Hm, 521ms versus 437ms - Overhead for the interpreter, I have to rework my modular exponentation again :(

Best regards and many thanks, Hagen
Posted on 2003-12-17 04:38:58 by Hagen
Hi BitRAKE,
there seem's big differences between Your 1.8 GHz P4 to my 1.5 GHz P4.

I have now downloaded a Mathematica Trial version and after some trial&errors I get your above test running.
On my P4 I get 1687 ms for the test.

With Random Integers in Range 2^4096 i get timings of 2186 Milliseconds.
Thus my Machine is far slower as 1.2 times as Your Machine ??!



I1 = Random[Integer, {2^4095, 2^4096 - 1}];
I2 = Random[Integer, {2^4095, 2^4096 - 1}];
I3 = Random[Integer, {2^4095, 2^4096 - 1}];
Timing[PowerMod[I1, I2, I3]]

= 2.186 Seconds.

The ratio between Mathematica to my Lib are now 2200/530 ~ 4 times faster, thats more probable :)

Maybe, could it be that I don't have Mathematica properly configured ??

Best Regards, Hagen
Posted on 2003-12-17 07:07:48 by Hagen
Hagen, my system is:
CPU: 2Ghz Athlon XP "Barton" (512K L2 cache)
Memory: 2GB DDR333

Posted on 2003-12-17 10:10:25 by bitRAKE
Ok, estimated then 2GHz/1.5GHz = 1.333 * 2 because on P4 i use SSE2 instead of native 486 asm = 2.666
2200ms / 2.666 = 830ms thus ~ 2 times difference remains ! Your DDR RAM seems to me realy realy fast, because I have no DDR installed.

It is possible to post here a small Demo EXE with my code, eg. Pi/Factorials and so on, and to try it on your Athlon ?

Best Regards, Hagen
Posted on 2003-12-17 16:56:04 by Hagen

It is possible to post here a small Demo EXE with my code, eg. Pi/Factorials and so on, and to try it on your Athlon ?
Yes, please do. SSE2 isn't supported, but everything else should work.

1 Mhz Athlon != 1 Mhz P4

In general I'd estimate:

2 Ghz Athlon == 2.6 Ghz P4

...and then DDR makes up the difference.
Posted on 2003-12-17 18:24:03 by bitRAKE