To the 2901 bit slicers out there

Jon Elson elson at pico-systems.com
Thu Dec 28 11:18:08 CST 2017


> On 28.12.2017 05:35, Randy Dawson via cctalk wrote:
>> What about the coding tools?  ADASM? Looks long gone, how 
>> do you do microcode today?
>>
>
>
Oh, I first started out with an assembler macro hack, but it 
was insanely cumbersome, due to the limits of the macro 
capability of the Z-80 assembler I was using.

I then wrote a micro-assembler in Pascal.  It was better, 
but still pretty cumbersome.  There's a good chance that if 
I actually had gotten to writing the full 360 microcode I 
would have come up with improvements.

Here's a little test program for the old asm macro-based 
assembler:

;DIM LIGHT PROG FOR MICRO ASSEMBLER
     MACLIB    UASM
     INIT
;  DEFINE 2 CONSTANTS IN SCRATCH PAD
SPONE    DEFCON    1
SP4096    DEFCON    4096
;  END OF CONSTANTS
DIM    REGS    GPR0,GPR0,GPR0        ;CLEAR GPR0
     FUNC    ZERO
     CTRL    CONT
     GEN
     REGS    ACONST,GPR1,GPR1    ;COPY SPONE TO GPR1
     DEST    PASSWRT
     FUNC    APLUSCY
     CONST    SPONE
     CTRL    CONT
     GEN
     REGS    ACONST,GPR3,GPR3    ;COPY SP4096 TO GPR3
     DEST    PASSWRT
     FUNC    APLUSCY
     CONST    SP4096
     CTRL    CONT
     GEN
BIGLP    REGS    GPR0,GPR0,GPR0        ;INC GPR0, MOVE TO 
COUNTER
     DEST    PASSWRT
     FUNC    APLUSCY
     CY    ONE
     CTRL    LDRCONT
     NA    0,I            ;CBUS => CTR
     GEN
HILOOP    REGS    GPR2,GPR2,GPR2        ;ALL ONES TO CBUS
     DEST    PASSWRT
     FUNC    ALLONE
     CTRL    REPLNA
     NA    HILOOP

The GEN command generates one micro-instruction word from 
all the keywords above it.


Here's what my 2nd gen Pascal micro-assembler code looked like :
     CODE
;    TITLE    muldiv
;This program does multiply and divide on 2903/2904 32 bit CPU
; UNSIGNED MULTIPLY 32 X 32 => 64
; CLEAR R0 FOR MS PRODUCT
; MULTIPLICAND IN R1, MULTIPLIER IN R2
; PRODUCT IN R0 (MS) AND R1 (LS)
     ORG    0
UMULT    R=GPR0,GPR0,GPR0,F=ZERO,D=W        ;CLEAR GPR0
     R=GPR2,GPR2,GPR2,D=LDQ&W,F=A+CY,Q=LDRCONT,N=31    ;mOVE 
MPLIER TO Q REG
                         ;#MPLY CYCLES -1 => CTR
UMLP    R=GPR1,GPR0,GPR0,D=MPYUSG,F=SPEC,Q=REPLNA,N=UMLP
;*****    SHFT    USGMPY
     R=GPR1,GPR1,GPR1,F=Q+CY,D=W        ;COPY LS PROD IN Q TO R1
UWAIT    D=PASS,F=ZERO,Q=JUMP,N=UWAIT        ;hang for user 
to check result
; END OF UNSIGNED MULTIPLY
;
;
; SIGNED MULTIPLY 32 X 32 => 64
; CLEAR R0 FOR MS PRODUCT
; MULTIPLICAND IN R1, MULTIPLIER IN R2
; PRODUCT IN R0 (MS) AND R1 (LS)
SMULT    r=GPR0,GPR0,GPR0,F=ZERO,D=W        ;CLEAR GPR0
     r=GPR2,GPR2,GPR2,D=LDQ&W,F=A+CY,Q=LDRCONT,N=30 ;mOVE 
MPLIER TO Q REG
                             ;#MPLY CYCLES -2 => CTR
SMLP    r=GPR1,GPR0,GPR0,D=MPY2CM,F=SPEC,Q=REPLNA,N=SMLP ;2S 
COMPL MPY
;*****    SHFT    MPY2CM
     r=GPR1,GPR0,GPR0,D=MPY2LST,F=SPEC    ;2S COMPL MPY LAST 
CYCLE
     r=GPR1,GPR1,GPR1,F=Q+CY,D=W        ;COPY LS PROD IN Q TO R1
SWAIT    D=PASS,F=ZERO,Q=JUMP,N=SWAIT        ;hang for user 
to check result
; END OF UNSIGNED MULTIPLY
     END

This tests the multiply functions on the 2903.  The tab 
formatting got messed up pasting this here,
it looked better in the original form.

Jon


More information about the cctech mailing list