corrected ISA bug and improved speed

examples/slu4-minimal-64/slu4-minimal-64.yaml

@@ -1004,8 +1004,8 @@ instructions:
       operand_sets:
         list:
           - absolute_address
-  ble:
-    # branch on less than or equal
+  bgt:
+    # branch on greater than
     bytecode:
       value: 0x61
       size: 8
@@ -1014,8 +1014,8 @@ instructions:
       operand_sets:
         list:
           - absolute_address
-  bgt:
-    # branch on greater than
+  ble:
+    # branch on less than or equal
     bytecode:
       value: 0x62
       size: 8
@@ -1358,6 +1358,19 @@ macros:
         - "sts @ARG(0)+1"
         - "lda @ARG(1)+1"
         - "sts @ARG(0)+0"
+  cpy2ai:
+    - operands:
+        count: 2
+        operand_sets:
+          list:
+            - absolute_address
+            - immediate_16bit
+      instructions:
+        # Copies in little endian order
+        - "ldi BYTE0(@ARG(1))"
+        - "sta @ARG(0)+0"
+        - "ldi BYTE1(@ARG(1))"
+        - "sta @ARG(0)+1"
   cpy4as:
     - operands:
         count: 2
@@ -1441,3 +1454,35 @@ macros:
         - "sts @ARG(0)+2"
         - "lds @ARG(1)+3"
         - "sts @ARG(0)+3"
+  inc16a:
+    - operands:
+        count: 1
+        operand_sets:
+          list:
+            - absolute_address
+      instructions:
+        - "lda @ARG(0)+0"
+        - "inc"
+        - "sta @ARG(0)+0"
+        - "lda @ARG(0)+1"
+        - "aci 0"
+        - "sta @ARG(0)+1"
+  inc32a:
+    - operands:
+        count: 1
+        operand_sets:
+          list:
+            - absolute_address
+      instructions:
+        - "lda @ARG(0)+0"
+        - "inc"
+        - "sta @ARG(0)+0"
+        - "lda @ARG(0)+1"
+        - "aci 0"
+        - "sta @ARG(0)+1"
+        - "lda @ARG(0)+2"
+        - "aci 0"
+        - "sta @ARG(0)+2"
+        - "lda @ARG(0)+3"
+        - "aci 0"
+        - "sta @ARG(0)+3"

examples/slu4-minimal-64/software/math32lib.min64

@@ -2,36 +2,26 @@
 
 ; compare32
 ;   Compares two 32-bit values to determine equality
+;       X ? Y
 ;
 ;   Arguments
-;       sp+3 : left X value (4 bytes)
-;       sp+7 : right Y value (4 bytes)
-;       sp+11 : results flags placeholder
+;       sp+3 : right Y value (4 bytes)
+;       sp+7 : left X value (4 bytes)
 ;
 ;   Returns
-;       sp+11 : results flags:
-;                   bit 0 - Zero Flag (values are equal)
-;                   bit 1 - Carry Flag (X > Y)
+;       flags will be set per comparison
 ;
 compare32:
-    ; init results
-    ldi 0 sta .flags    ; has to be RAM variable since stack ops alter flags
     ; first check high bytes, then others in sequence
     ; values on stack are stored big endian
-    lds (7+0) sta .rval lds (3+0) cpa .rval bne .done
-    lds (7+1) sta .rval lds (3+1) cpa .rval bne .done
-    lds (7+2) sta .rval lds (3+2) cpa .rval bne .done
-    lds (7+3) sta .rval lds (3+3) cpa .rval
-    ; check zero flag. flags still OK because no stack operatrion since CPA
-    bne .done   ldi %00000001 adb .flags    ; note this clears the flags
+    lds (3+0) sta .rval lds (7+0) cpa .rval bne .done
+    lds (3+1) sta .rval lds (7+1) cpa .rval bne .done
+    lds (3+2) sta .rval lds (7+2) cpa .rval bne .done
+    lds (3+3) sta .rval lds (7+3) cpa .rval
 .done:
-    ; check carry flag
-    bcc .return ldi %00000010 adb .flags
-.return:
-    lda .flags sts 11
     rts
 .rval: .byte 0
-.flags: .byte 0
+
 
 
 
@@ -159,25 +149,23 @@ multiply_int32:
 ;
 divide32:
     ; first check values for 0
-    phsi 0
     phs4i 0
-    phs4s (7+5)
+    phs4s (7+4)
     jps compare32
     pls4
-    lds 5 cpi 1 beq .divide_by_zero
-    phs4s (3+5)
+    beq .divide_by_zero
+    phs4s (3+4)
     jps compare32
     pls4
     pls4
-    lds 1 pls cpi 1 beq .return_zero
+    beq .return_zero
     ; check if divisor > dividend
-    phsi 0
-    phs4s (3+1)
-    phs4s (7+5)
+    phs4s (7+0)
+    phs4s (3+4)
     jps compare32
     pls4
     pls4
-    lds 1 pls cpi 2 beq .divisor_too_large
+    bgt .divisor_too_large
 .start_division:
     ; set up working stack:
     ;   little endian
@@ -195,11 +183,10 @@ divide32:
     sta .working_mem+0
     ldi 0 sta .carry_bit
     ; determine if we can do subtraction
-    phsi 0                 ; compare results
-    phs4s 7+1              ; divisor
-    phs4a .working_mem+4   ; working value high word
+    phs4s 7                ; divisor (left)
+    phs4a .working_mem+4   ; working value high word (right)
     jps compare32
-    lds 9 cpi 0 beq .div_loop_continue
+    bgt .div_loop_continue
 .div_loop_subtraction:
     ; working value is equal to or larger than divsior
     ; do the subtraction
@@ -211,7 +198,6 @@ divide32:
     ; clear stack
     pls4
     pls4
-    pls
     ; decrement counter and check for 0
     deb .counter
     lda .counter cpi 0 bne .div_loop
@@ -265,25 +251,18 @@ divide32:
 ;       sp+7 - value Y, 4 byte value
 ;
 ;   Return Value
-;       sp+2 - replace the original 4 byte value with the sum
+;       sp+3 - replace the original 4 byte value with the sum
 ;
 add32:
-    ; since interacting with the stack has an undefined impact
-    ; on the carry flags, we need to copy the values to local
-    ; variables first.
     ; stack is big endian, save locally little endian
-    cpy4as .xval, 3
     cpy4as .yval, 7
-    ; star subtractiosn with LSB
-    lda .xval+0 ada .yval+0 sta .xval+0
-    lda .xval+1 aca .yval+1 sta .xval+1
-    lda .xval+2 aca .yval+2 sta .xval+2
-    lda .xval+3 aca .yval+3 sta .xval+3
-    ; save results back to stack
-    cpy4sa 3, .xval
+    ; start addition with LSB, Remember, stack is big endian
+    lds 3+3 ada .yval+0 sts 3+3
+    lds 3+2 aca .yval+1 sts 3+2
+    lds 3+1 aca .yval+2 sts 3+1
+    lds 3+0 aca .yval+3 sts 3+0
     ; end return
     rts
-.xval: .4byte 0
 .yval: .4byte 0
 
 
@@ -299,22 +278,15 @@ add32:
 ;
 ;
 subtract32:
-    ; since interacting with the stack has an undefined impact
-    ; on the carry flags, we need to copy the values to local
-    ; variables first.
-    ; stack is big endian, save locally little endian
-    cpy4as .xval, 3
+    ; stack is big endian, save Y locally little endian
     cpy4as .yval, 7
     ; star subtractiosn with LSB
-    lda .xval+0 sba .yval+0 sta .xval+0
-    lda .xval+1 sca .yval+1 sta .xval+1
-    lda .xval+2 sca .yval+2 sta .xval+2
-    lda .xval+3 sca .yval+3 sta .xval+3
-    ; save results back to stack
-    cpy4sa 3, .xval
+    lds 3+3 sba .yval+0 sts 3+3
+    lds 3+2 sca .yval+1 sts 3+2
+    lds 3+1 sca .yval+2 sts 3+1
+    lds 3+0 sca .yval+3 sts 3+0
     ; end return
     rts
-.xval: .4byte 0
 .yval: .4byte 0
 
 

examples/slu4-minimal-64/software/primes.min64

@@ -25,12 +25,7 @@ start:
     phs4a _n_value jps print_value32 pls4
     phs2i is_prime_str jps _Print pls2
 .increment_n:
-    phs4a _n_value
-    phs4i 1
-    jps add32
-    cpy4as _n_value,1
-    pls4
-    pls4
+    inc32a _n_value
     jpa .n_loop
 
 _n_value: .4byte 0
@@ -64,6 +59,11 @@ print_value32:
 ;
 ;   returns
 ;       sp+7 : 0 or 1 depending on whether N  is prime
+
+_isquared_str: .cstr "I^2 = "
+_nprefix_str: .cstr ", N = "
+_eol_str: .cstr "\n"
+
 is_prime32:
     ; first check for 2 or 3
     lds 3 cpi 0 bne .modulo_two        ; check top byte of N for 0
@@ -76,19 +76,17 @@ is_prime32:
 .modulo_two:
     lds 6 lsr bcc .is_not_prime        ; see if N's least signficant bit is even or odd
 .modulo_three:
-    phsi 0                            ; results placeholder for compare later
     phs4i 3                           ; place divisor on stack
-    phs4s (3+5)                       ; place dividend on stack (from stack)
+    phs4s (3+4)                       ; place dividend on stack (from stack)
     jps divide32
     pls4
     ; check if remainder is 0
     phs4i 0 jps compare32 pls4
     pls4
-    pls cpi 1 beq .is_not_prime
+    beq .is_not_prime
 .loop_init:
     cpy4ai .current_i_val,5
 .loop:
-    phsi 0         ; push placeholder on stack for later compare
     phs4a .current_i_val
     phs4a .current_i_val
     jps multiply_uint32
@@ -98,40 +96,40 @@ is_prime32:
     lds 3 cpi 0 bne .iteration_loop_done
     lds 4 cpi 0 bne .iteration_loop_done
     pls4
-    ; now compare low 4 bytes of I*I result to N
-    phs4s (3+5) jps compare32 pls4
+    cpy4as .isquared,1
     pls4
-    ; if N < I*I, we are done
-    pls cpi 0 beq .loop_done_is_prime
+    ; now compare low 4 bytes of I*I result to N
+    phs4a .isquared phs4s 3+4 jps compare32 pls4 pls4
+    ; if I*I > N, we are done
+    bgt .loop_done_is_prime
+.n_gte_i_squared:
     ; now check various modulos.
     ; check N % I == 0
-    phsi 0         ; push placeholder on stack for later compare
     phs4a .current_i_val   ; I
-    phs4s (3+5)            ; N
+    phs4s (3+4)            ; N
     jps divide32
     pls4                   ; quotient
     phs4i 0
     jps compare32
     pls4
     pls4
-    pls cpi 1 beq .loop_done_is_not_prime
+    beq .loop_done_is_not_prime
     ; check N % (I+2) == 0
     phs4a .current_i_val   ; I
     phs4i 2
     jps add32
     cpy4as .temp_val, 1
     pls4
     pls4
-    phsi 0
     phs4a .temp_val
-    phs4s (3+5)
+    phs4s (3+4)
     jps divide32
     pls4                   ; quotient
     phs4i 0
     jps compare32
     pls4                   ; zero
     pls4                   ; remainder
-    pls cpi 1 beq .loop_done_is_not_prime
+    beq .loop_done_is_not_prime
     ; add 6 to I and loop
     phs4a .current_i_val
     phs4i 6
@@ -144,7 +142,6 @@ is_prime32:
     ; get rid of I*I stack
     pls4
     pls4
-    pls
 .loop_done_is_prime:
 .is_prime:
     ldi 1 sts 7
@@ -156,6 +153,7 @@ is_prime32:
 .current_i_val: .4byte 0
 .temp_val: .4byte 0
 
+.isquared: .4byte 0
 
 #include "math32lib.min64"
 #include "stringlib.min64"