Mad Pascal

zbyti · February 29, 2020


program MonteCarloPi;

uses crt;

{$define FAST}
{$f $70}

var
    rtClock1              : byte absolute 19;
    rtClock2              : byte absolute 20;
    rndNumber             : byte absolute $D20A;

{$ifdef FAST}
    stop : word absolute $e0;
    i                     : word absolute $e0;
    r                     : word absolute $e2;
    x                     : word absolute $e4;
    y                     : word absolute $e6;
    bingo                 : word absolute $e8;
    probe                 : word absolute $ea;
    foundPi               : word absolute $ec;
    n                     : byte absolute $ee;
{$else}
    stop, i, r, x, y      : word;
    bingo, probe, foundPi : word;
    n                     : byte;
{$endif}

begin
    bingo := 0;
    r := 127 * 127;
    probe := 10000;

    Pause;    
    rtClock1 := 0; rtClock2 := 0;
    
    for i := 0 to probe do begin
        n := (rndNumber or %10000000) xor %10000000;
        x := n * n;
        n := (rndNumber or %10000000) xor %10000000;        
        y := n * n;
        if (x + y) <= r then Inc(bingo);
    end;
    
    foundPi := 4 * bingo;
    stop := (rtClock1 * 256) + rtClock2;
    
    {$ifdef FAST}
    WriteLn('Variables on zero page');
    {$endif}
    WriteLn('Probe size ', probe);
    WriteLn('Points in circle ', bingo);
    Write('Found pi approximation ', foundPi);
    WriteLn(chr(30),chr(30),chr(30),chr(30),chr(255),chr(44));
    WriteLn('Frames counter = ', stop);
    ReadKey;
end.


PROBE=10000:B=0:R=127*127:F=PEEK(20)
WHILE F=PEEK(20)
WEND
POKE 20,0:POKE 19,0
FOR I=0 TO PROBE
 N=PEEK($D20A):N=N!128:N=N EXOR 128
 X=N*N
 N=PEEK($D20A):N=N!128:N=N EXOR 128
 Y=N*N
 IF (X+Y)<=R THEN INC B
NEXT I
P=4*B
T=TIME
? P;CHR$(30);CHR$(30);CHR$(30);CHR$(30);CHR$(255);CHR$(46)
? T;" FRAMES"

YoshPlus&Pi - Benchmarks.ods

monte-carlo-pi-sources.zip

Edited March 1, 2020 by zbyti
sources

zbyti · March 2, 2020

#include <stdio.h>

void main(void)
{
  register unsigned int stop, i, x, y;
  register unsigned int b, p, r;
  register unsigned char n;
  register unsigned char* rndp = 0xd20a;
  register unsigned char* tick = 0x14;
  register unsigned char* tack = 0x13;

  b = 0;
  r = 127 * 127;
  p = 10000;

  n = *tick;
  while (*tick == n) { ; }

  printf("\nMonte-Carlo PI cc65 V2.18\n");

  asm(" lda #0");
  asm(" sta $13");
  asm(" sta $14");

  for (i = 0 ; i < p; ++i)
  {
    n = (*rndp | 128) ^ 128;
    x = n * n;
    n = (*rndp | 128) ^ 128;
    y = n * n;
    if ((x + y) <= r)
      ++b;
  }
  b = 4 * b;
  stop = *tick + (*tack * 256);

  printf("%d%c%c%c%c%c%c\n", b, 30, 30, 30, 30, 255, 46);
  printf("%d ticks\n", stop);

  infinite:
    goto infinite;
}

cl65 -t atari -o mcpi.xex mcpi.c           602 ticks
cl65 -t atari -Osir -Cl -o mcpi.xex mcpi.c 303 ticks

YoshPlus&Pi - Benchmarks.ods

Edited March 2, 2020 by zbyti

zbyti · March 2, 2020

Mad Pascal 1.6.4 updated to 01.03.2020 master.

Edited March 2, 2020 by zbyti

zbyti · March 2, 2020

#include <stdio.h>
#include <peekpoke.h>

#define p 10000
#define r 16129
#define tick		(PEEK(0x14))
#define tack		(PEEK(0x13) * 256)
#define rndp 		(PEEK(0xd20a))

void main(void)
{
  register unsigned int stop, i, x, y;
  register unsigned int b = 0;
  register unsigned char n;

  n = tick;
  while (tick == n) { ; }

  printf("\nMonte-Carlo PI cc65 V2.18\n");

  asm(" lda #0");
  asm(" sta $13");
  asm(" sta $14");

  for (i = 0 ; i < p; ++i)
  {
    n = (rndp | 128) ^ 128;
    x = n * n;
    n = (rndp | 128) ^ 128;
    y = n * n;
    if ((x + y) <= r)
      ++b;
  }
  b *= 4;
  stop = tick + tack;

  printf("%d%c%c%c%c%c%c\n", b, 30, 30, 30, 30, 255, 46);
  printf("%d ticks\n", stop);

  infinite:
    goto infinite;
}

Edited March 2, 2020 by zbyti
add macros

Wrathchild · March 2, 2020

n = (rndp | 128) ^ 128

This is producing:

   lda $D20A
   ora #$80
   eor #$80

is that not the same as:

n = rndp & 127

either of the optimizer flags doesn't detect and replace this?

drac030 · March 2, 2020

Also, does not CC65 support such use of pointers as:

n = *(unsigned char *)0xd20a & 0x7f;

or (instead of the asm snippet):

*(int *)0x13 = 0;

Also, this:

 infinite:
    goto infinite;

would be better said as something like that:

 for(;;)
   ;

(although it in no way affects performance, of course).

zbyti · March 2, 2020

@Wrathchild no, it's not the same.

@drac030 thx :]


#include <stdio.h>

#define p 10000
#define r 16129

#define RTCLOK1 (*(volatile unsigned char*)0x13)
#define RTCLOK2 (*(volatile unsigned char*)0x14)
#define RANDOM  (*(volatile unsigned char*)0xd20a)

void main(void)
{
  register unsigned int stop, i, x, y;
  register unsigned int b = 0;
  register unsigned char n;

  n = RTCLOK2;
  while (RTCLOK2 == n) { ; }

  printf("\nMonte-Carlo PI cc65 V2.18\n");

  RTCLOK1 = 0; RTCLOK2 = 0;
  for (i = 0 ; i < p; ++i)
  {
    n = (RANDOM | 128) ^ 128;
    x = n * n;
    n = (RANDOM | 128) ^ 128;
    y = n * n;
    if ((x + y) <= r)
      ++b;
  }
  b *= 4;
  stop = RTCLOK2 + 256 * RTCLOK1;

  printf("%u\x1e\x1e\x1e\x1e\xff.\n%u ticks\n", b, stop);

  for (;;);
}

bocianu · March 3, 2020

I'm pretty sure that:

n = (rndp | 128) ^ 128

is equivalent of:

n = rndp & 127

and it's quicker.

Try and compare results

drac030 · March 3, 2020

Yes, I see no difference either: n is unsigned char, so random | 128 ^ 128 = LDA RANDOM / ORA #$80 / EOR #$80 which effectively, when bit 7 is set, sets it again, and inverts (clears) it later; and when it is clear - first sets it and inverts (clears) later. So it is AND #$7F = &0x7F (as said already above).

Edited March 3, 2020 by drac030
typos, many typos

dmsc · March 3, 2020

Hi!

3 hours ago, zbyti said:

@Wrathchild no, it's not the same.

@drac030 thx :]

A little more "idiomatic" C :

#include <stdio.h>
#include <atari.h>

#pragma static-locals(on)

#define p 10000
#define r 16129

void main(void)
{
  register unsigned int x, y;
  unsigned int stop, i, b = 0;
  register unsigned char n;

  n = OS.rtclok[2];
  while (OS.rtclok[2] == n) { ; }

  printf("\nMonte-Carlo PI cc65 V2.18\n");

  OS.rtclok[1] = 0; OS.rtclok[2] = 0;
  for (i = 0 ; i < p; ++i)
  {
    n = (POKEY_READ.random | 128) ^ 128;
    x = n * n;
    n = (POKEY_READ.random | 128) ^ 128;
    y = n * n;
    if ((x + y) <= r)
      ++b;
  }
  b *= 4;
  stop = OS.rtclok[2] + 256 * OS.rtclok[1];

  printf("%u.%04u\n%u ticks\n", b/10000,b%10000, stop);

  for (;;);
}

Also, writing (n = POKEY_READ.random & 0x7F) would be a little faster.

Have Fun!

zbyti · March 3, 2020

@dmsc very nice! thx :] I wrote Monte-Carlo PI as benchmark this way to make it easier for me to transfer code to other programming languages.

Edited March 3, 2020 by zbyti

zbyti · March 3, 2020

@Wrathchild & @bocianu & @drac030 shame on me of course you are right.

Edited March 3, 2020 by zbyti
rnd & 127

Wrathchild · March 3, 2020

@zbyti no worries

POKEY's random number generator should in no way be seen as a good candidate for use in a Monte Carlo simulation

zbyti · March 3, 2020

47 minutes ago, Wrathchild said:

@zbyti no worries

POKEY's random number generator should in no way be seen as a good candidate for use in a Monte Carlo simulation

It's true, but SID it's no better

//cl65 -t c64 -Osir -Cl -o mcpi.prg mcpi-c64.c
#include <stdio.h>

#define p 10000
#define r 16129
#define RTCLOK1 (*(volatile unsigned char*)0xa1)
#define RTCLOK2 (*(volatile unsigned char*)0xa2)
#define RANDOM  (*(volatile unsigned char*)0xd41b)

void main(void)
{
  register unsigned int stop, i, x, y;
  register unsigned int b = 0;
  register unsigned char n;

//SID's Random Number Generator
  asm(" lda #$ff");     //maximum frequency value
  asm(" sta $D40E");    //voice 3 frequency low byte
  asm(" sta $D40F");    //voice 3 frequency high byte
  asm(" lda #$80");     //noise waveform, gate bit off
  asm(" sta $D412");    //voice 3 control register

  n = RTCLOK2;
  while (RTCLOK2 == n) { ; }

  printf("\nMonte-Carlo PI cc65 V2.18\n");

  RTCLOK1 = 0; RTCLOK2 = 0;
  for (i = 0 ; i < p; ++i)
  {
    n = RANDOM & 127;
    x = n * n;
    n = RANDOM & 127;
    y = n * n;
    if ((x + y) <= r)
      ++b;
  }
  b *= 4;
  stop = RTCLOK2 + 256 * RTCLOK1;

  printf("%u.%04u\n%u ticks\n", b/p,b%p, stop);

  for (;;);
}

Edited March 3, 2020 by zbyti

tebe · March 7, 2020

MP Instruction in English by Uwo Oertel, thx

http://mads.atari8.info/doc/madpascal_en.html

stepho · March 12, 2020

On 3/3/2020 at 8:53 AM, bocianu said:

n = (rndp | 128) ^ 128

When I want to turn bit 7 off I usually write it as:

n = rndp & ~(1<<7)

Turning it on is like:

n = rndp | (1<<7)

And testing it is like:

if( rndp & (1<<7) != 0 )

the != 0 is not strictly necessary.

Wrathchild · March 12, 2020

12 hours ago, stepho said:

n = rndp | (1<<7)

Thankfully compilers are intelligent enough to pre-calc the constant math, even without the optimiser enabled

    lda     $D20A
    ora     #$80

jvas · March 22, 2020

Hi,

Can somebody tell me what is missing? I'm trying to set pixels, but nothing happens on screen (just a black/blank one).

Thanks in advance!

Cheers,

Jozsi

test1.pas

tebe · March 27, 2020

current color palette is clear

 SetRGBPalette(0);                  // set index to 0

 SetRGBPalette(0,0,0);            // palette[0] = rgb(0,0,0)
 SetRGBPalette(100,100,100); // palette[1] = rgb(100,100,100)

test_vbxe.zip

+Stephen · March 27, 2020

42 minutes ago, tebe said:
current color palette is clear
 SetRGBPalette(0);                  // set index to 0

 SetRGBPalette(0,0,0);            // palette[0] = rgb(0,0,0)
 SetRGBPalette(100,100,100); // palette[1] = rgb(100,100,100)
test_vbxe.zip 2.01 kB · 0 downloads

Is there a way to determine which of the 4 palettes is being used?

tebe · March 27, 2020

default = 1

Tickled_Pink · April 10, 2020

On 2/25/2020 at 10:17 AM, zbyti said:

Here are my tests on atari800 PAL :] Sources mostly included

Programming languages in the latest official version for February 20, 2020.

A8-Benchmarks.zip 59.57 kB · 5 downloads

For me, the most interesting thing about this was just how slow PL65 was. It was my language of choice. Still have my original disk here somewhere. I suspect that the problem was that it constantly jumped to its runtime library which, IIRC, was quite large and clearly not very well optimised.

+DjayBee · April 10, 2020

2 hours ago, Tickled_Pink said:

Still have my original disk here somewhere.

If you find it and know someone with the right equipment, then please have it dumped.

PL65 is not yet available as a preservation dump. I know of only one .PRO-file here on AtariAge at all.

Tickled_Pink · April 10, 2020

2 hours ago, DjayBee said:

If you find it and know someone with the right equipment, then please have it dumped.

PL65 is not yet available as a preservation dump. I know of only one .PRO-file here on AtariAge at all.

Yeah. That was probably the one I originally dumped when I had an APE interface. I'm not sure how to dump it properly into a standard XEX file.

tebe · April 13, 2020

https://github.com/tebe6502/Mad-Pascal


chessboard 70 ticks

romsum 73-79 ticks

sieve 607 ticks

mp_bench.7z

Mad Pascal

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Join the conversation

Recently Browsing 0 members