Update description of IR in README

Improve README style
2020-03-21 23:02:38 +01:00 · 2020-03-21 22:39:57 +01:00 · 2020-03-21 21:09:55 +01:00 · 2019-05-07 17:28:34 +02:00 · 2019-05-06 10:31:33 +02:00 · 2019-05-06 10:22:55 +02:00
22 changed files with 3508 additions and 87 deletions
--- a/.gitignore
+++ b/.gitignore
@ -0,0 +1,5 @@
+META-INF/
+*.class
+*.jar
+*.out
+build.mia
--- a/13
+++ b/13
@ -0,0 +1,13 @@
+KCOMPILER=kotlinc
+KEXEC=kotlin
+TEMPLATE=weaver.jar compiler.jar microcompiler.jar
+
+$(TEMPLATE): %: $(%:.jar=.kt)
+	$(KCOMPILER) $(@:.jar=.kt) -d $@
+
+all: weaver.jar compiler.jar microcompiler.jar
+
+clean:
+	rm -f weaver.jar
+	rm -f compiler.jar
+	rm -f microcompiler.jar
--- a/README.md
+++ b/README.md
@ -0,0 +1,616 @@
+# LMIA advanced development kit
+
+This project was developed as a simple development kit for programming and
+microprogramming the LMIA system.
+
+
+## μASM instruction set
+
+### NOP
+No-operation. This wastes one clock cycle.
+
+
+### MOV [regA] \[regB\]
+Move value in *regA* to *regB*
+
+*This operation uses the bus*
+
+*If regB is [**LC**](#lc), no other [**LC**](#lc) operation can be specified in the
+same cycle*
+
+
+### MVN {[reg] | [const]}
+Move the inverse of a value (from register or constant) into register
+[**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+
+### MVZ
+Set [**AR**](#ar) to zero.
+
+*Sets flags: [**Z**](#z), [**N**](#n)*
+
+
+### ADD {[reg] | [const]}
+Add value (from register or constant) to register [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**O**](#o), [**C**](#c)*
+
+
+### SUB {[reg] | [const]}
+Subtract value (from register or constant) from register [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**O**](#o), [**C**](#c)*
+
+
+### AND {[reg] | [const]}
+Perform bitwise AND with given value (from register or constant) and register
+**AR**. *reg* is ignored and only prevalent due to a microcompiler quirk;
+expect it to be removed in future releases.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n)*
+
+
+### ORR {[reg] | [const]}
+Perform bitwise Or with given value (from register or constant) and register
+**AR**.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n)*
+
+
+
+### ADN {[reg] | [const]}
+Add value (from register or constant) to register [**AR**](#ar) without
+updating flags.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+
+### LSL
+Performs a logical shift left of [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### ISL
+Performs a logical shift left of [**AR**](#ar) and [**HR**](#hr) as if they
+were one 32-bit register where [**AR**](#ar) corresponds to the
+most-significant bits.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### ASR
+Performs an arithmetic shift right on [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### ISR
+Performs an arithmetic shift right of [**AR**](#ar) and [**HR**](#hr) as if
+they were one 32-bit register where [**AR**](#ar) corresponds to the
+most-significant bits.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### LSR
+Performs a logical shift right of [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### ROL
+Performs an arithmetic rotate right on [**AR**](#ar).
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### IRL
+Performs an arithmetic shift left of [**AR**](#ar) and [**HR**](#hr) as if they
+were one 32-bit register where [**AR**](#ar) corresponds to the
+most-significant bits.
+
+*This operation uses the bus*
+
+*If a constant is passed, this operation cannot be parallellized*
+
+*Sets flags: [**Z**](#z), [**N**](#n), [**C**](#c)*
+
+
+### LCSET [const]
+Set [**LC**](#lc) to value of constant.
+
+
+### CONST [const]
+Set [**AR**](#ar) to value of cosntant.
+
+*This operation cannot be parallelized*
+
+
+### INCPC
+Increment value in [**PC**](#pc) by one.
+
+
+### DECLC
+Decrement value in [**LC**](#lc) by one.
+
+
+### CALL [label]
+Move value in [**uPC**](#upc) to [**uSP**](#usp) and set value in [**uPC**](#upc) to point
+to the address of the given label.
+
+
+### RET
+Move value in [**uSP**](#usp) into [**uPC**](#upc).
+
+
+### HALT
+Stop execution and set value in [**uPC**](#upc) to **0**.
+
+
+### BRA [label]
+Perform an unconditional branch to the address of the given label.
+
+
+### BNZ [label]
+Branch to address of label if [**Z-flag**](#z) is **0**.
+
+
+### BRZ [label]
+Branch to address of label if [**Z-flag**](#z) is **1**.
+
+
+### BRN [label]
+Branch to address of label if [**N-flag**](#n) is **1**.
+
+
+### BRC [label]
+Branch to address of label if [**C-flag**](#c) is **1**.
+
+
+### BRO [label]
+Branch to address of label if [**O-flag**](#o) is **1**.
+
+
+### BLS [label]
+Branch to address of label if [**L-flag**](#l) is **1**.
+
+
+### BNC [label]
+Branch to address of label if [**C-flag**](#c) is **0**.
+
+
+### BNO [label]
+Branch to address of label if [**O-flag**](#o) is **0**.
+
+
+### BOP
+Branch to address specified by entry in optable pointed to by highest [**OP**](#op)
+in [**IR**](#ir).
+
+
+### BAM
+Branch to address specified by entry in addressing mode pointed to by [**M**](#m)
+in [**IR**](#ir).
+
+
+### BST
+Branch to start. This sets value in [**uPC**](#upc) to **0**.
+
+
+### RESET [reg]
+Sets all bits in the specified register to **1**.
+
+*This operation uses the bus*
+
+
+## Compiler/weaver directives
+
+### \#define \[name] [const]
+Define a compile-time constant. This will replace all (valid) constant
+declarations with the given name with the value supplied here.
+
+**NOTE**: Constant names are case-insensitive; i.e. *FOO* and *foo* are
+functionally indistinguishable to the compiler.
+
+
+### \#data \[address] [const]
+Define an initial value in the program memory at the given address.
+
+
+### $[label]
+Define a compile-time label at the given position in the microprogram. Labels
+can be referenced using an '@' symbol.
+
+For example: `$BAR` would declare a label *BAR* which can be referenced with
+`@BAR`.
+
+**NOTE**: Label names are case-insensitive; i.e. `@FOO` and `@foo` are
+functionally indistinguishable to the compiler.
+
+
+### \#optable \[index] {[label] | [const]}
+Declare an entry in the opcode jump table (**K1**).
+
+**NOTE**: The given index must be at most 15 and at least 0 and in the case of
+a constant being supplied as the value, this value my not be negative nor be
+greater than 127.
+
+
+### \#amode \[index] {[label] | [const]}
+Declare an entry in the addressing mode jump table (**K2**).
+
+**NOTE**: The given index must be at most 3 and at least 0 and in the case of
+a constant being supplied as the value, this value my not be negative nor be
+greater than 127.
+
+
+### \#pmgen [python]
+Define a script to be run for each address in program memory. This can be used
+to generate program memory value mappings more efficiently than manually
+defining values. The script is run in a for-loop with the index variable
+`address`. I.e. the variable `address` can be used in the specified script to
+refer to the current program memory address. To emit a value, simply print a
+string in the format "\[address] [value]" where the value can be in the range
+-32768 to 65535 (inclusive).
+
+
+### \#emit
+Define a script to emit uASM code at compile-time. Each following line preceded
+by a `>` will be included in the script; the first line not preceded by `>`
+(or EOF) will define the end of the script. To emit code, simply print the code
+that should be emitted.
+
+Example:
+
+```
+mov ar asr
+#emit
+>for i in range(4):
+>  print("lsl; mov ar ir")
+>  print("add pm")
+sub gr
+```
+
+the above code will be converted into the the following code by the preprocessor:
+
+```
+mov ar asr
+lsl; mov ar ir
+add pm
+lsl; mov ar ir
+add pm
+lsl; mov ar ir
+add pm
+lsl; mov ar ir
+add pm
+sub gr
+```
+
+
+## Flags
+Flags - *aside from [**L**](#l)* - are set based on ALU operations, so they depend
+on [**AR**](#ar) and the **BUS**. Henceforth, unless otherwise implied or
+stated, [**AR**](#ar) will refer to the state/value of [**AR**](#ar) *after* an
+ALU operation. Flags retain their state until an instruction which is declared
+as modifying said flag is executed.
+
+### Z
+Set if [**AR**](#ar) == **0**.
+
+### N
+Set if sign bit in [**AR**](#ar) is set.
+
+### O
+Set if sign of [**AR**](#ar) differs from signs of both [**AR**](#ar) *before*
+the arithmetic operation and of the **BUS**.
+
+### C
+Set if [**AR**](#ar) is less than or equal to [**AR**](#ar) *before* the
+arithmetic operation.
+
+### L
+Set if [**LC**](#lc) == **0**.
+
+
+## Registers
+Available registers for read/write operations are documented below. Unless
+otherwise specified, the registers are directly accessible via the bus for read
+and write operations.
+
+### AR
+The accumulator register. This register can only be written to as the result of
+an ALU operation. This is to say, that **AR** is indirectly writable via the ALU,
+but is nonetheless directly readable via the bus.
+
+### PM
+The program-memory pseudo-register allows you to read/write values from the
+currently accessed program memory address (see [**ASR**](#asr-1)).
+
+### ASR
+The address register; this register is used to specify which address of
+program-memory to be accessible via [**PM**](#pm).
+
+**NOTE**: This register cannot be read.
+
+### HR
+The help register. This is a general-purpose register which is useful for
+storing ephemeral or intermediate values during a computation.
+
+### IR
+The instruction register. This register offers extra functionality such as
+**K1**- and **K2**-table addressing via the [**OP**](#op) and [**M**](#m)
+bits respectively. The [**GRx**](#grx) and [**M**](#m) bits can also be used
+to address a specific general register via the [**GR**](#gr) multiplexer.
+
+Bit-level layout of IR (from MSB to LSB, left-to-right):
+
+| **OP** | **GRx** | **M**  | **ADR** |
+|:------:|:-------:|:------:|:-------:|
+| 4 bits | 2 bits  | 2 bits | 8 bits  |
+
+
+##### OP
+Machine instruction. Indexes **K1**-table.
+
+##### GRx
+[**GR**](#gr) multiplexer selector.
+
+##### M
+Addressing mode. Indexes **K2**-table.
+
+##### ADR
+[**ASR**](#asr-1) address argument.
+
+
+
+### GR
+This is a shorthand for accessing the general register currently made available
+by the GR multiplexer when said MUX is controlled by the [**GRx**](#grx) bits in
+[**IR**](#ir).
+
+**NOTE**: Only one GR can be accessed per cycle. Which register this is (of
+the four available registers) is determined by the value in [**IR**](#ir).
+
+### GRM
+This is a shorthand for accessing the general register currently made available
+by the GR multiplexer when said MUX is controlled by the [**M**](#m) bits in
+[**IR**](#ir).
+
+**NOTE**: Only one GR can be accessed per cycle. Which register this is (of
+the four available registers) is determined by the value in [**IR**](#ir).
+
+### LC
+The loopcounter register is a special register that can only be modified in
+three ways: all three ways are dictated by the L-field in the microprogram.
+Additionally, the counter can remain unchanged by simply not specifying an
+action to take for the register. The three ways of modifying it are as follows:
+
+* Decrement counter by one. The corresponding uASM instruction for this is
+`declc`
+* Load value from bus. The uASM instruction being `mov [reg] LC`
+* Load a 7-bit constant from micromemory. The uASM instruction for this is
+`lcset [const]`
+
+**NOTE**: LC cannot be read and can only be written to as detailed above. The
+value in LC can, though, to some degree be inferred from the [**L-flag**](#l)
+(see [*Flags*](#flags)).
+
+### PC
+The program counter. This register is only 8 bits wide. Writing a value larger
+than 8 bits to this register will simply truncate the binary string to 8 bits.
+Attemping to read a value from PC to a register of a larger width will result
+in the value in PC populating the lowesr 8 bits and the rest being filled with
+0's.
+
+### uPC
+The microprogram counter; sometimes also referred to as **MyPC**. This register
+cannot directly be read. It can, though, be written to in a variety of ways.
+These have been specified as *branch* instructions (such as [`BRA`](#bra-label)), as
+well as subroutine instructions (such as [`ret`](#ret)).
+
+**NOTE**: This register is not connected to the bus.
+
+### uSP
+The microstack pointer; sometimes also referred to as **MySPC**. This register
+cannot be directly read or written to. To write to it, a call to a subroutine
+must be issued (using [`call [label]`](#call-label)). The only way to "read" this
+register is to issue a subroutine-return instruction ([`ret`](#ret)) which
+copies the value in this register to [**uPC**](#upc).
+
+**NOTE**: This register is not connected to the bus. The value in this register
+cannot be directly acted upon insofar as it cannot be directly read or copied
+to any other register than [**uPC**](#upc), meaning that the actual value in
+this register (and consequently in [**uPC**](#upc) aswell) cannot be directly
+accessed or transformed. I.e. the value in this register cannot be copied to,
+for example, a general register and thus, its value can only be inferred, not
+directly measured.
+
+
+## Sorting algorithms
+For the sorting competition, we have chosen to focus on implementing bucketsort
+with an inline insertionsort when inserting values into corresponding buckets.
+
+### bucksort.uc
+
+This was the first attempt at a sorting implementation in uASM. It doesn't do
+more than a simple hash and possibly updating some bucket-specific values.
+Nonetheless, it formed a clear basis for future implementations.
+
+Pros:
+* N/A
+
+Cons:
+* N/A
+
+**Average cycle count: N/A**
+
+
+### bsrt.uc
+
+The first successful uASM bucketsort implementation. It uses the aforementioned
+inline insertionsort. Additionally, it employs a lookup table for pointing to
+buckets; this had the benefit of significantly decreasing the cycles required
+to hash values, as well as allowing for buckets of sizes other than even
+exponents of 2 (as opposed to other implementations). On top of this, it used a
+parallel-hashing system which allowed the rotation step of the hash algorithm
+to be applied to two elements of the list (to be sorted) at once, further
+decreasing the amount of cycles required to perform a hash (per element).
+
+Pros:
+* Variable bucket size
+* Arbitrary bucket arrangement
+* Parallel-hash implementation
+* Uses the most recent merge algorithm (rated around 400 cycles)
+
+Cons:
+* Heavy bookkeeping due to lookup table paired with parallel-hash
+* Inefficiency in merge due to lookup table (average loss of 80 cycles)
+
+**Average cycle count: 1250**
+
+
+### sort2.uc
+
+A second iteration of the *bsrt* implementation, this one refines the dual-hash
+by omitting the lookup table. This means that bookkeeping can be minimized, as
+it was only necessary due to a lack of available registers. This, of course,
+comes at the cost of not being able to have variable-sized buckets; i.e. their
+sizes must an exponent of 2. A minor optimization that was created for this
+algorithm was the hash-based bookkeeping, wherein a small optimization to
+bookkeeping could be done during the hashing of the list elements. Any
+bookkeeping which does not require the ALU can, in fact, be performed during the
+hashing, since the rotation steps only require AR and HR to be untouched. All
+other operations are permitted.
+
+Pros:
+* Merge: (320-330 cycles)
+    * Direct addressing fits optimally due to register limitations
+    * Optimized regster allocations
+* Minimal bookkeeping since it is mostly performed during the hashing operation
+
+Cons:
+* Bucket sizes must be exponents of 2
+* Fixed bucket indices based on hash
+* Two merge operations required (negative + positive)
+* 95 unused program-memory addresses
+
+**Average cycle count: 1100**
+
+
+### sort3.uc
+
+A third iteration of the *bsrt* implementation: this time entirely scrapping
+the bucket header, opting to delegate this behaviour to the lookup-table itself.
+I.e. the lookup table no long points to the start of a bucket (as this can be
+inferred later), but rather points to the last element of the bucket.
+Additionally, this implementation scraps the insertion-sort performed in the
+bucket sort, delegating this task to the merge stage of the sorting algorithm.
+
+Pros:
+* Constant-time bucket sort (406 cycles)
+* More optimally used LUT
+* No bookkeeping during bucket-sort
+
+Cons:
+* Non-constant-time merge
+* Optimized LUT requires marginally more arithmetic operations over other variants
+
+**Average cycle count: N/A (not fully implemented)**
+
+
+### bsrt2.uc
+
+A fourth iteration developed in parallel with *sort3* as a proof-of-concept of
+the recently designed lookup-table replacement for hashing. This version is an
+almost direct copy of *bsrt*, except without a LUT, a K1 jump-table and absolute
+sizes in the bucket headers (as opposed to a pointer to the last element of a
+bucket).
+
+Pros:
+* Highly optimized K1 jump-table
+* Subroutined jump-table for possible reuse elsewhere
+* 13 elements per bucket (+1 over *bsrt*)
+* Low instruction count
+
+Cons:
+* Only sorts one element per iteration
+
+**Average cycle count: 1050**
+
+
+### sort4.uc
+
+A fifth iteration of the common bucketsort algorithm. This one is, as its name
+implies, based on the *sort2* algorithm. It improves upon it by making heavy
+use of the `call` and `ret` instructions. I.e. by moving the entire bucketsort
+implementation to a subroutine, it effectively allows 16 calls to sort values
+per iteration of the outermost loop. This conversion to a subroutine is done at
+zero cost to performance, as it exploits the fact that a call to a jumptable-
+based hashing algorithm is made and rather than returning to the algorithm
+after the hashing has taken place, the hash-table simply jumps to the
+insertionsort immediately, after which a `ret` is used to return back to normal
+execution. This has the effect of reducing a bucketsort of a single value be
+two instructions to dereference the value to sort, during which a call to the
+subroutine is made and after sorting, it continues execution at the instruction
+immediately after the dereference.
+
+Pros:
+* Optimized K1 jump-table with inline bucketsort
+* Highly un-rollable bucketsort implementation
+* Highly efficient bus use (almost always saturated)
+* Dynamic bucket placement, allowing for very fast merge operations
+* Efficient use of general registers to reduce arithmetic operations requiring
+constant values
+
+Cons:
+* Maximum of 6 elements per bucket
+* 96 unused program-memory addresses (+1 per bucket)
+
+**Average cycle count: 758**
--- a/bucksort.cpp
+++ b/bucksort.cpp
@ -0,0 +1,172 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+
+#define BUCKETS 0b1000
+#define BUCKET_SIZE 20
+#define LENGTH 32
+
+int count = 0;
+bool LT(int a, int b)
+{
+	count++;
+	return a < b;
+}
+
+// TODO: parallellize rotate-left instruction (BSL)
+// This can be done by partially unrolling the loop ;)
+// Bucket/Insertion sort (except final merge). C-style pseudocode
+void bi_sort(short *data){
+    short *buckets;
+    short pc_, ir_, ar, gr, hr, lc_, ir_;
+
+    for(pc_ = 0xE0; pc_ != 0; ++pc_){
+        // Fetch and hash
+        ar = data[pc_];
+        gr = data[pc_];
+        
+        ar = ar >> 9;
+        ar = ar & 0b1111000;
+        hr = ar;
+
+        ar = buckets[ar];    // Index 0 is pointer to end of array
+        
+        ++ar;
+        buckets[hr] = ar;    // Increment pointer (since we're inserting a new value)
+        --ar;
+
+        ir_ = pc_;           // Push pc
+        pc_ = ar;            // Copy start index to pc (for fast indexing)
+        ar = ar - hr;        // Compute length
+
+        lc_ = ar;            // Loop AR-times
+
+        --lc_;
+
+        while(lc_ != 0){
+            ++pc_;
+            ar = data[pc_];
+            if((ar -= gr) < 0){
+                ar += gr;                
+                data[pc_] = gr;
+                // Insert here
+                while(lc_ != 0){
+                    ++pc_;
+                    gr = data[pc_];
+                    data[pc_] = ar;
+                    ar = gr;
+                    --lc_;
+                }
+
+                goto END;
+            }
+            --lc_;
+        }
+
+        ++pc_;
+
+        data[pc_] = gr;      // GR was the biggest. Insert it at the end
+
+END:
+        pc_ = ir_;           // Pop pc
+    }
+}
+
+void sort(short *data, int length)
+{
+	short buckets[BUCKETS][BUCKET_SIZE] = {};
+	short a_, pc, c, d;
+
+	// Bucketsort
+	for (a_ = 0; a_ < length; a_++)
+	{
+		c = data[a_];
+		d = (c >> 13) & 0b111;
+		pc = buckets[d][0];
+		pc++;
+		buckets[d][0] = pc;
+		buckets[d][pc] = c;
+	}
+
+	for (short q = 0; q < BUCKETS; q++)
+	{
+		short *curr = buckets[q];
+		int length = curr[0];
+		printf("buck: %d\tlength: %hd\n", q, curr[0]);
+		for (short i = 0; i < length; i++)
+		{
+			printf("%hd, ", curr[i + 1]);
+		}
+		printf("\n");
+	}
+#if 0
+i ← 1
+while i < length(A)
+	x ← A[i]
+	j ← i - 1
+	while j >= 0 and A[j] > x
+		A[j+1] ← A[j]
+		j ← j - 1
+	end while
+	A[j+1] ← x
+	i ← i + 1
+end while
+#endif
+	
+	// Insertion Sort
+	for (short q = 0; q < BUCKETS; q++)
+	{
+		short length = buckets[q][0];
+		short *curr = buckets[q] + 1;
+		a_ = 1;
+		while (a_ < length)
+		{
+			c = curr[a_];
+			pc = a_ - 1;
+			while (pc >= 0 && LT(curr[pc], c))
+			{
+				curr[pc+1] = curr[pc];
+				pc--;
+			}
+			curr[pc+1] = c;
+			a_++;
+		}
+	}
+
+	// Merge the buckets
+	pc = 0;
+	int h = 0b100;
+	for (short q = 0; q < BUCKETS; q++)
+	{
+		short *curr = buckets[h];
+		h = (h + 1) & 0b111;
+		a_ = curr[0];
+		while (a_ >= 1)
+		{
+			data[pc] = curr[a_];
+			pc++;
+			a_--;
+		}
+	}
+}
+
+int main(int *argc, char **argv)
+{
+	short data[LENGTH] = {};
+
+	srand(clock());
+	for (int i = 0; i < LENGTH; i++)
+	{
+		data[i] = rand() % 0xFFFF;
+	}
+
+	sort(data, LENGTH);
+
+	printf("Num compares: %d\n", count);
+	for (int i = 0; i < LENGTH; i++)
+	{
+		printf("%hd, ", data[i] & 0xFFFF);
+	}
+	printf("\n");
+	return 0;
+}
--- a/build.sh
+++ b/build.sh
@ -0,0 +1,121 @@
+#!/bin/bash
+
+MODULE_PATH=$(realpath ${0%/*})
+
+# JARs
+UCOMP=microcompiler.jar
+COMP=compiler.jar
+WEAVER=weaver.jar
+
+# Main classes
+UCOMPMAIN=MicrocompilerKt
+COMPMAIN=CompilerKt
+WEAVERMAIN=WeaverKt
+
+# Random number generator
+RGEN=rand_gen.py
+
+# Intermediate file
+INTER=comp.out
+
+
+usage(){
+    echo -e "Usage:\n\t$0 [TARGET] {TYPE | OUTFILE}\n\t$0 [TARGET] [MICROTARGET] [OUTFILE]\n\t$0 [TARGET] [TYPE] [OUTFILE] [COMBINE]\n\t$0 [TARGET] asm [OUTFILE] [COMBINE] [MICROTARGET]\n\nWhere:\n\tTARGET: main compilation target file\n\tTYPE: either \"asm\" or \"micro\"\n\tOUTFILE: output .mia file\n\tCOMBINE: existing .mia file to combine compilation output with\n\tMICROTARGET: explicit microinstruction target"
+}
+
+path_of(){
+    echo "$MODULE_PATH/$1"
+}
+
+if [ $# -eq 0 ]; then
+    >&2 echo "Not enough arguments!"
+    usage
+    exit
+fi
+
+
+TARGET=$1
+
+case $# in
+    1)
+        TYPE="asm"
+        OUTPUT="build.mia"
+        ;;
+
+    2)
+        if [ "$2" = "asm" ] || [ "$2" = "micro" ] ; then
+            TYPE=$2
+            OUTPUT="build.mia"
+        else
+            TYPE="asm"
+            OUTPUT=$2
+        fi
+        ;;
+
+    3)
+        if [ "$2" = "asm" ] || [ "$2" = "micro" ]; then
+            TYPE=$2
+            OUTPUT=$3
+        else
+            TYPE="asm"
+            OUTPUT=$2
+            MICRO=$3
+        fi
+        ;;
+
+    4)
+        TYPE=$2
+        OUTPUT=$3
+        COMBINE=$4
+        ;;
+
+    5)
+        if [ "$2" != "asm" ]; then
+            >&2 echo "TYPE must be asm"
+            usage
+            exit
+        fi
+
+        TYPE=$2
+        OUTPUT=$3
+        COMBINE=$4
+        MICRO=$5
+        ;;
+
+    *)
+        >&2 echo "Too many arguments!"
+        usage
+        exit
+        ;;
+esac
+
+(cd $MODULE_PATH && make all)
+
+if [ "$TYPE" = "asm" ]; then
+    KJAR=$COMP
+    KCLASS=$COMPMAIN
+else
+    KJAR=$UCOMP
+    KCLASS=$UCOMPMAIN
+fi
+
+kotlin -classpath $(path_of $KJAR) $KCLASS $TARGET > $INTER || exit
+python "$(path_of $RGEN)" >> $INTER
+
+if [ "$COMBINE" = "" ]; then
+    if [ "$MICRO" != "" ]; then
+        kotlin -classpath $(path_of $UCOMP) $UCOMPMAIN $MICRO >> $INTER || exit
+    fi
+
+    kotlin -classpath $(path_of $WEAVER) $WEAVERMAIN $INTER $OUTPUT || exit
+else
+    if [ "$MICRO" != "" ]; then
+        kotlin -classpath $(path_of $UCOMP) $UCOMPMAIN $MICRO >> $INTER || exit
+    fi
+    kotlin -classpath $(path_of $WEAVERJAR) $WEAVERMAIN $INTER $COMBINE $OUTPUT || exit
+fi
+
+# Remove intermediate compilation file
+rm -f $INTER
+
+echo "Compiled successfully to $OUTPUT"
--- a/code.eda
+++ b/code.eda
@ -1,31 +1,39 @@
-AND GR1, 0
+#AND GR1, 0
+#
+#LOAD GR0, *0xFE
+#AND GR0, 0xF
+#STORE GR0, *0xFF
+#ADD GR1, *0xFF
+#
+#LOAD GR0, *0xFE
+#LSR GR0, *4
+#AND GR0, 0xF
+#STORE GR0, *0xFF
+#ADD GR1, *0xFF
+#
+#LOAD GR0, *0xFE
+#LSR GR0, *8
+#AND GR0, 0xF
+#STORE GR0, *0xFF
+#ADD GR1, *0xFF
+#
+#LOAD GR0, *0xFE
+#LSR GR0, *12
+#AND GR0, 0xF
+#STORE GR0, *0xFF
+#ADD GR1, *0xFF
+#
+#STORE GR1, *0xFF
+#
+LOAD GR0, *0x00
+LOAD GR1, *0x00

-LOAD GR0, *0xFE
-AND GR0, 0xF
-STORE GR0, *0xFF
-ADD GR1, *0xFF
+CMP GR0, GR1
+BEQ @TEST

-LOAD GR0, *0xFE
-LSR GR0, *4
-AND GR0, 0xF
-STORE GR0, *0xFF
-ADD GR1, *0xFF
+LOAD GR0, *0xFF

-LOAD GR0, *0xFE
-LSR GR0, *8
-AND GR0, 0xF
-STORE GR0, *0xFF
-ADD GR1, *0xFF
+TEST:
+HALT

-LOAD GR0, *0xFE
-LSR GR0, *12
-AND GR0, 0xF
-STORE GR0, *0xFF
-ADD GR1, *0xFF
-
-STORE GR1, *0xFF
-
-HALT GR0, *0
-
-STORE GR1, *0xFF

--- a/compiler.kt
+++ b/compiler.kt
@ -0,0 +1,306 @@
+import java.io.*
+
+enum class OpCode(val opcode: Int, val useM: Boolean, val useADR: Boolean, val useReg: Boolean = true) {
+	LOAD(0, true, true),
+	STORE(1, true, true),
+	ADD(2, true, true),
+	SUB(3, true, true),
+	AND(4, true, true),
+	LSR(5, true, true),
+	BRA(6, false, true),
+	BNE(7, false, true),
+	CMP(8, true, false),
+	BEQ(9, false, true),
+	HALT(15, false, false, false);
+
+    companion object {
+        fun fromString(name: String) = OpCode.values().firstOrNull{ it.name == name }
+    }
+}
+
+enum class Mode(val id: Int) {
+	DIRECT(0), IMMEDIATE(1), INDIRECT(2), INDEXED(3)
+}
+
+abstract class Instruction(val words: Int) {
+	abstract fun getData(insns: Iterable<Instruction>): ShortArray
+}
+
+class Label(val name: String): Instruction(0) {
+	override fun getData(insns: Iterable<Instruction>) = ShortArray(0)
+}
+
+class Operation(val code: OpCode, val reg: Int, val m: Mode, val adr: AddressReference, val immediate: Short? = null): Instruction(if(m == Mode.IMMEDIATE) 2 else 1) {
+	constructor(code: OpCode, reg: Int): this(code, reg, Mode.DIRECT, AddressReference(0)){
+        if(code.useM || code.useADR)
+            throw IllegalArgumentException("Not enough parameters specified for instruction: ${code.name}")
+    }
+    constructor(code: OpCode, m: Mode, adr: AddressReference, immediate: Short? = null): this(code, 0, m, adr, immediate){
+        if(code.useReg)
+            throw IllegalArgumentException("Not enough parameters specified for instruction: ${code.name}")
+    }
+    constructor(code: OpCode): this(code, 0, Mode.DIRECT, AddressReference(0)){
+        if(code.useM || code.useADR || code.useReg)
+            throw IllegalArgumentException("Not enough parameters specified for instruction: ${code.name}")
+    }
+
+    init {
+        if(m == Mode.IMMEDIATE && immediate == null)
+            throw IllegalArgumentException("No immediate argument passed!")
+    }
+	
+	override fun getData(insns: Iterable<Instruction>): ShortArray {
+		val array = ShortArray(words)
+        array[0] = code.opcode
+    			.and(0b1111)
+    			.shl(12)
+    			.or(
+    				if(code.useReg) reg.and(0b11).shl(10)
+    				else 0
+    			)
+    			.or(
+    				if(code.useM) m.id.and(0b11).shl(8)
+    				else 0
+    			)
+    			.or(
+    				if(code.useADR) adr.getAddress(insns).and(0b11111111)
+    				else 0
+    			)
+    			.toShort()
+        if(m == Mode.IMMEDIATE) array[1] = immediate!!
+
+        return array
+	}
+}
+
+class AddressReference(private val label: Label?, private val absolute: Int?) {
+	constructor(label: Label): this(label, null)
+	constructor(absolute: Int): this(null, absolute)
+
+	fun getAddress(insns: Iterable<Instruction>): Int {
+		if(absolute != null) return absolute
+		var addrOff = 0
+        for(insn in insns)
+			if(insn == label) return addrOff
+			else addrOff += insn.words
+		
+        throw RuntimeException("Found reference to undeclared label!")
+	}
+}
+
+class CompilationUnit {
+    private val instructions = ArrayList<Instruction>()
+    private val labels = ArrayList<Label>()
+
+    fun declareLabel(name: String) {
+        if(instructions.firstOrNull{ it is Label && it.name == name } != null)
+            throw IllegalStateException("Attempt to declare the same label twice!")
+        registerInstruction(getLabel(name))
+    }
+    
+    fun getLabel(name: String): Label {
+        if(labels.firstOrNull{ it.name == name } == null)
+            labels.add(Label(name))
+        return labels.first{ it.name == name }
+    }
+
+    fun registerInstruction(insn: Instruction){
+        instructions.add(insn)
+    }
+
+    fun compile(): ShortArray {
+        var dat = 0
+        for(insn in instructions)
+        	dat += insn.words
+        
+        if(dat > 256)
+            throw RuntimeException("Instruction overflow")
+    
+        val rawData = ShortArray(dat)
+        var index = 0
+    
+        for(insn in instructions)
+            for(short in insn.getData(instructions))
+                rawData[index++] = short
+    
+        return rawData
+    }
+}
+
+enum class ArgType {
+    REG, LABEL, INDEX, NUMBER
+}
+
+fun parseRegister(arg: String): Int? {
+    if(!arg.toUpperCase().startsWith("GR")) return null
+    try{
+        val reg = Integer.parseInt(arg.substring(2))
+        if(reg > 3 || reg < 0)
+            throw IllegalArgumentException("Register index out of range!")
+        return reg
+    }catch(e: NumberFormatException){
+        throw IllegalArgumentException("Invalid register value")
+    }
+}
+
+fun parseLabelReference(arg: String, unit: CompilationUnit): Pair<Label, Mode>? {
+    if(arg.length < 2 || Character.isDigit(arg[1])) return null
+    if(arg.startsWith("@")) return unit.getLabel(arg.substring(1)) to Mode.DIRECT
+    if(arg.startsWith("*")) return unit.getLabel(arg.substring(1)) to Mode.INDIRECT
+    return null
+}
+
+fun parseIndex(arg: String): Pair<Int, Mode>? {
+    if(arg.startsWith("[") && arg.endsWith("]")){
+        val literal = arg.substring(1, arg.length - 1)
+        return parseNumber(literal) to Mode.INDEXED
+    }
+    return null
+}
+
+fun parseNumber(literal: String): Int {
+    return if(literal.startsWith("0") && literal.length > 1){
+                    when(literal[1]){
+                        'x' -> Integer.parseInt(literal.substring(2), 16)
+                        'b' -> Integer.parseInt(literal.substring(2), 2)
+                        else -> Integer.parseInt(literal.substring(2), 8)
+                    }
+                }else{
+                    Integer.parseInt(literal)
+                }
+}
+
+fun checkProperSize(number: Int, max: Int): Int {
+    if(number > max || number < 0)
+        throw IllegalArgumentException("Parsed value out of range!")
+    return number
+}
+
+fun parseNumberLiteral(arg: String): Pair<Int, Mode>? {
+    if(arg.startsWith("$")) return checkProperSize(parseNumber(arg.substring(1)), 65535) to Mode.IMMEDIATE
+    if(arg.startsWith("*")) return checkProperSize(parseNumber(arg.substring(1)), 255) to Mode.INDIRECT
+    return try{
+        checkProperSize(parseNumber(arg), 255) to Mode.DIRECT
+    }catch(e: NumberFormatException){
+        null
+    }
+}
+
+fun parseInstruction(line: String, unit: CompilationUnit): Instruction? {
+    if(line.length == 0) return null
+    if(line.replace(" ", "").endsWith(":")){
+        val label = line.substring(0, line.length - 1)
+        if(label.toUpperCase().startsWith("GR") || label.startsWith("$") || label.startsWith("*") || label.startsWith("@"))
+            throw IllegalArgumentException("Label uses reserved prefix")
+        return unit.getLabel(label)
+    }
+    val firstSpace = line.indexOf(" ")
+    if(firstSpace == -1){
+        val opcode = OpCode.fromString(line.toUpperCase())
+        if(opcode == null)
+            throw IllegalArgumentException("Unknown instruction: $line")
+        if(opcode.useReg || opcode.useADR || opcode.useM)
+            throw IllegalArgumentException("Not enough arguments passed to $line")
+        return Operation(opcode)
+    }else{
+        val opcode = OpCode.fromString(line.substring(0, firstSpace).toUpperCase())
+        if(opcode == null)
+            throw IllegalArgumentException("Unknown instruction: $line")
+        val args = line.substring(firstSpace + 1).replace(" ", "").split(",").toTypedArray()
+        when(args.size){
+            1 -> {
+                var arg = parseArguments(args[0], null, unit)
+                return when(arg.first.first){
+                    ArgType.REG -> Operation(opcode, arg.first.second as Int)
+                    ArgType.LABEL -> Operation(opcode, (arg.first.second as Pair<Label, Mode>).second, AddressReference((arg.first.second as Pair<Label, Mode>).first))
+                    ArgType.INDEX -> Operation(opcode, (arg.first.second as Pair<Int, Mode>).second, AddressReference((arg.first.second as Pair<Int, Mode>).first))
+                    ArgType.NUMBER -> if((arg.first.second as Pair<Int, Mode>).second == Mode.IMMEDIATE)
+                                            Operation(opcode, (arg.first.second as Pair<Int, Mode>).second, AddressReference(0), (arg.first.second as Pair<Int, Mode>).first.toShort())
+                                        else Operation(opcode, (arg.first.second as Pair<Int, Mode>).second, AddressReference((arg.first.second as Pair<Int, Mode>).first))
+                }
+            }
+            2 -> {
+                val arg = parseArguments(args[0], args[1], unit)
+                val first = arg.first
+                val second = arg.second!!
+                
+                if(first.first != ArgType.REG)
+                    throw IllegalArgumentException("First argument must be a register")
+
+                return when(second.first){
+                    ArgType.REG -> Operation(opcode, first.second as Int, Mode.values()[second.second as Int], AddressReference(0), 0)
+                    ArgType.LABEL -> Operation(opcode, first.second as Int, (second.second as Pair<Label, Mode>).second, AddressReference((second.second as Pair<Label, Mode>).first))
+                    ArgType.NUMBER -> if((second.second as Pair<Int, Mode>).second == Mode.IMMEDIATE)
+                                            Operation(opcode, first.second as Int, (second.second as Pair<Int, Mode>).second, AddressReference(0), (second.second as Pair<Int, Mode>).first.toShort())
+                                        else Operation(opcode, first.second as Int, (second.second as Pair<Int, Mode>).second, AddressReference((second.second as Pair<Int, Mode>).first))
+                    ArgType.INDEX -> Operation(opcode, first.second as Int, (second.second as Pair<Int, Mode>).second, AddressReference((second.second as Pair<Int, Mode>).first))
+                }
+            }
+            else -> throw IllegalArgumentException("Too many arguments specified")
+        }
+    }
+}
+
+fun parseArgument(arg: String, unit: CompilationUnit): Pair<ArgType, Any>? {
+    var resolve: Any? = parseRegister(arg)
+    if(resolve != null)  return ArgType.REG to resolve
+    resolve = parseIndex(arg)
+    if(resolve != null)  return ArgType.INDEX to resolve
+    resolve = parseLabelReference(arg, unit)
+    if(resolve != null)  return ArgType.LABEL to resolve
+    resolve = parseNumberLiteral(arg)
+    if(resolve != null)  return ArgType.NUMBER to resolve
+
+    return null // Unresolved
+}
+
+fun parseArguments(arg0: String, arg1: String?, unit: CompilationUnit): Pair<Pair<ArgType, Any>, Pair<ArgType, Any>?> {
+    val resolve0 = parseArgument(arg0, unit)
+    if(resolve0 == null) throw IllegalArgumentException("Invalid argument: $arg0")
+    
+    val resolve1 = if(arg1 != null) parseArgument(arg1, unit) else null
+    if(arg1 != null && resolve1 == null) throw IllegalArgumentException("Invalid argument: $arg0")
+    
+    return resolve0 to resolve1
+}
+
+fun parseInstructions(fileData: String): ShortArray {
+    val unit = CompilationUnit()
+    val lines = fileData.replace("\r", "").replace("\t", "").split('\n').toTypedArray()
+    for(index in lines.indices){
+        val commentIndex = lines[index].indexOf("#")
+        if(commentIndex > 0)
+            lines[index] = lines[index].substring(0, commentIndex)
+        try{
+            val insn = parseInstruction(lines[index], unit)
+            if(insn != null) unit.registerInstruction(insn)
+        }catch(e: Exception){
+            print("An error occurred when compiling (line $index)")
+            val message = e.message
+            if(message != null) print(":\n\t$message")
+            println()
+        }
+    }
+    return unit.compile()
+}
+
+fun main(args: Array<String>){
+	// Ensure correct argument length 'n stuff
+	if(args.size != 1){
+		System.err.println("Invalid argument length!")
+		System.exit(-1)
+	}
+
+	val file = File(args[0])
+
+	// Make sure we have a valid file
+	if(!file.isFile){
+		System.err.println("Given file doesn't exist!")
+		System.exit(-2)
+	}
+    for(insn in parseInstructions(file.readText()))
+        println(insn.toUHex())
+}
+
+
+fun Short.toUHex() = toInt().and(0xFFFF.toInt()).or(1.shl(30)).toString(16).substring(4)
--- a/compiler.py
+++ b/compiler.py
@ -3,14 +3,16 @@
 # Compiles ED-Assembly to Machinecode
 #

-# LOAD    GRx, M, ADR
-# STORE   GRx, M, ADR
-# ADD     GRx, M, ADR
-# SUB     GRx, M, ADR
-# AND     GRx, M, ADR
-# LSR     GRx, M, Y # STEG
+# LOAD    GRx, ADR
+# STORE   GRx, ADR
+# ADD     GRx, ADR
+# SUB     GRx, ADR
+# AND     GRx, ADR
+# LSR     GRx, Y # STEG
 # BRA     ADR
 # BNE     ADR
+# BEQ     ADR
+# CMP     GRx, GRy

 #
 #   One machine instruction
@ -125,6 +127,10 @@ OPCODE_TABLE = {
    "LSR"   : 0b0101,
    "BRA"   : 0b0110,
    "BNE"   : 0b0111,
+    "CMP"   : 0b1000,
+    "BEQ"   : 0b1001,
+    "BLT"   : 0b1010,
+    "CMI"   : 0b1011,
    "HALT"  : 0b1111,
 }

@ -229,6 +235,34 @@ def parse_instruction(args):
    """ 
    Parse an assembly instruction into an instruction object.
    """
+    if args[0] in ["HALT"]:
+        return Instruction(OPCODE_TABLE[args[0]], 0, 0, 0)
+    if args[0] in ["LOAD", "STORE", "ADD", "SUB", "AND", "LSR"]:
+        return parse_tripple_instuction(args)
+    if args[0] in ["BRA", "BNE", "BEQ", "BLT"]:
+        return parse_jump(args)
+    if args[0] in ["CMP"]:
+        return parse_dual_registers(args)
+    if args[0] in ["CMI"]:
+        return parse_compare_imediate(args)
+    else:
+        print("Foregotten instruction: ", args[0])
+        assert False
+
+def parse_compare_imediate(args):
+    if len(args) < 3:
+        raise SyntaxError("Not enough arguments.")
+    if len(args) > 3:
+        raise SyntaxError("Trash at end of line \"{}\"".format(" ".join(args[3:])))
+    opcode = OPCODE_TABLE[args[0]]
+    if args[1].upper() not in REGISTERS:
+        raise SyntaxError("Invalid register name" + args[1])
+    register = REGISTERS[args[1].upper()]
+    address, mode = parse_operand(args[2])
+    return Instruction(opcode, register, ADDRESS_MODES["DIRECT"], address)
+    
+
+def parse_tripple_instuction(args):
    if len(args) < 3:
        raise SyntaxError("Not enough arguments.")
    if len(args) > 3:
@ -240,6 +274,26 @@ def parse_instruction(args):
    address, mode = parse_operand(args[2])
    return Instruction(opcode, register, mode, address)

+def parse_jump(args):
+    if len(args) < 2:
+        raise SyntaxError("Not enough arguments.")
+    if len(args) > 2:
+        raise SyntaxError("Trash at end of line \"{}\"".format(" ".join(args[3:])))
+    opcode = OPCODE_TABLE[args[0]]
+    address, mode = parse_operand(args[1])
+    return Instruction(opcode, 0, ADDRESS_MODES["DIRECT"], address)
+
+def parse_dual_registers(args):
+    if len(args) < 3:
+        raise SyntaxError("Not enough arguments.")
+    if len(args) > 3:
+        raise SyntaxError("Trash at end of line \"{}\"".format(" ".join(args[3:])))
+    opcode = OPCODE_TABLE[args[0]]
+    if args[1].upper() not in REGISTERS:
+        raise SyntaxError("Invalid register name" + args[1])
+    register_a = REGISTERS[args[1].upper()]
+    register_b = REGISTERS[args[2].upper()]
+    return Instruction(opcode, register_a, register_b, 0)

 def main():
    success = True
--- a/labb3.mia
+++ b/labb3.mia
@ -1,31 +1,31 @@
 PM:
-00: 4500
-01: 0000
-02: 00fe
-03: 4100
-04: 000f
-05: 10ff
-06: 24ff
-07: 00fe
-08: 5004
-09: 4100
-0a: 000f
-0b: 10ff
-0c: 24ff
-0d: 00fe
-0e: 5008
-0f: 4100
-10: 000f
-11: 10ff
-12: 24ff
-13: 00fe
-14: 500c
-15: 4100
-16: 000f
-17: 10ff
-18: 24ff
-19: 14ff
-1a: f000
+00: 00df
+01: 2100
+02: 0001
+03: b0ff
+04: 9014
+05: 1015
+06: 0a15
+07: 1415
+08: 2500
+09: 0001
+0a: 1417
+0b: 0e17
+0c: 8b00
+0d: a008
+0e: 1a17
+0f: 1e15
+10: 0a15
+11: b4ff
+12: 7008
+13: 6001
+14: f000
+15: 0020
+16: 0000
+17: 0000
+18: 0000
+19: 00ff
+1a: 0000
 1b: 0000
 1c: 0000
 1d: 0000
@ -272,7 +272,7 @@ MyM:
 0c: 0380000
 0d: 0880000
 0e: 0130180
-0f: 000020f
+0f: 0000780
 10: 0480000
 11: 11c0001
 12: 0980000
@ -286,17 +286,17 @@ MyM:
 1a: 0000299
 1b: 0000400
 1c: 0058180
-1d: 0000000
-1e: 0000000
-1f: 0000000
-20: 0000000
-21: 0000000
-22: 0000000
-23: 0000000
-24: 0000000
-25: 0000000
-26: 0000000
-27: 0000000
+1d: 0380000
+1e: 0b84180
+1f: 000041c
+20: 0000180
+21: 000049c
+22: 0000180
+23: 03c00ff
+24: 0c40000
+25: 0128000
+26: 0380000
+27: 0b40180
 28: 0000000
 29: 0000000
 2a: 0000000
@ -395,10 +395,10 @@ K1:
 05: 17
 06: 1c
 07: 1b
-08: 00
-09: 00
-0a: 00
-0b: 00
+08: 1d
+09: 1f
+0a: 21
+0b: 23
 0c: 00
 0d: 00
 0e: 00
@ -411,22 +411,22 @@ K2:
 03: 07

 PC:
-1b
-
-ASR:
 00

-AR:
-001a
+ASR:
+01

-HR:
+AR:
 0000

+HR:
+0001
+
 GR0:
-000a
+0001

 GR1:
-002e
+0000

 GR2:
 0000
@ -435,16 +435,16 @@ GR3:
 0000

 IR:
-f000
+b001

 MyPC:
-0f
+00

 SMyPC:
 00

 LC:
-ff
+00

 O_flag:

@ -455,4 +455,4 @@ N_flag:
 Z_flag:

 L_flag:
-End_of_dump_file
+End_of_dump_file
--- a/microcompiler.kt
+++ b/microcompiler.kt
@ -0,0 +1,620 @@
+enum class ALU(val value: Int, val parameters: Int = 1) {
+    NOP(0b0000, 0), // No operation
+    MOV(0b0001),    // Move from bus to AR
+    MVN(0b0010),    // Move inverse of bus to AR
+    MVZ(0b0011, 0), // Set AR to zero
+    ADD(0b0100),    // Add bus to AR
+    SUB(0b0101),    // Subtract bus from AR
+    AND(0b0110),    // Bitwise AND with bus and AR
+    ORR(0b0111),    // Bitwise OR with bus and AR
+    ADN(0b1000),    // Add bus to AR without setting flags
+    LSL(0b1001, 0), // Logical shift left AR
+    ISL(0b1010, 0), // Shift contents of (AR and HR) left (32-bit shift)
+    ASR(0b1011, 0), // Arithmetic shift right
+    ISR(0b1100, 0), // Signed big shift right
+    LSR(0b1101, 0), // Logical shift right AR
+    ROL(0b1110, 0), // Rotate AR left
+    IRL(0b1111, 0); // Rotate ARHR left (32-bit rotate)
+
+    companion object {
+        fun locate(value: Int) = values().first{ it.value == value }
+        fun matchName(name: String) = values().firstOrNull{ it.name.toLowerCase() == name.toLowerCase() }
+    }
+}
+
+
+enum class ToBus(val value: Int) {
+    NONE(0b000),
+    IR(0b001),
+    PM(0b010),
+    PC(0b011),
+    AR(0b100),
+    HR(0b101),
+    GR(0b110),
+    CONST(0b111);
+
+    companion object {
+        fun locate(value: Int) = values().first{ it.value == value }
+        fun matchName(name: String) = values().first{ it.name.toLowerCase() == name.toLowerCase() }
+    }
+}
+
+enum class FromBus(val value: Int) {
+    NONE(0b000),
+    IR(0b001),
+    PM(0b010),
+    PC(0b011),
+    // No AR
+    HR(0b101),
+    GR(0b110),
+    ASR(0b111);
+
+    companion object {
+        fun locate(value: Int) = values().first{ it.value == value }
+    }
+}
+
+enum class LoopCounter(val value: Int) {
+    NONE(0b00),
+
+    DEC(0b01),
+    BUS(0b10),
+    U(0b11);
+
+    companion object {
+        fun locate(value: Int) = values().first{ it.value == value }
+    }
+}
+
+enum class SEQ(val value: Int) {
+    INC(0b0000),
+    BOP(0b0001), // Branch on opcode
+    BAM(0b0010), // Branch on addressing mode
+    BST(0b0011), // Branch to start (set uPC = 0)
+    CAL(0b0110),
+    RET(0b0111),
+    BRA(0b0101),
+    BNZ(0b0100),
+    BRZ(0b1000),
+    BRN(0b1001),
+    BRC(0b1010),
+    BRO(0b1011),
+    BLS(0b1100),
+    BNC(0b1101),
+    BNO(0b1110),
+    HALT(0b1111);
+
+    companion object {
+        fun locate(value: Int) = values().first{ it.value == value }
+        fun matchName(name: String) = values().firstOrNull{ it.name.toLowerCase() == name.toLowerCase() }
+    }
+}
+
+open class MicroInstruction(private val _compiledValue: Int, val isConstInstr: Boolean = false) {
+    constructor(
+            aluOP: ALU,
+            toBus: ToBus,
+            fromBus: FromBus,
+            muxControl: Boolean,
+            increment: Boolean,
+            lc: LoopCounter,
+            seq: SEQ
+    ): this((
+            (aluOP.value shl 21) or
+            (toBus.value shl 18) or
+            (fromBus.value shl 15) or
+            (muxControl.toBit() shl 14) or
+            (increment.toBit() shl 13) or
+            (lc.value shl 11) or
+            (seq.value shl 7)
+        ) and (-1 ushr 7))
+
+    constructor(aluOP: ALU, toBus: ToBus, setM: Boolean = false): this(aluOP, toBus, FromBus.NONE, setM, false, LoopCounter.NONE, SEQ.INC)
+    constructor(aluOP: ALU, constant: Int): this((aluOP.value shl 21) or (ToBus.CONST.value shl 18) or (constant onlyBits 16), true)
+    constructor(aluOP: ALU): this(aluOP, 0)
+    constructor(toBus: ToBus, fromBus: FromBus, setM: Boolean = false): this(ALU.NOP, toBus, fromBus, setM, false, LoopCounter.NONE, SEQ.INC)
+    constructor(toBus: ToBus, setM: Boolean = false): this(ALU.NOP, toBus, FromBus.NONE, setM, false, LoopCounter.BUS, SEQ.INC)
+    
+    var addressReference: AddressReference? = null
+
+    val aluOP: ALU
+        get() = ALU.locate(_compiledValue ushr 21 onlyBits 4)
+
+    val toBus: ToBus
+        get() = ToBus.locate(_compiledValue ushr 18 onlyBits 2)
+
+    val fromBus: FromBus
+        get() = FromBus.locate(_compiledValue ushr 15 onlyBits 2)
+    
+    val muxControl = _compiledValue.getBitAt(14)
+    val increment = _compiledValue.getBitAt(13)
+
+    val lc: LoopCounter
+        get() = LoopCounter.locate(_compiledValue ushr 11 onlyBits 2)
+
+    val seq: SEQ
+        get() = SEQ.locate(_compiledValue ushr 7 onlyBits 4)
+
+    val address = _compiledValue onlyBits 7
+
+    val usesBus = isConstInstr || toBus != ToBus.NONE
+    val readsBus = isConstInstr || fromBus != FromBus.NONE
+
+    val needsUADR = isConstInstr || (lc == LoopCounter.U) || (seq.value in 0b0100..0b1110)
+    private val rawValue: Int = _compiledValue onlyBits 25 or (if(addressReference == null || addressReference!!.actualAddress == -1) 0 else addressReference!!.actualAddress)
+    val compiledValue: Int
+        get(){
+            if(addressReference != null && addressReference!!.actualAddress == -1)
+                throw RuntimeException("Unresolved label reference: ${addressReference!!.labelName}")
+            return _compiledValue onlyBits 25 or (if(addressReference == null) 0 else addressReference!!.actualAddress)
+        }
+
+
+    infix fun withReference(ref: AddressReference): MicroInstruction {
+        addressReference = ref
+        return this
+    }
+    
+    infix fun withReference(name: String): MicroInstruction {
+        addressReference = AddressReference.makeReference(name)
+        return this
+    }
+
+    infix fun merge(uInstr: MicroInstruction?): MicroInstruction? {
+        if(uInstr == null) return this
+        if(isConstInstr || uInstr.isConstInstr) throw RuntimeException("CONST-instructions are fundamentally un-parallellizable")
+        if(usesBus && uInstr.usesBus && (toBus != uInstr.toBus)) throw RuntimeException("Instructions cannot share bus (output)") // Instructions can't share bus
+        if(readsBus && uInstr.readsBus && (fromBus != uInstr.fromBus)) throw RuntimeException("Instructions cannot share bus (input)") // Instructions cannot both read from bus to differing outputs
+        if(seq != SEQ.INC && uInstr.seq != SEQ.INC && seq != uInstr.seq) throw RuntimeException("SEQ mismatch") // We can merge an INC with something else, but we cannot merge more than this
+        if(needsUADR && uInstr.needsUADR && address != uInstr.address) throw RuntimeException("uADR mismatch") // Instructions cannot depend on differing uADR values
+
+        val newInstr = MicroInstruction(rawValue or uInstr.rawValue)
+
+        val adrRef = if(needsUADR) addressReference else if(uInstr.needsUADR) uInstr.addressReference else null
+        if(adrRef != null && newInstr.needsUADR) newInstr withReference adrRef
+
+        return newInstr
+    }
+}
+
+enum class Register(val busValue: Int, val canRead: Boolean = true) {
+    ASR(0b111, false),
+    IR(0b001),
+    PM(0b010),
+    PC(0b011),
+    AR(0b100),
+    HR(0b101),
+    GR(0b110);
+
+    companion object {
+        fun lookup(name: String) = if(name.toLowerCase() == "grm") (GR to true) else (values().firstOrNull{ it.name.toLowerCase() == name.toLowerCase() } to false)
+    }
+}
+
+class AddressReference{
+    private val address: Int
+    val labelName: String?
+
+    private constructor(address: Int?, labelName: String?){
+        this.address = address ?: -1
+        _actualAddress = this.address
+        this.labelName = labelName
+        if(address == null && labelName == null) throw RuntimeException("Reference must be accessible by name or address")
+    }
+
+    val actualAddress: Int
+        get() {
+            if(_actualAddress == -1)
+                throw RuntimeException("Unresolved label reference: $labelName")
+            return _actualAddress
+        }
+
+    private var _actualAddress: Int
+
+    private fun checkResolved(){
+        if(_actualAddress != -1) throw RuntimeException("Address already resolved")
+    }
+
+    fun resolveConstant(value: Int){
+        checkResolved()
+        _actualAddress = value
+    }
+
+    fun resolveAddress(target: Int){
+        if(target > 128 || target < 0) throw RuntimeException("Invalid target address")
+        checkResolved()
+        _actualAddress = target
+    }
+
+    companion object {
+        private val registry = ArrayList<AddressReference>()
+        fun makeReference(address: Int) = AddressReference(address, null)
+        fun makeReference(labelName: String): AddressReference {
+            val ref = registry.firstOrNull{ it.labelName == labelName }
+            if(ref != null) return ref
+            val reference = AddressReference(null, labelName)
+            registry.add(reference)
+            return reference
+        }
+    }
+}
+
+fun Boolean.toBit() = if(this) 1 else 0
+fun Int.toInstruction() = or(0x10000000).toString(16).substring(1)
+infix fun Int.onlyBits(bits: Int) = and(-1 ushr (32 - bits))
+fun Int.getBitAt(index: Int) = ushr(index).and(1) == 1
+
+fun parseMOV(instr: String): MicroInstruction {
+    val args = instr.split(" ")
+    if(args.size != 3) throw RuntimeException("MOV instruction requires two arguments: $instr")
+    if(args[1] == args[2]) throw RuntimeException("Cannot move from register being moved to: $instr")
+
+    val (a, setMA) = Register.lookup(args[1])
+    if(!a!!.canRead) throw RuntimeException("Cannot read from register: $instr")
+    
+    
+    if(args[2] == "lc") return MicroInstruction(ToBus.locate(a.busValue), setMA)
+
+    val (b, setMB) = Register.lookup(args[2])
+    if(a == b!! && setMA != setMB)
+        throw RuntimeException("Cannot directly move data between multiplexed registers!")
+    if(b == Register.AR) return MicroInstruction(ALU.MOV, ToBus.locate(a.busValue), setMA || setMB)
+
+    return MicroInstruction(ToBus.locate(a.busValue), FromBus.locate(b.busValue), setMA || setMB)
+}
+
+fun readNumber(literal: String, max: Int): Int {
+    val number = if(literal.startsWith("0") && literal.length > 1){ // Parse special number
+        if(literal[1] == 'x') Integer.parseInt(literal.substring(2), 16)
+        else if(literal[1] == 'b') Integer.parseInt(literal.substring(2), 2)
+        else Integer.parseInt(literal.substring(1), 8)
+    }else{
+        Integer.parseInt(literal, 10)
+    }
+
+    if(number > max || number < 0) throw RuntimeException("Value out of range: $literal")
+
+    return number
+}
+
+fun parseCONST(instr: String): MicroInstruction {
+    val args = instr.split(" ").toTypedArray()
+    if(args.size != 2) throw RuntimeException("Unexpected arguments: $instr")
+
+    return MicroInstruction(ALU.MOV) withReference try{ AddressReference.makeReference(readNumber(args[1], 65535)) } catch(e: NumberFormatException){ AddressReference.makeReference(args[1]) }
+}
+
+fun parseALU(instr: String): MicroInstruction {
+    val args = instr.split(" ")
+
+    val aluOperation = ALU.matchName(args[0])!!
+
+    if(args.size != (1 + aluOperation.parameters)) throw RuntimeException("Unexpected arguments: $instr")
+
+    if(aluOperation.parameters == 0) return MicroInstruction(aluOperation, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.INC)
+
+    try{
+        val num = readNumber(args[1], 65535)
+        return MicroInstruction(aluOperation, num)
+    }catch(e: NumberFormatException){
+        // NaN
+    }
+
+    val (source, setM) = Register.lookup(args[1])
+    if(source == null) return MicroInstruction(aluOperation) withReference AddressReference.makeReference(args[1])
+    if(!source.canRead) throw RuntimeException("Cannot read from source: $instr")
+
+    return MicroInstruction(aluOperation, ToBus.locate(source.busValue), setM)
+}
+
+fun parseLabelReference(ref: String): AddressReference? {
+    return if(ref.startsWith("@")) AddressReference.makeReference(ref.substring(1)) else null
+}
+
+fun parseCALL(instr: String): MicroInstruction {
+    val args = instr.split(" ")
+    if(args.size != 2) throw RuntimeException("Unexpected arguments: $instr")
+
+    return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.CAL) withReference parseAddressReference(args[1])
+}
+
+fun parseBranch(instr: String): MicroInstruction {
+    val args = instr.split(" ")
+    if(args.size != 2) throw RuntimeException("Unexpected arguments: $instr")
+
+    return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.matchName(args[0])!!) withReference parseAddressReference(args[1])
+}
+
+fun parseLCSet(instr: String): MicroInstruction {
+    val args = instr.split(" ")
+    if(args.size != 2) throw RuntimeException("Unexpected arguments: $instr")
+
+    val ref = try{
+        AddressReference.makeReference(readNumber(args[1], 0xFF))
+    }catch(e: NumberFormatException){
+        AddressReference.makeReference(args[1])
+    }
+
+    return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.U, SEQ.INC) withReference ref
+}
+
+fun parseAddressReference(arg: String): AddressReference {
+    return parseLabelReference(arg) ?: AddressReference.makeReference(readNumber(arg, 0xFF))
+}
+
+fun parseInstruction(line: String): MicroInstruction? {
+    val subInstructions = line.split(";")
+    if(subInstructions.size == 1){
+        var shave = subInstructions[0].toLowerCase()
+        while(shave.startsWith(" ") || shave.startsWith("\t")) shave = shave.substring(1)
+        while(shave.endsWith(" ") || shave.endsWith("\t")) shave = shave.substring(0, shave.length - 1)
+
+        if(shave.length == 0) return null
+
+        if(shave.startsWith("mov ")) return parseMOV(shave)
+        else if(shave.startsWith("const ")) return parseCONST(shave)
+        else if(shave == "incpc") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, true, LoopCounter.NONE, SEQ.INC)
+        else if(ALU.matchName(shave.substring(0, if(shave.indexOf(" ") == -1) shave.length else shave.indexOf(" "))) != null) return parseALU(shave)
+        else if(shave.startsWith("call ")) return parseCALL(shave)
+        else if(shave == "ret") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.RET)
+        else if(shave == "bop") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.BOP)
+        else if(shave == "bam") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.BAM)
+        else if(shave == "bst") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.BST)
+        else if(shave.startsWith("b") && shave.indexOf(" ") != -1 && SEQ.matchName(shave.substring(0, shave.indexOf(" "))) != null) return parseBranch(shave)
+        else if(shave == "halt") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.NONE, SEQ.HALT)
+        else if(shave.startsWith("lcset")) return parseLCSet(shave)
+        else if(shave == "declc") return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.NONE, false, false, LoopCounter.DEC, SEQ.INC)
+        else if(shave.startsWith("reset")){
+            val (resetReg, setM) = Register.lookup(shave.substring(6))
+            return MicroInstruction(ALU.NOP, ToBus.NONE, FromBus.locate(resetReg!!.busValue), setM, false, LoopCounter.NONE, SEQ.INC)
+        }
+        else throw RuntimeException("Unknown instruction: $shave")
+    }else{
+        var result: MicroInstruction? = null
+        for(rawInstruction in subInstructions){
+            val parsed = parseInstruction(rawInstruction)
+            if(parsed == null) continue
+            result = parsed merge result
+            if(result == null) throw RuntimeException("Instructions ($line) could not be merged!")
+        }
+        return result!!
+    }
+}
+
+fun error(message: String, line: Int){
+    System.err.println("Error on line $line:\n\t$message")
+    System.exit(1)
+}
+
+fun main(args: Array<String>){
+    if(args.size != 1) throw RuntimeException("Bad arguments :(")
+    val file = java.io.File(args[0])
+
+    if(!file.isFile) throw RuntimeException("File not found :(")
+    val builder = StringBuilder("@u\n")
+    var currentLine = 0
+    var lineCount = 0
+    val insns = ArrayList<MicroInstruction>()
+    val k1 = HashMap<Int, AddressReference>()
+    val k2 = HashMap<Int, AddressReference>()
+
+    var emitting = false
+    val emitter = ArrayList<String>()
+
+    val rawData = file.readText().replace("\r", "").split("\n")
+    val pregen = StringBuilder()
+
+    fun emitCode(){
+          val emission = StringBuilder()
+          for(line in emitter)
+              emission.append(line).append("\n")
+          
+          val (output, error) = emission.toString().runPython()
+          if(error.length > 0)
+              throw RuntimeException("Error when attemping to emit code:\n    ${error.replace("\n", "\n    ")}")
+
+          pregen.append(output)
+
+          emitter.clear()
+
+    }
+
+    // Process code emission first
+    for(line in rawData){
+        if(line == "#emit") emitting = true
+        else{
+            if(emitting){
+                if(line.startsWith(">")){
+                    emitter.add(line.substring(1))
+                    continue
+                }
+                else{
+                    emitCode()
+                    emitting = false
+                }
+            }
+
+            pregen.append(line).append("\n")
+        }
+    }
+
+    if(emitter.size > 0) emitCode()
+
+    println("@p")
+    for(line in pregen.toString().replace("\r", "").split("\n")){
+        ++lineCount
+        try{
+            val actualCode = (if(line.indexOf("//") != -1) line.substring(0, line.indexOf("//")) else line).toLowerCase()
+            if(actualCode.length == 0) continue
+    
+            if(actualCode.startsWith("$")){
+                AddressReference.makeReference(actualCode.substring(1)).resolveAddress(currentLine)
+                continue
+            }
+    
+            // Define compile-time constant
+            if(actualCode.startsWith("#define ")){
+                val substr = actualCode.substring(8)
+                if(substr.indexOf(" ") == -1) throw RuntimeException("Bad compile-time constant definition: $actualCode")
+                val name = substr.substring(0, substr.indexOf(" "))
+                val value = substr.substring(substr.indexOf(" ") + 1).replace(" ", "").replace("\t", "")
+                try{
+                    val constant = readNumber(value, 65535)
+                    
+                    AddressReference.makeReference(name).resolveConstant(constant)
+                    continue
+                }catch(e: NumberFormatException){
+                    throw RuntimeException("Cannot parse constant: $actualCode")
+                }
+            }
+
+            // Define Program-Memory constants
+            if(actualCode.startsWith("#data ")){
+                fun err(): Nothing = throw RuntimeException("Bad program-memory state definition: $actualCode")
+                val substr = actualCode.substring(6)
+                if(substr.indexOf(" ") == -1) err()
+
+                val address = substr.substring(0, substr.indexOf(" "))
+                val constant = substr.substring(substr.indexOf(" ") + 1).replace(" ", "").replace("\t", "")
+
+                try{
+                    println("@0x${readNumber(address, 255).toString(16)}\n${readNumber(constant, 65535).or(1 shl 30).toString(16).substring(4)}")
+                }catch(e: NumberFormatException){
+                    err()
+                }
+                continue
+            }
+
+            // Define optable (k1) entry
+            if(actualCode.startsWith("#optable ")){
+                // Parse K1 declaration
+                val substr = actualCode.substring(9)
+                if(substr.indexOf(" ") == -1) throw RuntimeException("Bad optable declaration: $actualCode")
+
+                val address = substr.substring(0, substr.indexOf(" "))
+                val constant = substr.substring(substr.indexOf(" ") + 1).replace(" ", "").replace("\t", "")
+
+                try{
+                    k1[readNumber(address, 0xF)] =
+                        if(constant.startsWith("@")) AddressReference.makeReference(constant.substring(1))
+                        else AddressReference.makeReference(readNumber(constant, 127))
+                }catch(e: NumberFormatException){
+                    throw RuntimeException("Unknown number format for optable declaration: $actualCode")
+                }
+                continue
+            }
+
+            // Define addressing mode (k2) entry
+            if(actualCode.startsWith("#amode ")){
+                // Parse K2 declarations
+                val substr = actualCode.substring(7)
+                if(substr.indexOf(" ") == -1) throw RuntimeException("Bad amode declaration: $actualCode")
+
+                val address = substr.substring(0, substr.indexOf(" "))
+                val constant = substr.substring(substr.indexOf(" ") + 1).replace(" ", "").replace("\t", "")
+
+                try{
+                    k2[readNumber(address, 0x3)] =
+                        if(constant.startsWith("@")) AddressReference.makeReference(constant.substring(1))
+                        else AddressReference.makeReference(readNumber(constant, 127))
+                }catch(e: NumberFormatException){
+                    throw RuntimeException("Unknown number format for amode declaration: $actualCode")
+                }
+                continue
+            }
+
+            if(actualCode.startsWith("#pmgen ")){
+                val code = actualCode.substring("#pmgen ".length)
+
+                val (output, error) = "for address in range(0, 256):\n\t$code".runPython()
+                
+                if(error.length > 0)
+                    throw RuntimeException("An error ocurred when running PM-generation script:\n    ${error.replace("\n", "\n    ")}")
+
+                val outStrings = output.split("\n")
+
+                for(directive in outStrings){
+                    if(directive.length == 0) continue
+                    val split = directive.split(" ")
+                    if(split.size != 2)
+                        throw RuntimeException("Unknown PM mapping: $directive")
+
+                    val address: Int
+                    val value: Int
+                    try{
+                        address = split[0].toInt()
+                        if(address < 0 || address > 255)
+                            throw RuntimeException("Could not parse address: $directive")
+                    }catch(e: NumberFormatException){
+                        throw RuntimeException("Could not parse address: $directive")
+                    }
+                    
+                    try{
+                        value = split[1].toInt()
+                        if(value < -32768 || address > 65535)
+                            throw RuntimeException("Could not parse value: $directive")
+                    }catch(e: NumberFormatException){
+                        throw RuntimeException("Could not parse value: $directive")
+                    }
+
+                    // TODO: Fix this. This works, but it's absolutely terrible
+                    println("@0x${address.toString(16)}\n${value.and(0xFFFF).or(1 shl 30).toString(16).substring(4)}")
+                }
+                continue
+            }
+    
+            val instr = parseInstruction(actualCode)
+            if(instr == null) continue // Blank line
+            insns.add(instr)
+    
+            ++currentLine
+        }catch(e: Throwable){
+            error(e.message ?: "Compilation failed with an unknown error", lineCount)
+        }
+    }
+    if(currentLine > 127){
+        System.err.println("Instruction count overflow: $currentLine instructions")
+        System.exit(1)
+    }
+
+    // Convert instructions to strings
+    // Also resolve address references here
+    for(instr in insns)
+        builder.append(instr.compiledValue.toInstruction()).append("\n")
+    
+    // Write K-table data ahead of instructions
+    for(k1Pair in k1)
+        println("@k1${k1Pair.key.toString(16)}${k1Pair.value.actualAddress.toPaddedHexString(2)}")
+
+    for(k2Pair in k2)
+        println("@k2${k2Pair.key.toString(16)}${k2Pair.value.actualAddress.toPaddedHexString(2)}")
+    
+    // Write instructions
+    print(builder.toString())
+    
+    // Informational message
+    System.err.println("INFO: Microcompilation succeeded! Instruction count: $currentLine")
+}
+
+
+fun Int.toPaddedHexString(len: Int): String {
+    val str = toString(16)
+    return if(str.length >= len) str else ("0".repeat(len - str.length) + str)
+}
+
+fun String.runPython(): Pair<String, String> {
+     val proc = ProcessBuilder("python", "-c", "$this")
+                    .redirectOutput(ProcessBuilder.Redirect.PIPE)
+                    .redirectError(ProcessBuilder.Redirect.PIPE)
+                    .start()
+
+    // Await termination of pm generator
+    proc.waitFor()
+    return proc.inputStream.readString() to proc.errorStream.readString()
+}
+
+fun java.io.InputStream.readString(): String {
+    val bytes = ByteArray(available())
+    read(bytes, 0, bytes.size)
+    return String(bytes)
+}
--- a/qs.c
+++ b/qs.c
@ -0,0 +1,34 @@
+#include <stdio.h>
+
+void quicksort(int *data, int low, int high){
+    while(low < high){
+        int pivot = data[high];
+        int i = low;
+
+        for(int j = low; j < high; ++j){
+            if(data[j] < pivot){
+                int tmp = data[j];
+                data[j] = data[i];
+                data[i] = tmp;
+                i += 1;
+            }
+        }
+    
+        int tmp = data[high];
+        data[high] = data[i];
+        data[i] = tmp;
+        
+        quicksort(data, low, i - 1);
+        low = i + 1;
+    }
+}
+
+int main(int argc, char ** argv){
+    int data[26] = {1, 2, 0, 3, 5, 9, 8, 7, 6, 11, 10, 4, 5, 6, 7, 9, 4, 6, 8, 7, 2, 4, 1, 12, 40, 25};
+    for(int i = 0; i < 26; ++i)
+        printf("#: %d\n", data[i]);
+
+    quicksort(data, 0, 25);
+    for(int i = 0; i < 26; ++i)
+        printf("%: %d\n", data[i]);
+}
--- a/rand_gen.py
+++ b/rand_gen.py
@ -0,0 +1,9 @@
+import random
+
+def getRandomHex():
+    return hex(random.randint(0, 0x10000) | (1 << 30))[-4:].upper()
+
+if __name__ == "__main__":
+    print("@p\n@0xE0")
+    for i in range(0, 32):
+        print(getRandomHex())
--- a/sorter.eda
+++ b/sorter.eda
@ -0,0 +1,58 @@
+# Beatuifull
+LOAD GR0 0x00DF # I
+
+OUTER:
+ADD GR0 1
+CMI GR0 0xFF
+BEQ @END
+
+STORE GR0 *@I
+LOAD GR2 **@I
+
+LOAD GR1 *@I
+
+INNER:
+ADD GR1 1
+STORE GR1 *@J
+LOAD GR3 **@J
+
+CMP GR2 GR3
+BLT @SKIPP
+
+STORE GR2 **@J
+STORE GR3 **@I
+LOAD  GR2 **@I
+SKIPP:
+CMI GR1 0xFF
+BNE @INNER 
+BRA @OUTER
+
+END:
+HALT
+
+# for (int i = 0; i < length; i++)
+# {
+# 	for (int j = i; j < length;)
+# 	{
+#		j++
+# 		if (data[i] < data[j])
+# 		{
+# 			int a = data[i];
+# 			data[i] = data[j];
+# 			data[j] = a;
+# 		}
+# 	}
+# }
+
+
+
+I:
+.	0020
+J:
+.	0000
+A: # Pointer to value at I
+.   0000
+B: # Pointer to value at J
+.   0000
+LAST:
+.	00FF
--- a/test.gs
+++ b/test.gs
@ -0,0 +1,3 @@
+LOAD GR0, 0xFF # Put FF into GR0
+CMP GR0, GR2
+HALT
--- a/ucode/bsrt.uc
+++ b/ucode/bsrt.uc
@ -0,0 +1,253 @@
+#define LIST_START 0xE0
+#define HASH_MASK 0b1111
+#define BUCKET_SIZE 13
+#define FIRST_BUCKET_ADDRESS 0x10
+#define LAST_BUCKET_ADDRESS  0xE0
+#define LAST_BUCKET_START_ADDR 0xD3
+
+// PM initial state
+#data 0x00 0x78
+#data 0x01 0x85
+#data 0x02 0x92
+#data 0x03 0x9f
+#data 0x04 0xac
+#data 0x05 0xb9
+#data 0x06 0xc6
+#data 0x07 0xd3
+#data 0x08 0x10
+#data 0x09 0x1d
+#data 0x0a 0x2a
+#data 0x0b 0x37
+#data 0x0c 0x44
+#data 0x0d 0x51
+#data 0x0e 0x5e
+#data 0x0f 0x6b
+
+#data 0x10 0x10
+#data 0x1d 0x1d
+#data 0x2a 0x2a
+#data 0x37 0x37
+#data 0x44 0x44
+#data 0x51 0x51
+#data 0x5e 0x5e
+#data 0x6b 0x6b
+#data 0x78 0x78
+#data 0x85 0x85
+#data 0x92 0x92
+#data 0x9f 0x9f
+#data 0xac 0xac
+#data 0xb9 0xb9
+#data 0xc6 0xc6
+#data 0xd3 0xd3
+
+
+// TODO: Include register initial-state compiler directive (saves, like, 2 cycles max)
+// Set PC to LIST_START
+const LIST_START
+mov ar pc
+
+
+//// ---- START OF DANK SORT ---- ////
+$DANK_SORT
+
+mov pc asr
+mov pc ir; incpc
+mov pm ar
+mov pc asr
+mov pm hr
+
+irl; mov hr gr
+irl
+irl
+irl
+and HASH_MASK
+mov ar asr
+mov pm ar; mov pm asr
+mov ar pc; mvn ar
+add 1
+add pm
+
+mov ar lc
+
+mov pc ar
+
+// Move 
+mov pm pc
+incpc
+mov pc pm
+mov ar pc
+
+
+// First insertion
+$DANK_INSERT_FIRST_START
+
+incpc; bls @DANK_INSERT_FIRST_END_BIGGEST
+
+mov pc asr
+mov pm ar
+sub gr
+adn gr; brn @DANK_INSERT_FIRST_BOTTOM
+
+mov gr pm; incpc
+
+$DANK_INSERT_FIRST_SHIFT
+mov pc asr; bls @DANK_INSERT_SECOND_START
+mov pm gr
+mov ar pm
+mov gr ar; declc; incpc; bra @DANK_INSERT_FIRST_SHIFT
+
+$DANK_INSERT_FIRST_BOTTOM
+declc; bra @DANK_INSERT_FIRST_START
+
+$DANK_INSERT_FIRST_END_BIGGEST
+mov pc asr
+mov gr pm
+
+$DANK_INSERT_SECOND_START
+// Finalize hash in HR
+mov hr ar
+and HASH_MASK
+mov ar asr
+mov pm hr
+mov pm asr
+
+mov pm pc; mov pm ar
+incpc; sub hr
+mov pc pm
+
+mov ar lc
+
+mov hr pc
+
+// Get the value we hashed
+mov ir asr
+mov pm gr
+
+// Second insertion
+$DANK_INSERT_SECOND_LOOP
+incpc; bls @DANK_INSERT_SECOND_END_BIGGEST
+
+mov pc asr
+mov pm ar
+sub gr
+adn gr; brn @DANK_INSERT_SECOND_BOTTOM
+
+mov gr pm; incpc
+
+$DANK_INSERT_SECOND_SHIFT
+mov pc asr; bls @DANK_INSERT_SECOND_END_NUMLET
+mov pm gr
+mov ar pm
+mov gr ar; declc; incpc; bra @DANK_INSERT_SECOND_SHIFT
+
+$DANK_INSERT_SECOND_BOTTOM
+declc; bra @DANK_INSERT_SECOND_LOOP
+
+$DANK_INSERT_SECOND_END_BIGGEST
+mov pc asr
+mov gr pm
+
+$DANK_INSERT_SECOND_END_NUMLET
+
+mov ir ar; mov ir pc
+sub 0xFE
+brz @BUCKET_SORT_END; incpc
+bra @DANK_SORT; incpc
+
+
+$BUCKET_SORT_END
+
+
+
+// New merge-algo. Saves about 100 cycles over previous version
+const LAST_BUCKET_START_ADDR
+mov ar hr
+const FIRST_BUCKET_ADDRESS
+mov ar pc
+const LIST_START
+
+$MERGE
+mov pc ir
+mov ar gr
+mov pc asr; incpc
+mov pm ar
+sub ir
+mov ar lc
+mov gr ar
+$MERGE_MOVE
+mov pc asr; bls @MERGE_BOTTOM
+mov pm gr
+mov ar asr
+add 1
+mov gr pm; declc; incpc; bra @MERGE_MOVE
+
+$MERGE_BOTTOM
+mov ar gr
+mov ir ar
+sub hr
+adn hr; brz @MERGE_END
+add BUCKET_SIZE
+mov ar pc
+mov gr ar; bra @MERGE
+
+$MERGE_END
+$BREAK
+halt
+
+
+
+// Old merge-algo
+
+// Set PC to write index
+//const 0xE0
+//mov ar pc
+//
+//
+//// ######## INSERTION ########## 
+//
+//// COPY NEGATIVE
+//
+//// Initialize GR as bucket pointer and IR as element pointer
+//const FIRST_BUCKET_ADDRESS
+//mov ar gr
+//
+//
+//$LOOP
+//// Dereference GR into IR and compute length
+//mov gr asr
+//mov pm ar
+//
+//// Subtract start from end and save (the resultant length) into LC
+//sub gr
+//mov ar lc
+//
+//// Calcuate the first element of the array and store in IR
+//mov gr ar
+//add 1
+//mov ar ir
+//
+//$COPY_BUCKET
+//bls @NEXT_BUCKET
+//
+//mov ir asr; mov ir ar
+//mov pm hr
+//
+//add 1
+//mov ar ir
+//
+//// Write value to be copied into write index (data[pc] = data[ir])
+//mov pc asr
+//mov hr pm; incpc; declc
+//bra @COPY_BUCKET
+//$NEXT_BUCKET
+//
+//
+//mov gr ar
+//add BUCKET_SIZE
+//mov ar gr
+//sub LAST_BUCKET_ADDRESS
+//bnz @LOOP
+//
+//
+//$BREAK
+//halt
--- a/ucode/bsrt2.uc
+++ b/ucode/bsrt2.uc
@ -0,0 +1,267 @@
+#define LIST_START 0xE0
+#define HASH_MASK 0b1111
+#define BUCKET_SIZE 14
+#define BUCKET_ADDRESS 0x00
+#define LAST_BUCKET_ADDRESS  0xE0
+#define LAST_BUCKET_START_ADDR 0xD2
+
+
+#optable 0x0 @OT_0
+#optable 0x1 @OT_1
+#optable 0x2 @OT_2
+#optable 0x3 @OT_3
+#optable 0x4 @OT_4
+#optable 0x5 @OT_5
+#optable 0x6 @OT_6
+#optable 0x7 @OT_7
+#optable 0x8 @OT_8
+#optable 0x9 @OT_9
+#optable 0xa @OT_A
+#optable 0xb @OT_B
+#optable 0xc @OT_C
+#optable 0xd @OT_D
+#optable 0xe @OT_E
+#optable 0xf @OT_F
+
+// PM initial state
+#data 0x00 0x00
+#data 0x0E 0x00
+#data 0x1C 0x00
+#data 0x2A 0x00
+#data 0x38 0x00
+#data 0x46 0x00
+#data 0x54 0x00
+#data 0x62 0x00
+#data 0x70 0x00
+#data 0x7e 0x00
+#data 0x8c 0x00
+#data 0x9a 0x00
+#data 0xa8 0x00
+#data 0xb6 0x00
+#data 0xc4 0x00
+#data 0xd2 0x00
+
+
+// TODO: Include register initial-state compiler directive
+// (saves, like, 2 cycles max)
+// Set PC to LIST_START
+const LIST_START
+mov ar pc
+
+
+//// ---- START OF DANK SORT ---- ////
+$SORT
+
+mov pc asr; incpc
+mov pm ir; call @JTABLE // Call jumptable. This also copies PC to HR
+
+// We return here after visiting jump table
+// AR contains a magic value and ASR contains the same value
+mov pm lc; mov pm pc
+
+// Move
+incpc
+mov pc pm
+mov ar pc
+
+
+// First insertion
+$INSERT_START
+
+// Load a value and compare with value to be inserted
+// If L is set, value is greater than all other values in bucket
+incpc
+mov pc asr; bls @INSERT_END_BIGGEST
+mov pm ar
+sub ir
+adn ir; brn @INSERT_BOTTOM
+
+// We found where to insert value. Insert it
+mov ir pm; incpc
+
+// Shift values after the value we inserted
+$INSERT_SHIFT
+mov pc asr; bls @INSERT_END_NUMLET
+mov pm gr
+mov ar pm
+mov gr ar; declc; incpc; bra @INSERT_SHIFT
+
+// Continue to next value in bucket for comparison
+$INSERT_BOTTOM
+declc; bra @INSERT_START
+
+// Value we are inserting was the biggest value in the bucket
+$INSERT_END_BIGGEST
+mov ir pm
+
+$INSERT_END_NUMLET
+
+// Check if we've sorted all values. If we have, merge then,
+// otherwise return to top of bucketsort
+mov hr ar; mov hr pc
+sub 0
+bnz @SORT
+
+
+$BUCKET_SORT_END
+
+
+const LAST_BUCKET_START_ADDR
+mov ar hr
+const BUCKET_ADDRESS
+mov ar pc
+const LIST_START
+
+$MERGE
+mov pc ir
+mov pc asr; incpc
+mov pm lc
+
+$MERGE_MOVE
+mov pc asr; bls @MERGE_BOTTOM
+mov pm gr
+mov ar asr
+add 1
+mov gr pm; declc; incpc; bra @MERGE_MOVE
+
+$MERGE_BOTTOM
+mov ar gr
+mov ir ar
+sub hr
+adn hr; brz @PROGRAM_END
+adn BUCKET_SIZE
+mov ar pc
+mov gr ar; bra @MERGE
+
+
+$PROGRAM_END
+$BREAK
+halt
+
+
+// Jump-table subroutine
+$JTABLE
+mov pc hr; bop
+
+// IR OP-field jump table
+$OT_0
+const 0x70
+mov ar asr; ret
+
+$OT_1
+const 0x7e
+mov ar asr; ret
+
+$OT_2
+const 0x8c
+mov ar asr; ret
+
+$OT_3
+const 0x9a
+mov ar asr; ret
+
+$OT_4
+const 0xa8
+mov ar asr; ret
+
+$OT_5
+const 0xb6
+mov ar asr; ret
+
+$OT_6
+const 0xc4
+mov ar asr; ret
+
+$OT_7
+const 0xd2
+mov ar asr; ret
+
+$OT_8
+const 0x00
+mov ar asr; ret
+
+$OT_9
+const 0x0e
+mov ar asr; ret
+
+$OT_A
+const 0x1c
+mov ar asr; ret
+
+$OT_B
+const 0x2a
+mov ar asr; ret
+
+$OT_C
+const 0x38
+mov ar asr; ret
+
+$OT_D
+const 0x46
+mov ar asr; ret
+
+$OT_E
+const 0x54
+mov ar asr; ret
+
+$OT_F
+const 0x62
+mov ar asr; ret
+
+
+
+// Old merge-algo
+
+// Set PC to write index
+//const 0xE0
+//mov ar pc
+//
+//
+//// ######## INSERTION ########## 
+//
+//// COPY NEGATIVE
+//
+//// Initialize GR as bucket pointer and IR as element pointer
+//const BUCKET_ADDRESS
+//mov ar gr
+//
+//
+//$LOOP
+//// Dereference GR into IR and compute length
+//mov gr asr
+//mov pm ar
+//
+//// Subtract start from end and save (the resultant length) into LC
+//sub gr
+//mov ar lc
+//
+//// Calcuate the first element of the array and store in IR
+//mov gr ar
+//add 1
+//mov ar ir
+//
+//$COPY_BUCKET
+//bls @NEXT_BUCKET
+//
+//mov ir asr; mov ir ar
+//mov pm hr
+//
+//add 1
+//mov ar ir
+//
+//// Write value to be copied into write index (data[pc] = data[ir])
+//mov pc asr
+//mov hr pm; incpc; declc
+//bra @COPY_BUCKET
+//$NEXT_BUCKET
+//
+//
+//mov gr ar
+//add BUCKET_SIZE
+//mov ar gr
+//sub LAST_BUCKET_ADDRESS
+//bnz @LOOP
+//
+//
+//$BREAK
+//halt
--- a/ucode/bucksort.uc
+++ b/ucode/bucksort.uc
@ -0,0 +1,154 @@
+#define BUCKET_SIZE 0b1000
+#define BUCKET_COUNT 15
+#define LIST_START 0xE0
+#define HASH_MASK 0b1111000
+
+#data 0x00 0x00
+#data 0x08 0x08
+#data 0x10 0x10
+#data 0x18 0x18
+#data 0x20 0x20
+#data 0x28 0x28
+#data 0x30 0x30
+#data 0x38 0x38
+#data 0x40 0x40
+#data 0x48 0x48
+#data 0x50 0x50
+#data 0x58 0x58
+#data 0x60 0x60
+#data 0x68 0x68
+#data 0x70 0x70
+
+
+//// FOR TESTING PURPOSES ONLY ////
+
+// TODO: Buckets store last index instead of length.
+
+//// ---- Bucket sort ----
+
+// Set PC to E0
+const LIST_START
+mov ar pc
+
+$MOVE_TO_BUCKETS
+mov pc asr; mov pc ar;
+add pc; incpc; 
+brz @MOVE_TO_BUCKETS_END 
+mov pm ar; mov pm gr
+
+// The hash
+rol
+rol
+rol
+rol
+rol
+rol
+rol
+and HASH_MASK
+
+// Push PC
+mov pc ir 
+
+// Set current address to the hash.
+mov ar asr
+mov ar pc
+
+// Increase length
+// Compare loop
+//  if a < b : keep looping
+//  else: swap a and b and shift the rest one step
+//  if at end, insert value and return.
+
+
+// Length of loop, and set up PC for first list index
+const 1
+add pm; incpc
+mov ar pm
+sub pc
+// Set loop counter
+mov ar lc
+
+$COMPARE_LOOP
+bls @INSERT_LAST_ELEMENT
+mov pc asr; incpc; declc   
+
+// TODO: Make sure this is optimal
+mov pm ar
+sub gr
+brn @COMPARE_LOOP
+add gr; mov gr pm
+
+mov pc asr; incpc; declc
+// call @BREAK
+
+$INSERT_LAST_ELEMENT
+mov gr pm
+
+// Pop PC
+mov ir pc
+bra @MOVE_TO_BUCKETS
+
+// mov ar asr
+// mov pc ir               // Push PC
+// mov pm pc; add pm
+// incpc
+// mov pc pm
+// mov ar asr
+// mov gr pm
+// mov ir pc               // Pop PC
+// bra @MOVE_TO_BUCKETS
+
+$MOVE_TO_BUCKETS_END
+halt
+
+// //// ---- Bubble sort ----
+// 
+// 
+// $BUBBLE_YEET
+// 
+// mov pm ir; incpc
+// 
+// mov pc asr
+// 
+// 
+// $SORT_LOOP
+// mov pc hr // Current lowest index
+// mov pm gr // Current lowest value
+// 
+// reset asr
+// mov pc pm
+// 
+// $LOOP
+// mov pc asr
+// mov gr ar
+// sub pm
+// 
+// // If PM < GR
+// brn @CHECK
+// mov pc hr // Current lowest index
+// mov pm gr // Current lowest value
+// 
+// $CHECK
+// incpc
+// mov pc ar
+// sub ir
+// bnz @LOOP
+// 
+// reset asr
+// mov pm pc
+// mov pm asr; incpc
+// mov pm ar
+// mov gr pm
+// mov hr asr
+// mov ar pm
+// reset asr
+// mov pc pm
+// declc
+// bls @END_SORT
+// bra @SORT_LOOP
+// 
+// $END_SORT
+// halt
+
+$BREAK
+halt
--- a/ucode/sort2.uc
+++ b/ucode/sort2.uc
@ -0,0 +1,219 @@
+// Hi! If you're reading this, you probably want to know what this program does.
+// To that, I say: Good luck and I hope you have patience and a strong will to
+// live, 'cause both negatively impacted by trying to read the code below.
+// This isn't a joke; I feel like someone telling a person to get off the edge
+// of a tall building here: reading the code below WILL negatively impact your
+// life. You have been warned.
+
+#define LIST_START 0xE0
+#define LIST_END 0xFE
+#define LIST_START_MODIFIED 0xDF
+#define LIST_END_MODIFIED 0xFD
+#define HASH_MASK 0b1111000
+#define NEGATIVE_START 0x40
+#define NEGATIVE_END 0x78
+#define POSITIVE_START 0x00
+#define POSITIVE_END 0x38
+#define BUCKET_SIZE 8
+#define BUCKET_INDEX_TRACKER 0b11111000
+
+
+// Set up buckets (with absolute sizes)
+// Bucket count: 16
+// Bucket size: 8
+// Motivation: direct mapping from hash to bucket at the cost of bucket size (-6 per bucket)
+#data 0x00 0x00 // 0 (POSITIVE)
+#data 0x08 0x00 // 1 (POSITIVE)
+#data 0x10 0x00 // 2 (POSITIVE)
+#data 0x18 0x00 // 3 (POSITIVE)
+#data 0x20 0x00 // 4 (POSITIVE)
+#data 0x28 0x00 // 5 (POSITIVE)
+#data 0x30 0x00 // 6 (POSITIVE)
+#data 0x38 0x00 // 7 (POSITIVE)
+#data 0x40 0x00 // 8 (NEGATIVE)
+#data 0x48 0x00 // 9 (NEGATIVE)
+#data 0x50 0x00 // A (NEGATIVE)
+#data 0x58 0x00 // B (NEGATIVE)
+#data 0x60 0x00 // C (NEGATIVE)
+#data 0x68 0x00 // D (NEGATIVE)
+#data 0x70 0x00 // E (NEGATIVE)
+#data 0x78 0x00 // F (NEGATIVE)
+
+
+// Shift list down by one address, freeing up 0xFF for random-access use
+reset asr
+mov pm hr
+
+const LIST_START_MODIFIED
+mov ar asr
+mov hr pm
+
+// Initialize PC to point at list
+//const LIST_START
+mov ar pc
+
+
+$BUCKET_SORT_START
+
+// AR = PM[PC]
+// HR = PM[PC + 1]
+mov pc asr
+mov pc ir; incpc
+mov pm ar
+mov pc asr
+mov pm hr
+
+// Hash HR, partially hash AR.
+// Result of HR-hash is in AR
+// Result of AR-hash is in HR
+irl; mov ar gr // PM[0xFF] = AR 
+irl; reset asr
+irl; mov gr pm
+irl; mov pc asr // Set GR to pre-hash value of HR
+irl; mov pm gr
+irl
+irl
+and HASH_MASK
+
+mov ar asr
+mov pm lc; mov pm pc
+incpc
+mov pc pm
+mov ar pc
+
+$FIRST_INSERTION
+
+incpc; bls @FIRST_INSERTION_END_BIGGEST
+
+mov pc asr
+mov pm ar
+sub gr
+adn gr; brn @FIRST_INSERTION_BOTTOM
+
+mov gr pm; incpc
+
+$FIRST_INSERTION_SHIFT
+mov pc asr; bls @FIRST_INSERTION_END_NOTBIGGEST
+mov pm gr
+mov ar pm
+mov gr ar; declc; incpc; bra @FIRST_INSERTION_SHIFT
+
+$FIRST_INSERTION_BOTTOM
+declc; bra @FIRST_INSERTION
+
+$FIRST_INSERTION_END_BIGGEST
+mov pc asr
+mov gr pm
+
+$FIRST_INSERTION_END_NOTBIGGEST
+
+// Prepare second insertion (minimal bookkeeping)
+mov hr ar
+and HASH_MASK
+mov ar asr
+mov pm pc; mov pm lc
+incpc; reset asr
+mov pm gr
+mov ar asr
+mov pc pm
+mov ar pc
+
+
+$SECOND_INSERTION
+
+incpc; bls @SECOND_INSERTION_END_BIGGEST
+
+mov pc asr
+mov pm ar
+sub gr
+adn gr; brn @SECOND_INSERTION_BOTTOM
+
+mov gr pm; incpc
+
+$SECOND_INSERTION_SHIFT
+mov pc asr; bls @SECOND_INSERTION_END_NOTBIGGEST
+mov pm gr
+mov ar pm
+mov gr ar; declc; incpc; bra @SECOND_INSERTION_SHIFT
+
+$SECOND_INSERTION_BOTTOM
+declc; bra @SECOND_INSERTION
+
+$SECOND_INSERTION_END_BIGGEST
+mov pc asr
+mov gr pm
+
+$SECOND_INSERTION_END_NOTBIGGEST
+mov ir ar; mov ir pc
+sub LIST_END_MODIFIED
+brz @BUCKET_SORT_END; incpc
+bra @BUCKET_SORT_START; incpc
+
+$BUCKET_SORT_END
+
+
+//call @BREAK
+
+
+// Merge negative values
+const NEGATIVE_END
+mov ar hr
+const NEGATIVE_START
+mov ar pc
+const LIST_START
+
+$NEGATIVE_MERGE
+mov pc ir
+mov pc asr; incpc
+mov pm lc
+$NEGATIVE_MERGE_MOVE
+mov pc asr; bls @NEGATIVE_MERGE_BOTTOM
+mov pm gr
+mov ar asr
+add 1
+mov gr pm; declc; incpc; bra @NEGATIVE_MERGE_MOVE
+
+$NEGATIVE_MERGE_BOTTOM
+mov ar gr
+mov ir ar
+sub hr
+adn hr; brz @POSITIVE_MERGE_INIT
+add BUCKET_SIZE
+mov ar pc
+mov gr ar; bra @NEGATIVE_MERGE
+
+
+
+
+// Merge positive values
+$POSITIVE_MERGE_INIT
+const POSITIVE_END
+mov ar hr
+const POSITIVE_START
+mov ar pc
+mov gr ar   // Pop AR
+
+$POSITIVE_MERGE
+mov pc asr; incpc
+mov pm lc
+$POSITIVE_MERGE_MOVE
+mov pc asr; bls @POSITIVE_MERGE_BOTTOM
+mov pm gr
+mov ar asr
+add 1
+mov gr pm; declc; incpc; bra @POSITIVE_MERGE_MOVE
+
+$POSITIVE_MERGE_BOTTOM
+mov ar gr
+mov pc ar
+and BUCKET_INDEX_TRACKER
+sub hr
+adn hr; brz @PROGRAM_END
+adn BUCKET_SIZE
+mov ar pc
+mov gr ar; bra @POSITIVE_MERGE
+
+
+$PROGRAM_END
+$BREAK
+halt
--- a/ucode/sort3.uc
+++ b/ucode/sort3.uc
@ -0,0 +1,119 @@
+#optable 0x0 @BREAK
+#optable 0x1 @BREAK
+#optable 0x2 @BREAK
+#optable 0x3 @BREAK
+#optable 0x4 @BREAK
+#optable 0x5 @BREAK
+#optable 0x6 @BREAK
+#optable 0x7 @BREAK
+#optable 0x9 @BREAK
+#optable 0xa @BREAK
+#optable 0xb @BREAK
+#optable 0xc @BREAK
+#optable 0xd @BREAK
+#optable 0xe @BREAK
+#optable 0xf @BREAK
+
+
+#define LIST_START 0xE0
+#define HASH_MASK 0b1111
+#define FIRST_BUCKET 0x10   // Index of start of first bucket
+#define LAST_BUCKET 0xD3    // Index of start of last bucket
+#define BUCKET_SIZE 13      // Each bucket an hold 13 elements before spilling
+
+// LUT entries contain pointers to last element in bucket
+// This allows the LUT to serve as both a jump-table and as bucket-headers
+// Look at me being all resourceful and stuff, huh? Inb4 slowdowns ;))
+#data 0x00 0x77
+#data 0x01 0x84
+#data 0x02 0x91
+#data 0x03 0x9e
+#data 0x04 0xab
+#data 0x05 0xb8
+#data 0x06 0xc5
+#data 0x07 0xd2
+#data 0x08 0x0f
+#data 0x09 0x1c
+#data 0x0a 0x29
+#data 0x0b 0x36
+#data 0x0c 0x43
+#data 0x0d 0x50
+#data 0x0e 0x5d
+#data 0x0f 0x6a
+
+const LIST_START
+mov ar pc
+
+// Two values sorted per iteration, so half the iterations obv ;)
+// For consistency, we just decrement LC twice per iteration, though
+// Had this course also stressed power use or other efficiency-related questions,
+// we probably wouldn't decrement twice per iteration for the sole purpose of "clarity"
+// But alas, Kent doesn't care, so why should we?
+lcset 32
+
+$BUCKET_SORT
+bls @MERGE_INIT
+
+mov pc asr; incpc
+mov pm ar; mov pm ir // Use whatever value we're sorting to index GR at complete random
+mov pc asr; incpc
+mov pm hr
+
+// Shift AR and HR
+irl; mov pm gr
+irl
+irl
+irl
+and HASH_MASK // Completely hash value from HR
+mov ar asr
+mov pm ar
+add 1
+mov ar pm
+mov ar asr
+mov gr pm; declc // Mark element as sorted by decrementing LC
+
+mov hr ar
+and HASH_MASK // Completely hash value from AR
+mov ar asr
+mov pm ar
+add 1
+mov ar pm
+mov ar asr
+mov ir pm; declc // Mark element as sorted by decrementing LC
+
+bra @BUCKET_SORT
+
+
+//// ---- MERGE ---- ////
+$MERGE_INIT
+
+// HR points to LUT
+// PC points to start of first bucket
+// AR points to list
+mov pc hr   // PC is always 0 after bucketsort because FF+1 mod 100 = 0. Imagine that!
+const FIRST_BUCKET
+mov ar pc
+const LIST_START
+
+$MERGE
+mov ar ir
+mov hr asr
+sub pm
+mov pc gr
+
+
+
+
+
+
+
+// If this was not the last bucket, branch back to start of merge
+sub LAST_BUCKET
+adn LAST_BUCKET
+bnz @MERGE
+
+
+$END
+
+$BREAK
+halt
--- a/ucode/sort4.uc
+++ b/ucode/sort4.uc
@ -0,0 +1,148 @@
+// Hi! If you're reading this, you probably want to know what this program does
+// To that, I say: Good luck and I hope you have patience and a strong will to
+// live, 'cause both will be negatively impacted by trying to read the code
+// below.
+// This isn't a joke; I feel like someone telling a person to get off the edge
+// of a tall building here: reading the code below WILL negatively impact your
+// life. You have been warned.
+
+// For clarity's sake, I thought I'd add a small preface here:
+// We use the optable as a replacement for a hashing algorithm. Check the
+// #optable directives if you're wondering what bucket the 4 highest bits of a
+// value end up pointing to. Each entry in the optable simply loads the
+// address of the corresponding bucket and initiates an insertion sort for said 
+// bucket.
+// Labels ending in "_SPEC" are edge-case optimization labels. They are
+// branched to in the case of special edge cases with the idea being that they
+// are much faster than following the reguar code-path (even if said path would
+// ultimately perform the same action).
+// For example, "call @JTABLE_SPEC" invokes a special jumptable subroutine
+// which parallelizes some of the boilerplate needed for the upcoming merge.
+
+
+#define LIST_START 0xE0
+#define HIGHEST_BUCKET 0xD2
+
+// Generate jump-table ;)
+#emit
+>for i in range(16): print("#optable "+hex(i)+" @OT_"+hex(i)[2::])
+
+// Generate program memory:
+//      Addresses 0xE0-0xFF ignored
+//      Addresses that are a multiple of 0xE are set to 0
+//      All other addresses contain a relative pointer to the next bucket start
+#pmgen if address < 0xE0: print(str(address)+" "+str(0 if (address % 14) == 0 else ((address - (address % 0xE)) - 0xE) if address > 0xD else -1))
+
+
+// Set all GR-registers to -1, so that we always can call "sub gr" to increment
+// AR, no matter what value we have in IR ;)
+const 0xB00         // GRx=10, M=11
+lsr; mov ar ir
+lsr; reset gr
+lsr; reset grm
+mov ar ir           // GRx=00, M=01
+reset gr
+reset grm
+
+
+// Initialize PC to point at list
+const LIST_START
+mov ar pc
+mov pc asr
+
+// Perform bucketsort for 32 elements
+#emit
+>for i in range(31):
+>  print("mov pm ir; incpc; call @JTABLE")
+mov pm ir; call @JTABLE_SPEC
+
+
+// Initialize state for merge
+const HIGHEST_BUCKET
+mov ar asr
+mov pm lc
+sub gr
+
+// Copy elements to list
+$MERGE
+mov ar asr; declc; bls @MB_SPEC
+mov pm ir
+mov pc asr                      // This branch improves stability
+mov ir pm; incpc; bls @MERGE_BOTTOM  // Transfer value and copy more elements
+sub gr; bra @MERGE
+
+$MERGE_BOTTOM
+sub gr
+mov ar asr
+
+$MB_SPEC
+mov pm ar; mov pm asr
+sub gr
+mov pm lc; bnz @MERGE
+
+// Breakpoint hook and program termination point
+$BREAK
+$END
+halt
+
+
+
+
+// Jump-table subroutine
+$JTABLE
+mov pc hr; bop
+
+$JTABLE_SPEC
+const LIST_START
+mov ar hr; bop
+
+
+// Generate jump table
+#emit
+>bucket_size=14
+>for i in range(16):
+>  print("$OT_"+hex(i)[2::])
+>  if i < 8:
+>    print("const "+str(bucket_size*(7-i)))
+>  else:
+>    print("const "+str(bucket_size*(23-i)))
+>  print("mov ar asr; bra @PREPARE_SORT")
+
+
+// Actual bucketsort
+$PREPARE_SORT
+mov pm pc; mov pm lc                    // Load bucket length
+sub gr; incpc                           // Increment bucket length
+mov pc pm; bls @IE_SPEC                 // Store new length
+
+mov ar pc; sub ar
+sub ir // Save -IR into AR (you'll see why)
+
+
+$INSERTION
+mov pc asr; incpc; declc; bls @INSERTION_END_BIGGEST
+add pm                  // Effectively, AR=PM-IR, except more like AR=-IR+PM
+sub pm; brn @INSERTION
+
+mov pm ar
+mov ir pm; bls @IEN_SPEC
+
+$INSERTION_SHIFT
+mov pc asr
+mov pm ir
+mov ar pm; bls @INSERTION_END_NOTBIGGEST
+mov ir ar; declc; incpc; bra @INSERTION_SHIFT
+
+$IEN_SPEC
+mov pc asr
+mov ar pm; bra @INSERTION_END_NOTBIGGEST
+
+$IE_SPEC
+mov ar asr
+
+$INSERTION_END_BIGGEST
+mov ir pm
+
+$INSERTION_END_NOTBIGGEST
+mov hr pc
+mov pc asr; ret
--- a/ucode/test.uc
+++ b/ucode/test.uc
@ -0,0 +1,31 @@
+#optable 0x0 @TABLE_0
+#optable 0x1 @TABLE_1
+#optable 0x2 @TABLE_2
+#optable 0x3 @TABLE_3
+
+mov pm ir
+reset asr
+mov pm grm
+const 0
+mov ar asr
+mov pm gr
+bop
+
+$TABLE_0
+const 1
+bra @END
+
+$TABLE_1
+const 2
+bra @END
+
+$TABLE_2
+const 3
+bra @END
+
+$TABLE_3
+const 4
+
+
+$END
+halt
--- a/weaver.kt
+++ b/weaver.kt
@ -0,0 +1,211 @@
+import java.io.*
+
+class MachineState {
+    val programMemory = ShortArray(256)// PM
+    val microMemory = IntArray(128)     // MyM
+    val k1 = ByteArray(16)             // K1
+    val k2 = ByteArray(4)              // K2
+    var pc: Byte = 0.toByte()         // PC
+    var asr: Byte = 0.toByte()        // ASR
+    var ar: Short = 0.toShort()       // AR
+    var hr: Short = 0.toShort()       // HR
+    var gr0: Short = 0.toShort()      // GR0
+    var gr1: Short = 0.toShort()      // GR0
+    var gr2: Short = 0.toShort()      // GR0
+    var gr3: Short = 0.toShort()      // GR0
+    var ir: Byte = 0.toByte()         // IR
+    var uPC: Byte = 0.toByte()        // MyPC
+    var uSP: Byte = 0.toByte()        // SMyPC
+    var lc: Byte = 0.toByte()         // LC
+
+    override fun toString(): String {
+        val builder = StringBuilder("PM:\n")
+        for(index in 0 until programMemory.size)
+            builder.append(index.toUHex()).append(": ").append(programMemory[index].toUHex()).append("\n")
+        
+        builder.append("\nMyM:\n")
+        for(index in 0 until microMemory.size)
+            builder.append(index.toUHex()).append(": ").append(microMemory[index].toShortUHex()).append("\n")
+
+        builder.append("\nK1:\n")
+        for(index in 0 until k1.size)
+            builder.append(index.toUHex()).append(": ").append(k1[index].toUHex()).append("\n")
+        
+        builder.append("\nK2:\n")
+        for(index in 0 until k2.size)
+            builder.append(index.toUHex()).append(": ").append(k2[index].toUHex()).append("\n")
+        
+        return builder
+            .append("\nPC:\n")
+            .append(pc.toUHex())
+            .append("\n\nASR:\n")
+            .append(asr.toUHex())
+            .append("\n\nAR:\n")
+            .append(ar.toUHex())
+            .append("\n\nHR:\n")
+            .append(hr.toUHex())
+            .append("\n\nGR0:\n")
+            .append(gr0.toUHex())
+            .append("\n\nGR1:\n")
+            .append(gr1.toUHex())
+            .append("\n\nGR2:\n")
+            .append(gr2.toUHex())
+            .append("\n\nGR3:\n")
+            .append(gr3.toUHex())
+            .append("\n\nIR:\nb")
+            .append(ir.toInt().or(1 shl 30).toString(2).substring(28))
+            .append("\n\nMyPC:\n")
+            .append(uPC.toUHex())
+            .append("\n\nSMyPC:\n")
+            .append(uSP.toUHex())
+            .append("\n\nLC:\n")
+            .append(lc.toUHex())
+            .append("\n\nO_flag:\n\nC_flag:\n\nN_flag:\n\nZ_flag:\n\nL_flag:\nEnd_of_dump_file")
+            .toString()
+    }
+
+    companion object {
+        fun parseState(rawState: String): MachineState {
+            val state = MachineState()
+            val lines = rawState.replace("\r", "").split("\n").toTypedArray()
+
+            // Read PM
+            for(index in 0 until 256)
+                state.programMemory[index] = Integer.parseInt(lines[index + 1].substring(4), 16).toShort()
+
+            // Read MyM
+            for(index in 0 until 128)
+                state.microMemory[index] = Integer.parseInt(lines[index + 3 + 256].substring(4), 16)
+            
+            // Read K1
+            for(index in 0 until 16)
+                state.k1[index] = Integer.parseInt(lines[index + 5 + 256 + 128].substring(4), 16).toByte()
+
+            // Read K2
+            for(index in 0 until 4)
+                state.k2[index] = Integer.parseInt(lines[index + 7 + 256 + 128 + 16].substring(4), 16).toByte()
+            
+            state.pc = Integer.parseInt(lines[413], 16).toByte()
+            state.asr = Integer.parseInt(lines[416], 16).toByte()
+            state.ar = Integer.parseInt(lines[419], 16).toShort()
+            state.hr = Integer.parseInt(lines[422], 16).toShort()
+            state.gr0 = Integer.parseInt(lines[425], 16).toShort()
+            state.gr1 = Integer.parseInt(lines[428], 16).toShort()
+            state.gr2 = Integer.parseInt(lines[431], 16).toShort()
+            state.gr3 = Integer.parseInt(lines[434], 16).toShort()
+            state.ir = Integer.parseInt(lines[437].substring(1), 2).toByte()
+            state.uPC = Integer.parseInt(lines[440], 16).toByte()
+            state.uSP = Integer.parseInt(lines[443], 16).toByte()
+            state.lc = Integer.parseInt(lines[446], 16).toByte()
+
+            return state
+        }
+    }
+}
+
+fun error(message: String){
+    System.err.println(message)
+    println("Usage:\n\tkotlin WeaverKt [machineCodeFile] [outputFile]\nor\n\tkotlin WeaverKt [machineCodeFile] [inputState] [outputFile]")
+    System.exit(1)
+}
+
+fun main(args: Array<String>){
+    if(args.size > 3) error("Too many arguments!")
+    if(args.size < 2) error("Too few arguments!")
+    
+    val weaveFile = File(args[0])
+    if(!weaveFile.isFile) error("Given machine code file doesn't exist!")
+
+    val state: MachineState
+
+    if(args.size == 2) state = MachineState()
+    else{
+        val file = File(args[1])
+        if(file.isFile) state = MachineState.parseState(file.readText())
+        else{
+            error("Machine state file (${args[1]}) doesn't exist!")
+            state = MachineState()
+        }
+    }
+
+    val weaveData = weaveFile.readText().replace("\r", "").split("\n").toTypedArray()
+    var weaveUCode = false
+    var pIndex = 0
+    var uIndex = 0
+    fun index()= if(weaveUCode) uIndex else pIndex
+    fun iIdx() {
+        if(weaveUCode) ++uIndex
+        else ++pIndex
+    }
+    fun sIdx(value: Int){
+        if(weaveUCode) uIndex = value
+        else pIndex = value
+    }
+    for(rawValue in weaveData){
+        val value = {
+            var v = rawValue.replace(" ", "").replace("\t", "")
+            
+            // Return
+            if(v.indexOf("#") > 0) v.substring(0, v.indexOf("#")) else v
+        }()
+        
+        if(value.length == 0) continue
+        else if(value.startsWith("@")){
+            if(value == "@u") weaveUCode = true
+            else if(value == "@p") weaveUCode = false
+            else if(value.startsWith("@k1")){
+                // Parse K1 table entry
+                if(value.length != 6)
+                    error("Badly formatted K1 declaration: $rawValue")
+                val index = value.substring(3, 4).toInt(16)
+                val k1Value = value.substring(4, 6).toInt(16).toByte()
+                if(k1Value < 0)
+                    error("Invalid K1 address pointer (must be in range 00-7F): $rawValue")
+
+                state.k1[index] = k1Value
+            }
+            else if(value.startsWith("@k2")){
+                // Parse K2 table entry
+                if(value.length != 6)
+                    error("Badly formatted K2 declaration: $rawValue")
+
+                val index = value.substring(3, 4).toInt(16)
+                if(index > 4)
+                    error("Invalid K2 index value (must be in range 0-3): $rawValue")
+
+
+                val k2Value = value.substring(4, 6).toInt(16).toByte()
+                if(k2Value < 0)
+                    error("Invalid K2 address pointer (must be in range 00-7F): $rawValue")
+
+                state.k2[index] = k2Value
+            }
+            else if(value.startsWith("@0x")) sIdx(Integer.parseInt(value.substring(3), 16))
+            else sIdx(Integer.parseInt(value.substring(1), 10))
+
+            continue
+        }
+        else if((weaveUCode && index() >= state.microMemory.size) || (!weaveUCode && index() >= state.programMemory.size)) error("Memory out of bounds: ${index()}! Did you pass too much data?")
+        else if((weaveUCode && value.length != 7) || (!weaveUCode && value.length != 4)) error("Cannot weave data of bad length: $value")
+        else try{
+            if(weaveUCode) state.microMemory[index()] = Integer.parseInt(value, 16)
+            else state.programMemory[index()] = Integer.parseInt(value, 16).toShort()
+        }catch(e: NumberFormatException){
+            error("Cannot weave non-hex data: $value")
+        }
+        iIdx()
+    }
+
+    val outputFile = if(args.size == 3) File(args[2]) else File(args[1])
+    if(outputFile.isFile) outputFile.delete()
+    outputFile.createNewFile()
+
+
+    val machineData = state.toString()
+    outputFile.bufferedWriter().use{ it.write(machineData, 0, machineData.length) }
+}
+
+fun Short.toUHex() = toInt().and(0xFFFF.toInt()).or(1.shl(30)).toString(16).substring(4)
+fun Int.toUHex() = and(0xFF).or(1.shl(30)).toString(16).substring(6)
+fun Int.toShortUHex() = and((-1 ushr 7)).or(1.shl(30)).toString(16).substring(1)
+fun Byte.toUHex() = toInt().and(0xFF.toInt()).or(1.shl(30)).toString(16).substring(6)
Author	SHA1	Message	Date
Gabriel Tofvesson	644007b002	Update description of IR in README	2020-03-21 23:02:38 +01:00
Gabriel Tofvesson	a7aa9348ef	Update description of IR in README	2020-03-21 22:39:57 +01:00
Gabriel Tofvesson	44328f7cb4	Improve README style	2020-03-21 21:09:55 +01:00
Gabriel Tofvesson	96f120a99d	Simplify code	2019-05-07 17:28:34 +02:00
Gabriel Tofvesson	a21c5c3320	Update README	2019-05-06 10:31:33 +02:00
Gabriel Tofvesson	d9c7ec81c3	Increase bucket sizes	2019-05-06 10:22:55 +02:00
Gabriel Tofvesson	4e3c6d725b	Implement script-based code generation	2019-05-06 10:22:37 +02:00
Gabriel Tofvesson	82d69b721e	Add script support for program-memory generation	2019-05-05 22:11:09 +02:00
Gabriel Tofvesson	660f10c867	Add clarifying comments	2019-04-30 18:40:54 +02:00
Gabriel Tofvesson	d0b8f2d978	Optimize final bucketsort call	2019-04-30 18:30:14 +02:00
Gabriel Tofvesson	37f1afe452	Optimize boilerplate	2019-04-30 18:19:56 +02:00
Gabriel Tofvesson	a2420be4c8	Optimize for special cases	2019-04-30 17:59:03 +02:00
Gabriel Tofvesson	d71a2bdb97	Update README	2019-04-30 17:50:37 +02:00
Gabriel Tofvesson	13b4827969	Optimize for empty buckets and improve stability for full buckets	2019-04-30 17:50:04 +02:00
Gabriel Tofvesson	ae3d25bf65	Optimize sort4 based on practical tests	2019-04-29 23:16:26 +02:00
Gabriel Tofvesson	d067a72c38	Add clickable links to README	2019-04-29 22:45:03 +02:00
Gabriel Tofvesson	676109cbc3	Add clickable links to README	2019-04-29 22:43:29 +02:00
Gabriel Tofvesson	1adf41888a	Add clickable links to README	2019-04-29 22:33:20 +02:00
Gabriel Tofvesson	c422e92ec7	Add clickable links to README	2019-04-29 22:11:46 +02:00
Gabriel Tofvesson	c6c866b86e	Update README	2019-04-29 22:05:23 +02:00
Gabriel Tofvesson	7b060062af	Improve bucketsort by removing redundant INSERTION_BOTTOM	2019-04-29 21:26:01 +02:00
Gabriel Tofvesson	800c7c065c	Improve stability	2019-04-27 02:51:54 +02:00
Gabriel Tofvesson	9ad268b619	Prepare sort4 for possible post-sort optimization	2019-04-27 02:51:42 +02:00
Gabriel Tofvesson	926e572a53	Fully unrolled sort-loop	2019-04-27 02:34:22 +02:00
Gabriel Tofvesson	6b33921e8d	Update README	2019-04-27 02:33:41 +02:00
Gabriel Tofvesson	ac78950784	Optimize bucketsort for unrolling by converting it to a zero-loss subroutine	2019-04-27 02:08:44 +02:00
Gabriel Tofvesson	40d1031392	Fix build script	2019-04-27 00:44:39 +02:00
Gabriel Tofvesson	0f5c426699	Optimize Sort4	2019-04-27 00:44:06 +02:00
Gabriel Tofvesson	fd188204b4	Implement M-bit selection of GR mux	2019-04-26 22:27:21 +02:00
Gabriel Tofvesson	2b9bdf568a	Update build-script to be directory-independent	2019-04-26 22:27:21 +02:00
Gabriel Tofvesson	76c2fbb344	Move microcode projects to separate directory	2019-04-26 22:27:21 +02:00
Gabriel Tofvesson	50de68941d	Update README	2019-04-24 23:00:00 +02:00
Gabriel Tofvesson	078f4b0736	Optimize bsrt2	2019-04-24 22:54:24 +02:00
Gabriel Tofvesson	99a3024b2a	Implement bsrt using K1 jump table	2019-04-24 22:25:13 +02:00
Gabriel Tofvesson	4dd6bf6115	Update README	2019-04-24 20:11:08 +02:00
Gabriel Tofvesson	e67be67d57	Update README	2019-04-24 20:05:58 +02:00
Gabriel Tofvesson	708bd3a7b9	Update instructions according to new naming scheme	2019-04-24 20:05:58 +02:00
Gabriel Tofvesson	a441e4bad2	Rename arithmetic 32-bit shift/rotate instructions	2019-04-24 20:05:58 +02:00
Gabriel Tofvesson	8e337fe933	Implement special branch instructions	2019-04-24 19:56:24 +02:00
Gabriel Tofvesson	772472d932	Add example of K1 and K2 declarations to sort3	2019-04-24 19:47:05 +02:00
Gabriel Tofvesson	aa13da3219	Add support for K1 and K2 declarations to microcompiler	2019-04-24 19:46:42 +02:00
Gabriel Tofvesson	60b1278f6a	Add support for K-table declarations to weaver	2019-04-24 19:03:57 +02:00
Gabriel Tofvesson	8b5d6e5ad4	Update README	2019-04-24 13:31:52 +02:00
Gabriel Tofvesson	fd6a442962	Implement bucket-sort without inline insertion sort	2019-04-24 13:26:19 +02:00
Gabriel Tofvesson	3ff3e8bd67	Update README	2019-04-19 23:52:45 +02:00
Gabriel Tofvesson	36c64a79bf	Optimize bookkeeping and merge	2019-04-19 23:52:32 +02:00
Gabriel Tofvesson	150906141f	Optimize merge	2019-04-19 23:52:09 +02:00
Gabriel Tofvesson	4db5192579	Implement optimized bucket-sort (1100 cycles)	2019-04-19 19:40:32 +02:00
Gabriel Tofvesson	59534c188f	Update algorithm parameters	2019-04-19 12:27:18 +02:00
Gabriel Tofvesson	46db67a678	Clean code	2019-04-12 10:33:17 +02:00
GabrielTofvesson	d436b965d8	Implement dual-hash and merge-optimization	2019-04-11 21:01:13 +02:00
Gabriel Tofvesson	94b4e7760c	Rearrange buckets	2019-04-11 18:47:20 +02:00
Gabriel Tofvesson	75efe84317	Update reference code	2019-04-11 10:02:41 +02:00
Gabriel Tofvesson	e4d45e470c	Add LUT	2019-04-11 10:01:45 +02:00
Gabriel Tofvesson	ec2b562610	Redesign to sort signe values	2019-04-10 15:55:11 +02:00
GabrielTofvesson	db786624d7	Implement final merge of buckets	2019-04-10 02:59:33 +02:00
GabrielTofvesson	e48304d978	Implement bucket/insertion sort	2019-04-10 02:04:22 +02:00
Gabriel Tofvesson	51ef1d1d15	Merge branch 'dev' of gitlab.ida.liu.se:edvth289/TSEA28-Microkod into dev	2019-04-10 00:14:14 +02:00
Gabriel Tofvesson	66e1898a0d	Start modifying BucketSort implementation	2019-04-10 00:12:48 +02:00
Gabriel Tofvesson	09ecc55898	Implement RESET uASM instruction	2019-04-10 00:11:55 +02:00
Gabriel Tofvesson	5586eec381	Fix number format bug for program-memory declarations	2019-04-10 00:09:23 +02:00
Gabriel Tofvesson	710927523e	Weave random data into programs as part of compilation pipeline	2019-04-10 00:07:02 +02:00
Gabriel Tofvesson	0277e26422	Add clearer pipeline output info	2019-04-10 00:06:09 +02:00
Gabriel Tofvesson	00379b472b	Add explicit memory type declaration to random generator	2019-04-10 00:04:33 +02:00
Gabriel Tofvesson	def8aa8e90	Add make rule	2019-04-10 00:03:17 +02:00
Edvard Thörnros	a977124736	Fix README.md	2019-04-09 15:23:56 +02:00
Gabriel Tofvesson	8012c9409e	Update README.md	2019-04-08 22:16:02 +02:00
GabrielTofvesson	7edc6f7582	Update readme	2019-04-08 22:23:39 +02:00
GabrielTofvesson	9d0b798c20	Update readme	2019-04-08 22:16:54 +02:00
GabrielTofvesson	12e5aa4913	Remove NOP argument requirement	2019-04-08 22:12:11 +02:00
GabrielTofvesson	2a8eff03a0	Update gitignore to omit default build-script output file	2019-04-08 22:06:17 +02:00
GabrielTofvesson	866d4eeb6b	Update bucket sort algorithm	2019-04-08 22:04:51 +02:00
GabrielTofvesson	2702d98253	Add support for program-memory initial-states in uASM file	2019-04-08 22:04:27 +02:00
GabrielTofvesson	cdc60b86d1	Add support for switching between program-memory and micro-memory	2019-04-08 22:03:33 +02:00
GabrielTofvesson	b240f29efa	Update sorting algorithm to use constant definitions	2019-04-08 21:33:50 +02:00
GabrielTofvesson	9292c91352	Add support for constant definitions	2019-04-08 21:33:23 +02:00
GabrielTofvesson	680e6eaf23	Add documentation	2019-04-08 20:32:00 +02:00
Gabriel Tofvesson	d661391132	Start implementing Bucket Sort in uASM	2019-04-08 18:17:01 +02:00
Gabriel Tofvesson	bb5cb61556	Fix build-script formatting	2019-04-08 18:15:28 +02:00
Gabriel Tofvesson	5fcb5f2a7f	Allow constant values for general ALU operations	2019-04-08 18:14:26 +02:00
Gabriel Tofvesson	343d028398	Update sample microprogram	2019-04-07 14:10:50 +02:00
Gabriel Tofvesson	6755bedf9a	Fix instruction-merging with address references	2019-04-07 14:10:32 +02:00
Gabriel Tofvesson	35d15eb478	Update gitignore	2019-04-07 13:55:22 +02:00
Gabriel Tofvesson	ddd50eb165	Add automated build scripts	2019-04-07 13:54:18 +02:00
Gabriel Tofvesson	c2dc8dec0d	Update sample microcode file	2019-04-07 13:53:52 +02:00
Gabriel Tofvesson	e900daa435	Remove debug print statements	2019-04-07 13:53:37 +02:00
Edvard Thörnros	d4638fc517	Merge branch 'dev' of gitlab.ida.liu.se:edvth289/TSEA28-Microkod into dev	2019-04-06 23:23:19 +02:00
Edvard Thörnros	59cbe4938a	C++ sorting implementation.	2019-04-06 23:22:46 +02:00
Gabriel Tofvesson	ff924891f2	Add program to random-generate hex values	2019-04-06 23:16:13 +02:00
Gabriel Tofvesson	650cb46417	Add sample uASM program	2019-04-06 23:15:14 +02:00
Gabriel Tofvesson	2c9e19b6fe	Add microcompiler	2019-04-06 23:14:45 +02:00
Gabriel Tofvesson	e2025ec55d	Add support for microcode definition	2019-04-06 23:14:26 +02:00
Gabriel Tofvesson	fef6fa0ecb	Update sorter algo	2019-04-06 23:13:55 +02:00
Gabriel Tofvesson	6fb40ff4e2	Decrement PC	2019-04-06 23:13:28 +02:00
Gabriel Tofvesson	d51763221a	Add competing ASM type ;)	2019-04-05 23:21:42 +02:00
Gabriel Tofvesson	e6a0351942	Add machine-code weaver	2019-04-05 23:19:05 +02:00
Gabriel Tofvesson	e2f3ab756d	Implement compilation from files	2019-04-05 21:34:49 +02:00
Gabriel Tofvesson	655832e672	Add gitignore	2019-04-05 21:27:19 +02:00
Gabriel Tofvesson	e6008f5d74	Finish alpha version of compiler	2019-04-05 21:26:21 +02:00
Gabriel Tofvesson	ca8716ac6d	Add BLT and CMi	2019-04-05 17:12:09 +02:00
Edvard Thörnros	9b86c68ebc	Fix bug in assembly.	2019-04-05 17:11:11 +02:00
Edvard Thörnros	883295bdfa	Fix bug.	2019-04-05 17:00:21 +02:00
Edvard Thörnros	b94229c504	Add in CMI instruction.	2019-04-05 16:51:50 +02:00
Gabriel Tofvesson	52ae1cc1fa	Update implementations	2019-04-05 16:24:26 +02:00
Gabriel Tofvesson	58f00e79d7	Merge branch 'dev' of gitlab.ida.liu.se:edvth289/TSEA28-Microkod into dev	2019-04-05 14:32:50 +02:00
Gabriel Tofvesson	3988f16ad4	Add compiler implemented in Kotlin	2019-04-05 14:06:30 +02:00
Gabriel Tofvesson	e40bc8675f	Complete BEQ implementation	2019-04-05 12:55:39 +02:00
Gabriel Tofvesson	dd7169d6ca	Fix HALT insn	2019-04-05 12:55:16 +02:00
Gabriel Tofvesson	5d30f44295	Remove redundant HALT insn	2019-04-05 12:53:45 +02:00
Gabriel Tofvesson	f5fdd52a71	Implement CMP and BEQ	2019-04-05 12:50:47 +02:00
Gabriel Tofvesson	4925cd654d	Add CMP and BEQ	2019-04-05 12:50:20 +02:00