Subject:
|
Re: h8/300 16-bit multiply
|
Newsgroups:
|
lugnet.robotics.rcx
|
Date:
|
Mon, 4 Mar 2002 19:32:54 GMT
|
Viewed:
|
2789 times
|
| |
|
|
Thanks. I have a version just 2 words longer that doesn't use the ROM
code. Basically it's the ROM's code, trimmed down after assuming high
words are zero initially. It's completely untested though, so there
may be a bug.
CODE UM*
\ 16x16=32 multiply
\ R6 is first argument, top of stack is second, overwritten with result
\ R0,R1,R2 are scratch
\
mov.w @R7, R5
mov.w R6, R1 \ keep copies of arguments
mov.w R5, R2
mov.b R6h, R6l \ Ah*Bh
mulxu.b R5h, R6
mulxu.b R1l, R5 \ Al*Bl
mov.b R1l, R0l \ Al*Bh
mulxu.b R2h, R0
add.b R0l, R5h \ add to result
addx.b R0h, R6l
addx.b #0, R6h
mov.b R1h, R0l \ Ah*Bl
mulxu.b R2l, R0
add.b R0l, R5h \ add to result
addx.b R0h, R6l
addx.b #0, R6h
mov.w R5, @R7 \ low word of result
next
END-CODE
Yes, it's for my own Forth; given that it's for Mindstorms, reinventing
the wheel is supposed to be half the fun.
By the way: in your code a "MOV.W #0x0000,r5" takes 4 bytes,
whereas "SUB.W r5,r5" takes only 2 bytes.
--
Darin Johnson
Support your right to own gnus.
|
|
Message has 1 Reply:
 | | RE: h8/300 16-bit multiply
|
| (...) <snipped code...> Are you sure that it works for signed numbers :-) (...) If you like, I can give you some notes on how my Forth was developed. I'm using a simple pseudo-assembler that lets me port this Forth to a new processor in less than a (...) (23 years ago, 4-Mar-02, to lugnet.robotics.rcx)
|
Message is in Reply To:
| | RE: h8/300 16-bit multiply
|
| (...) The easiest thing is to cast your 16 bit values to 32 bits and use the built-in 32x32 multiply... Here's the code from pbForth that does it. (It's written in a pseudo- assebler, but you can probably figure it out...) #---...--- # UM* UM* ( u1 (...) (23 years ago, 4-Mar-02, to lugnet.robotics.rcx)
|
4 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
