000 cons cells,
> "write/safe" is generally 10% slower than "write/fast".  At 400,000 cons
> cells, "write/safe" takes 33% longer (6100 msec), and at 4 million, 65%
> longer (7600 msec).  This seems consistent with my estimate of "O(n) -> O(n
> log n), or perhaps O(n), depending on exactly how hash-eqs work".  And
> incidentally, at 4 cons cells, write/fast took 5400 and write/safe took
> 6400 msec (20% slower).

If you have the data handy, how much space did the hash-eq occupy,
as distinct from how many elements it had in it?

John Cowan        http://ccil.org/~cowan   cowan@x
Lope de Vega: "It wonders me I can speak at all.  Some caitiff rogue
did rudely yerk me on the knob, wherefrom my wits yet wander."
An Englishman: "Ay, belike a filchman to the nab'll leave you
crank for a spell." --Harry Turtledove, Ruled Britannia