Re: Scheme style auto-resizing hashtable

[Jay Glascoe <jglascoe@jay.giss.nasa.gov>]

On 26 Oct 1998, Harvey J. Stein wrote:

> STk's hash tables are faster than Guile's (on the order of 20% faster
> when computing runtime-overhead, and on the order of 50% when doing
> runtime-gc-overhead, assuming (gc-run-time) returns time in
> milliseconds). 
> 

my hash tables are a good deal faster than Guile's:

the heart of the test (run ten times):

            ;; <hash table is created>
	    (for-each (insert-entry mytab insert-proc) lines)
	    (for-each (find-entry mytab find-proc) lines)
	    (for-each (insert-entry mytab insert-proc) lines)
	    (for-each (find-entry mytab find-proc) lines)

...

for x in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20; do
    head -n "${x}0000" rand-nos.bak | tail -n 10000 > rand-nos

    echo "hashtab"
    ./myguile -s httest.scm
    echo

    echo "Guile hash"
    ./myguile -s gtest.scm 8219
    echo
done

hashtab
*time: 539*

Guile hash
*time: 810*

hashtab
*time: 517*

Guile hash
*time: 808*

hashtab
*time: 549*

Guile hash
*time: 809*

hashtab
*time: 554*

Guile hash
*time: 808*

hashtab
*time: 537*

Guile hash
*time: 810*

hashtab
*time: 535*

Guile hash
*time: 809*

etc.