Text Mandelbrot is available for the TI-99/4 now

[Tursi]

7 hours ago, vol said:

It is interesting that the standard TI Basic (+Editor/Assembler cartridge) is noticeably faster than Extended Basic.  I have gotten the next results

 

     d7,1   691     TI-99/4A + Editor/Assembler cartridge (Classic99 emu)
     d7,1   774     TI-99/4A + Extended Basic cartridge (Classic99 emu)

 

I  tried to use 't40xb' utility but it doesn't also allow to use 2 edge columns for PRINT.  

How did you enter lines 164 and 166? Both BASIC and XB give me "LINE TOO LONG", which means too many tokens (rather than characters).
 

I broke them up and ran on real hardware... I also get different results under Classic99. Are you on an older version? Current is 399.038.

 

TI BASIC (+EA): Hardware: 739  Classic99.038: 732  MAME (Mizapf): 728

XB: 821  Classic99.038: 815   MAME (Mizapf): 811

 

There's only a 1% delta between those... since CPU clock can vary by +/- 5% (per datasheet and observation seems to match), that's within spec for all of them.

 

I admit I am surprised. XB is faster at some things, but apparently not this math intensive work...

 

 

MANDLE

Edited February 28, 2021 by Tursi

[vol]

5 hours ago, TheBF said:

I am curious. Do you use the interrupt to poll the timer and accumulate in a memory location?

 

This is exactly how it is.

[vol]

4 hours ago, Tursi said:

How did you enter lines 164 and 166? Both BASIC and XB give me "LINE TOO LONG", which means too many tokens (rather than characters).
 

I broke them up and ran on real hardware... I also get different results under Classic99. Are you on an older version? Current is 399.038.

 

TI BASIC (+EA): Hardware: 739  Classic99.038: 732  MAME (Mizapf): 728

XB: 821  Classic99.038: 815   MAME (Mizapf): 811

 

There's only a 1% delta between those... since CPU clock can vary by +/- 5% (per datasheet and observation seems to match), that's within spec for all of them.

 

Thank you very much for your results.  I use Classic.037, now I am rechecking my results, maybe I made a mistake somewhere.

I think your problem with long lines happened because you used the editor where you pasted the text.  The editor has more limitations on the line length.  I used xbas99 (from there) to convert the text into Basic.

[vol]

10 hours ago, Tursi said:

TI BASIC (+EA): Hardware: 739  Classic99.038: 732  MAME (Mizapf): 728

XB: 821  Classic99.038: 815   MAME (Mizapf): 811

I have been able to figure out why your results differ from mine.  It is because your code has an extra line 168.  However I can't explain why this line affects timings.  It is an odd mystery for me.

BTW line 540 is also redundant. But it is good to add line 560 IF B=0 THEN 550.

Edited February 28, 2021 by vol

[vol]

I have been able to persuade T40XB to build the 38x21 Mandelbrot!

 

100 REM A BASIC, ASCII MANDELBROT
110 REM
120 REM THIS IMPLEMENTATION COPYRIGHT (C) 2019, GORDON HENDERSON
130 REM
140 REM PERMISSION TO USE/ABUSE ANYWHERE FOR ANY PURPOSE GRANTED, BUT
150 REM IT COMES WITH NO WARRANTY WHATSOEVER. GOOD LUCK!
160 REM TI-99/4A (T40) VARIANT BY LITWR, 2021
162 CALL LINK("T40")
163 CALL LINK("CHAR",126,"0000205408")
164 CALL LOAD(-20480,96,130,200,2,176,42,200,2,176,44,4,204,29,0,2,2,63,255,200,2,176,40,5,204)
165 CALL LOAD(-20456,51,130,6,76,30,0,2,2,177,0,200,2,131,196,4,91)
166 CALL LOAD(-20224,4,204,29,0,55,194,30,0,9,18,195,32,176,40,200,2,176,40,99,2,2,76,63,255)
167 CALL LOAD(-20200,168,12,176,44,23,2,5,160,176,42,4,96)
170 C$ = ".,'~=+:;[/<&?oxOX# " 
175 REM 'PALLET' LIGHTEST TO DARKEST...
180 SO = 1
185 REM SET TO 0 IF YOUR MID$() INDEXES FROM 0.
190 MI = LEN(C$)
200 MX = 4
210 LS = -2.0
220 TP = 1.25
230 XS = 2.5
240 YS = -2.5
250 W = 38
260 H = 21
270 SX = XS / W
280 SY = YS / H
282 CALL PEEK(-31804,T0,T1)
284 IF T0<>177 THEN CALL LOAD(-20188,T0,T1)
290 CALL LOAD(-31804,176,0)
300 FOR Y = 0 TO H
310   CY = Y * SY + TP
320   FOR X = 0 TO W
330     CX = X * SX + LS
340     ZX = 0
350     ZY = 0
360     CC = SO
370     X2 = ZX * ZX
380     Y2 = ZY * ZY
390     IF CC > MI THEN 460
400     IF (X2 + Y2) > MX THEN 460
410     T = X2 - Y2 + CX
420     ZY = 2 * ZX * ZY + CY
430     ZX = T
440     CC = CC + 1
450     GOTO 370
460     P$=P$&SEG$(C$, CC - SO, 1)
470   NEXT X
480   CALL LINK("PRINT",25,1,P$)
485   P$=""
490 NEXT Y
510 CALL PEEK(-20438,T0,T1,T2,T3)
530 CALL LINK("PRINT",25,1,((T0*256+T1)*65536+T2*256+T3)/46875)
550 CALL KEY(0,X,Y)
560 IF Y=0 THEN 550

This code can be copy-pasted.

I am very impressed by T40XB.  It has even a command to show a directory listing - CALL LINK("CAT") - a very useful feature.

However T40XB is still not polished enough.  Its glyph for the tilde character is corrupted so I had to define it in this program.  

 

Edited March 1, 2021 by vol

[Tursi]

On 2/28/2021 at 7:12 AM, vol said:

I have been able to figure out why your results differ from mine.  It is because your code has an extra line 168.  However I can't explain why this line affects timings.  It is an odd mystery for me.

BTW line 540 is also redundant. But it is good to add line 560 IF B=0 THEN 550.

Ahh, that makes sense. TI BASIC uses a line number lookup table to find code, it probably impacts the search time.

 

Yeah, I noticed 540 was pointless, but I didn't care enough to worry about it.

 

[senior_falcon]

in post #30, VOL wrote:

However T40XB is still not polished enough.  Its glyph for the tilde character is corrupted so I had to define it in this program.  

 

Thanks for catching that. In the T40XB included with the Betelgeuse3 package the tilde is one pixel higher than the one you defined and there is an extra dot in the bottom pixel row. The version of T40XB built into XB 2.8 G.E.M. has a tilde exactly like yours.

I will see why this is happening and correct it.