Spécial ! |
Version mobile de l'aldweb Site m.aldweb.com Ni Français, ni Anglais ?Essayez donc l'un de ces drapeaux :
|
|
Météo / Spam / www |
Météo Lyon
aldweb contre le SPAM
|
|
|
|
|
|
|
- Forum - Pocket Computer
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 257
Sujet n° 1512 |
Speed Comparison |
le 18/02/2007 @ 23:17 par Xerxes
|
Hello,
thanks to the collaboration of some other forums there is a list with a lot of calculators available now. If you miss your calculator in the table, more results are welcome.
http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/articles.cgi?read=700 |
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 1 -------- le 19/02/2007 @ 20:36 par aldweb
visiteur |
Hello Xerxes,
Interesting benchmark database, thank you to bring it to our knowledge here.
Direct clickable link for visitors: http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/articles.cgi?read=700
It is very interesting to see, for the PC-1360 which is my favourite device ever that the benchmark runs in 477 seconds in Basic and in 0.796 second when performed with the assembler code, that is 600 times faster in assembler!
Another interesting benchmark page, with less models benchmarked but with more tests performed is the one from Daniel here: http://collection.sasfepu.free.fr/collection/index.html (in French, click on the "Une page de TEST pour comparer la vitesse des différents Ordinateurs de Poche" link in the main right frame)
Cheers, aldweb
|
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 2 -------- le 19/02/2007 @ 23:26 par Xerxes
visiteur |
The PC-1360 is my favorite Sharp pocket too. My favorite Casio is the PB-2000C with the Pascal-Compiler-ROM-Card installed.
One reason of the high speed in Assembler is that I used the CPU-RAM for all variables, so no access to the external RAM was neccessary. But in comparison to the other CPUs the SC61860 is not very fast due to its instruction set. This is also the reason for the slow Basic interpreter.
The sense of this benchmark is to compare the often used basic elements of the programming languages only and not special duties like an empty loop that is not really convincing for a real program execution.
What is missing is a float test. I am working on it, but it is not so easy to find a accurate algorithm.
Before I started making this test, I looked at the other benchmarks available and found this interesting site too. My problem was, that there was no common test with many of calculators to have a better view.
|
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 3 -------- le 20/02/2007 @ 21:50 par aldweb
visiteur |
Hello,
You could also give a look to my own Bench2 experiment in this area, for Palm devices.
You could also work with some pseudo random numbers generator, starting with the same seed you are sure to reach the same result after the same number of iterations, which is very convenient to check an implementation of the algorithm.
Cheers, aldweb
|
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 4 -------- le 21/02/2007 @ 13:32 par Xerxes
visiteur |
Your Bench2 test gives a good overview to choose a programming language for a palm. It's interesting to see the differences. Good work. |
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 5 -------- le 13/03/2007 @ 15:34 par Xerxes
visiteur |
I have updated the list with a result for PC-1360 fans:
PC-1360 turbo: 235 sec
I have replaced the original resonator by a 1.6 MHz resonator. This is the maximum speed possible without problems.
I have also tested the PC-1360 with ROM version 2 (PEEK &7FFF=3). It's a bit slower with 486 sec vs 477 sec at ROM version 1 (PEEK &7FFF=2). Due to the bugs of version 1, its more accurate to use version 2 for timing.
|
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 6 -------- le 04/04/2007 @ 00:19 par aldweb
visiteur |
Thank you Xerxes for keeping us updated with your work.
With no big surprise, the assembler version for your test runs twice faster when the 1.6 MHz resonator versus the original 768 KHz processor frequency. The performance is strictly proportinal to the CPU beat.
Something I did not understand in your table of results is the "[turbo x.y]" wording. What do you mean by "turbo"? And why different versions of it?
Cheers, aldweb
|
|
|
Warning: A non-numeric value encountered in /web5/aldweb/www/aldweb_com/www/thread.php on line 497
Réponse n° 7 -------- le 04/04/2007 @ 12:54 par Xerxes
visiteur |
The table will be updated constantly till there is no more calculator to test. ;)
If the calculator is speeded up by software, it commented with "Fast Mode" like the FX-9860GSD or TI-59.
[turbo x.y] means that the calculator is speeded up by hardware modification, like changing resonator e.g. PC-1360, resistor e.g. FX-502P or capacitor e.g. HP-15C.
x.y represents the speeding up factor. The reason of different factors is that the maximum possible overclocking rate is very different, usually about factor 2 but on some calculators up to factor 3.5!
|
|
|
sujet actif
sujet clos
Important!
Nouveau message -
Rectifier message
Clôturer sujet
Remonter
|
|
|
|
|
|