The Museum of HP Calculators

HP Articles Forum

[Return to the Index ]
[ Previous | Next ]

41-MCODE: Dr. Jekyll & Mr. Hyde

Posted by Ángel Martin on 7 July 2012, 3:48 a.m.

I recently spent a little time experimenting with ideas to overcome the 64-function FAT limitation.

As the title suggest, this is about using alternative (or parallel) FATs for the same module (say it´s a 4k ROM). The situation is therefore that there is empty room left in the page to add more functions, but no more FAT entries are available (yes, we´ve all been here before many times...)

The idea is to use a couple of routines and their corresponding tables to re-write the FAT "on demand", thus the dual-personality concept. Such a dynamic re-writing can only be done when the module is located in RAM, be that the HEPAX, the MLDL2k, or - of course- the CL sRAM, plenty of pages there...

FAT-A would include the "MRHYDE" function, so you can switch to the FAT-B arrangement; and conversely the FAT-B would have the "DJEKYLL" routine to switch back.

This can also be implemented as an auto check upon the calculator-ON event, using the polling points to prompt which choice you want to start with. This would not require the two functions, although they´ve grown on me already, and besides they can always be used for safety.

Note that there´s no bank-switching involved. I first considered this approach but unfortunately (for this purpose but very fortunately for many others) the CL system always returns to the first bank when the calculator goes into ligh sleep, thus a hypothetical bank-2 FAT even if it existed wouldn´t hold a chance; let´s say it wouldn´t "stick" :-)

As example, let´s see how the ALPHA ROM outgrows its capabilities to have a total of 70 functions, split in 41 and 29 for each group.

The switching routines are as follows:


Header	A05B	08C	"L"	
Header	A05C	00C	"L"	
Header	A05D	019	"Y"	
Header	A05E	00B	"K"	
Header	A05F	005	"E"	
Header	A060	00A	"J"	
Header	A061	004	"D"	
DJEKIL	A062	35D	?NC XQ  	Refresh C[3,6]
	A063	000	->00D7  	PCTOC 
	A064	09C	PT=  5	
	A065	04A	C=0  R<-	
	A066	0FA	C<>B  M 	Write-To adr in B[M]
	A067	0DA	C=B  M	
	A068	3D0	LD@PT- F	
	A069	090	LD@PT- 2	"pF1F"
	A06A	1D0	LD@PT- 7	address of FAT-B table
	A06B	083	JNC +16d	
Header	A06C	085	"E"	
Header	A06D	004	"D"	
Header	A06E	019	"Y"	
Header	A06F	008	"H"	
Header	A070	012	"R"	
Header	A071	00D	"M"	
MRHYDE	A072	35D	?NC XQ  	Refresh C[3,6]
	A073	000	->00D7  	PCTOC 
	A074	09C	PT=  5	
	A075	04A	C=0  R<-	clear rest of ADR field
	A076	0FA	C<>B  M 	Write-To adr
	A077	0DA	C=B  M  	keep it in C[ADR]
	A078	390	LD@PT- E	
	A079	390	LD@PT- E	"pEE8"
	A07A	210	LD@PT- 8	address of FAT-A table
	A07B	330	FETCH S&X	read source word
	A07C	0FA	C<>B  M 	swap addresses
	A07D	040	WROM            write word to target
	A07E	23A	C=C+1  M	increase target adr
	A07F	0FA	C<>B  M 	recall source adr
	A080	23A	C=C+1  M	increase source adr
	A081	2F6	?C#0  XS	end of table?
	A082	3CB	JNC  -07	no, loop back
	A083	3E0	RTN     	yes, done.

and the tables are just an exact copy of the respective FAT´s, including the two zero words at the end, and followed by any word with a non-zero value in the XS field - which marks the end of the table to the read-write routines.

The FATS have each one extra function, to activate the parallel versions. So FAT-A includes ¨MRHYDE¨ and FAT-B includes ¨DJEKYLL¨.

The requirement impossed here is obviously the additional space required by the two tables and the switching routines (snipets really). A maximum of 128 functions (of which two are the housekeeping ones) can be accomodated, requiring 272 (2x136) words for the tables - but that´s the worst-case scenario.

a FOCAL program requiring a function on the "other" FAT will simple use the "personality function" to activate it first, then back (or not depending on what comes next). To be on the safe side one can always add such call with flag 25 to trap errors in case the FAT is already the active one.

Really can´t be any simpler, and it´s a lot of fun to watch. Email me if you want a copy of the ALPHA44 module to play with, and by all means let´s keep thinking on a way to eliminate that nasty restriction in better ways...

Cheers, AM

Edited: 9 July 2012, 9:08 a.m.

Password:

[ Return to the Message Index ]

Go back to the main exhibit hall