Changed mul_low algorithm to radix-16-1 algorithm with swap

Reviewed-on: TCShenanigans/symphony_stdlib#9

README.md

@@ -3,7 +3,7 @@
 This is a standard library for symphony.
 It is both intended as a practical toolkit to develop more complex software as well as a teaching resource.
 
-If you just want to use the standard library [[stdlib.asm]] is your main header, include it after your code.
+If you just want to use the standard library [[src/stdlib.asm]] is your main header, include it after your code.
 
 If you are using it as a learning resource have a look at the [teaching folder](teaching).
 

examples/README.md

@@ -0,0 +1,3 @@
+# Examples
+
+Examples of how to use the standard library to accomplish a task.

src/imath.asm

@@ -4,22 +4,53 @@
 ; r2 - The second value
 ; Result:
 ; r1 - The lower 32 bits of the result
-; Clobbers: r2, r3, r4, r5
+; Clobbers: r2, r3, r4, r5, r6
 pub mul_low:
-	mov r3, 0 ; result
+	cmp r1, r2
-	mov r4, 31 ; loop counter
+	jbe mul_low_noswap
-	
+	xor r1, r2, r1
-	mul_low_loop:
+	xor r2, r1, r2
-		asr r5, r2, 31
+	xor r1, r2, r1
-		and r5, r5, r1
+	; Passthrough
-		lsl r5, r5, r4
+
-		add r3, r3, r5
+; Multiplies r1 and r2, returning the lower part of the result
-		lsl r2, r2, 1
+; This method assumes r1 is smaller than r2, which results in faster execution
-		sub r4, r4, 1
+; Arguments:
-		cmp r4, 0
+; r1 - The first value
-		jge mul_low_loop
+; r2 - The second value
-	mov r1, r3
+; Result:
+; r1 - The lower 32 bits of the result
+; Clobbers: r2, r3, r4, r5, r6
+pub mul_low_noswap:
+	mov r4, 0 ; r4 has the result
+	mov r6, mul_loop_end
+	add r3, r2, r2
 	
+	mul_loop:
+		and r5, r1, 14 ; Taking the first four bits, discarding the odd
+		lsl r5, r5, 1 ; 0000 - 0, 0001 - 2, 0010 - 4, 0011 - 8, etc.
+		sub r5, r6, r5
+		jmp r5
+		
+		add r4, r4, r3
+		add r4, r4, r3
+		add r4, r4, r3
+		add r4, r4, r3
+		add r4, r4, r3
+		add r4, r4, r3
+		add r4, r4, r3
+		mul_loop_end:
+			mov flags, r1
+			jne mul_skip_one
+			add r4, r4, r2
+			mul_skip_one:
+			lsl r2, r2, 4
+			lsl r3, r3, 4
+			lsr r1, r1, 4
+			cmp r1, zr
+			jne mul_loop
+
+	mov r1, r4
 	jmp r13
 
 ; Calculates the absolute value of the value provided in the r1 register

tests/README.md

@@ -0,0 +1,3 @@
+# Tests
+
+Tests for the standard library go here, tests are allowed to depend on the recommended spec.isa changes.