speed optimization of multiplication routine

STATUS.md

@@ -199,7 +199,7 @@ Floating point mathematical routines - not official, but well known and broadly
 | `$B947`   | `inv_FAC1_mantissa`    | NOT DONE |                                                    |
 | `$B9EA`   | `log_FAC1`             | NOT DONE |                                                    |
 | `$BA28`   | `mul_MEM_FAC1`         | DONE     |                                                    |
-| `$BA2B`   | `mul_FAC2_FAC1`        | PARTIAL  | very imprecise; not yet clear, why                 |
+| `$BA2B`   | `mul_FAC2_FAC1`        | PARTIAL  | sometime imprecise - like PI*PI                    |
 | `$BA8C`   | `mov_MEM_FAC2`         | DONE     |                                                    |
 | `$BA90`   | `get_FAC2_via_INDEX`   | NOT DONE |                                                    |
 | `$BAE2`   | `mul10_FAC1`           | DONE     |                                                    |

c64/basic/math/mul_FAC2_FAC1_8x8.s

@@ -7,13 +7,20 @@
 // - https://codebase64.org/doku.php?id=base:8bit_multiplication_16bit_product_fast_no_tables
 //
 // input: .A, INDEX+3
-// output: .X (high byte), .A and INDEX+2 (low byte), carry always clear
+// output: .X (high byte), .A (low byte), carry always clear
 //
 
 mul_FAC2_FAC1_8x8:
 
 	sta INDEX+2
 
+	// Special case (optimization) - handle 0
+
+	lda INDEX+2
+	beq mul_FAC2_FAC1_8x8_zero
+	lda INDEX+3
+	beq mul_FAC2_FAC1_8x8_zero
+
 	lda #$00
 
 	dec INDEX+3	// decrement because we will be adding with carry set for speed (an extra one)
@@ -62,6 +69,14 @@ mul_FAC2_FAC1_8x8:
 
 	tax
 	lda INDEX+2
-	clc
 
+	clc
+	rts
+
+mul_FAC2_FAC1_8x8_zero:
+
+	lda #$00
+	tax
+
+	clc
 	rts