[futurebasic] Re: [FB] FB^3 wins Hands Down re Mandelbrot Test

Message: < previous - next > : Reply : Subscribe : Cleanse
Home   : September 2000 : Group Archive : Group : All Groups

From: Chris Stasny <staz@...>
Date: Wed, 20 Sep 2000 09:54:57 -0500
Wow! Thanks ted. I guess we have been a bit conservative in 
estimating our speed advantages over other basics. I've always said 
that we are at least 4 times faster than any other BASIC.

>Hi Gang:
>
>I wondered when someone was going to compare the RB Mandelbrot application
>with the FB^3 Mandelbrot application, but no one came forward (except
>Robert who wrote the FB^3 version).
>
>So, I decided to see which one was faster for myself. It was a little
>difficult because to compute and display the Mandelbrot graphic is the time
>consuming ordeal for these programs -- as such, the comparison should be
>done on both applications generating somewhat equal regions within the
>Mandelbrot function.
>
>With that said, the RB application typically generates the Mandelbrot
>graphic and then asks the user to check the upper and lower regions of the
>area the user wants to see and it then generates another Mandelbrot graphic
>to match the users coordinates. However, the FB^3 application just simply
>asks the user to click on a point and then it generates a two-fold
>magnification of the area thereby generating another Mandelbrot graphic.
>
>As such, to be fair about the comparison, both programs should be evaluated
>in terms of what was generated, what was displayed, and what time it took
>to preform the act. It should also be noted that the RB version reports
>times in seconds whereas the FB^3 version reports times in ticks (60ths of
>a second -- a slight hint as to which was faster).
>
>My methodology was simply to run the FB^3 version and have it generate a
>graphic; record the time; and then run the RB version selecting a region as
>close as I could to the region generated by the FB^3 version and then
>repeat the process four to six levels deep. I think I did a pretty fair job
>of selecting equal areas. The time comparisons follows (all times are in
>ticks).
>
>Test 1:
>
>	FB^ 3	RB
>	64	144 ticks
>	28	171
>	13	122
>	15	111
>	21	186
>	39	357
>
>	Average:
>	30	181  <--  FB^3 was 6 times faster
>
>Test 2:
>	FB^3	RB
>	64	144 ticks
>	54	358
>	28	300
>	19	180
>	15	163
>	18	228
>
>	Average:
>	33	228  <-- FB^3 was 6.9 times faster
>
>I hope that this test proves interesting to all involved. If anyone wants
>copies of the programs used, please email me and I'll place the entire
>experiment on my web site for downloading. It is my opinion that if one
>continued generating tests, such as these, one would find that the FB^3
>version is approximately 6+ times faster than the RB version for six levels
>deep in the Mandelbrot function. Deeper levels were not tested because it
>became obvious that FB^ was much faster than RB -- so what's the point?
>
>Many thanks to Robert (FB^3) and Geoff (RB) for their talents and efforts
>to make this test possible.
>
>tedd
>
>PS: For those interested, I am running a PPC 9500/132 with Mac OS 8.6
>
>
>
>
>
>
>
>http://sperling.com/
>_______________________________________________________
>Thinking about buying Jewelry?
>Try our site: http://earthstones.com
>
>
>
>--
>To unsubscribe, send ANY message to <futurebasic-unsubscribe@...>



Best,

-STAZ  ~)~

800.348.2623 Orders  http://www.stazsoftware.com
228.255.7086 FAX     mailto:staz@...