OpenEuphoria

Re: Testing Sequences for Uniqueness

• Posted by "Hawke'" <mikedeland at NETZERO.NET> Aug 24, 2000
• 435 views

methinks, mayhaps, thine benchmark hath flaw :(((((

you are entering the realm of cached data since you
are not comparing different sequences each run...

> x=compare ({1,2,3,4,5,6},{1,2,3,4,5,6})
is the problem child line....

so my suggestion is to create a large sequence of
pairs of 6 element sequences and u can step thru
them while timing the equality of those pairs

-----------tested code
sequence data,result
atom now
object temp
constant tries = 250000
constant dataset = {6,6,6,6,6,6}

--gotta store the result or EU will possibly
--throw away the equality test in an effort
--to optimize dead code
puts(1,"Creating empty dataset\n")
data   = repeat({},tries)
result = repeat({},tries)

puts(1,"Filling dataset with randomness\n")
for i = 1 to tries do
   data[i] = {rand(dataset),rand(dataset)}
end for

puts(1,"Beginning benchmark\n")
--we use the temp variable as it is
--less burden on the benchmark
--as we dont slice as much this way
now = time()
for i = 1 to tries do
   temp   = data[i]
   result[i] = equal(temp[1],temp[2])
end for
now = time() - now
printf(1,"%d tries in %f seconds",{tries,now})

------------end code

on a pIII733 with 128mb 133mhz ram
250000 tries in 0.270000 seconds

on a p200mmx with 48mb EDO ram
75000 tries in 0.220000 seconds

but it took several seconds to create the dataset
for each run on both machines....

i'm thinking equal() & compare() are fast nuff ;)
and who said EU was slow?
--Hawke'


_____NetZero Free Internet Access and Email______
   http://www.netzero.net/download/index.html