diff --git a/src/main/java/dk/camelot64/kickc/Compiler.java b/src/main/java/dk/camelot64/kickc/Compiler.java
index 02f2794ae..a84672de1 100644
--- a/src/main/java/dk/camelot64/kickc/Compiler.java
+++ b/src/main/java/dk/camelot64/kickc/Compiler.java
@@ -236,7 +236,6 @@ public class Compiler {
       optimizations.add(new PassNAddNumberTypeConversions(program));
       optimizations.add(new PassNAddArrayNumberTypeConversions(program));
       optimizations.add(new Pass2InlineCast(program));
-      //optimizations.add(new Pass2NopCastInlining(program));
       optimizations.add(new PassNCastSimplification(program));
       optimizations.add(new PassNFinalizeNumberTypeConversions(program));
       optimizations.add(new PassNTypeInference(program));
@@ -307,6 +306,7 @@ public class Compiler {
       List<Pass2SsaOptimization> constantOptimizations = new ArrayList<>();
       constantOptimizations.add(new PassNStatementIndices(program));
       constantOptimizations.add(new PassNVariableReferenceInfos(program));
+      constantOptimizations.add(new Pass2NopCastInlining(program));
       constantOptimizations.add(new Pass2MultiplyToShiftRewriting(program));
       constantOptimizations.add(new Pass2ConstantInlining(program));
       constantOptimizations.add(new Pass2ConstantAdditionElimination(program));
diff --git a/src/main/java/dk/camelot64/kickc/passes/Pass2NopCastInlining.java b/src/main/java/dk/camelot64/kickc/passes/Pass2NopCastInlining.java
index 8d1fe6a05..8af187c3a 100644
--- a/src/main/java/dk/camelot64/kickc/passes/Pass2NopCastInlining.java
+++ b/src/main/java/dk/camelot64/kickc/passes/Pass2NopCastInlining.java
@@ -4,6 +4,7 @@ import dk.camelot64.kickc.model.ControlFlowBlock;
 import dk.camelot64.kickc.model.Program;
 import dk.camelot64.kickc.model.statements.Statement;
 import dk.camelot64.kickc.model.statements.StatementAssignment;
+import dk.camelot64.kickc.model.symbols.Variable;
 import dk.camelot64.kickc.model.types.SymbolType;
 import dk.camelot64.kickc.model.types.SymbolTypeInference;
 import dk.camelot64.kickc.model.types.SymbolTypePointer;
@@ -12,8 +13,7 @@ import dk.camelot64.kickc.model.values.RValue;
 import dk.camelot64.kickc.model.values.SymbolRef;
 import dk.camelot64.kickc.model.values.VariableRef;
 
-import java.util.LinkedHashMap;
-import java.util.ListIterator;
+import java.util.*;
 
 /**
  * Compiler Pass inlineng cast assignments that has no effect (byte to/from signed byte, word to/from signed word)
@@ -24,12 +24,19 @@ public class Pass2NopCastInlining extends Pass2SsaOptimization {
       super(program);
    }
 
+   /** We can either inline intermediate vars or expressions - not both at once */
+   enum Mode { VARS, EXPR }
+
    /**
     * Inline cast assignments that has no effect (byte to/from signed byte, word to/from signed word)
     */
    @Override
    public boolean step() {
-      LinkedHashMap<SymbolRef, RValue> castAliasses = new LinkedHashMap<>();
+      LinkedHashMap<SymbolRef, RValue> replace1 = new LinkedHashMap<>();
+      LinkedHashMap<SymbolRef, RValue> replace2 = new LinkedHashMap<>();
+      Set<SymbolRef> delete = new HashSet<>();
+
+      Mode mode = null;
 
       for(ControlFlowBlock block : getGraph().getAllBlocks()) {
          ListIterator<Statement> stmtIt = block.getStatements().listIterator();
@@ -58,20 +65,59 @@ public class Pass2NopCastInlining extends Pass2SsaOptimization {
                      isNopCast = true;
                   }
                   if(isNopCast && assignment.getlValue() instanceof VariableRef) {
-                     getLog().append("Inlining Noop Cast " + assignment.toString(getProgram(), false));
-                     // Add the alias for replacement
-                     castAliasses.put((VariableRef) assignment.getlValue(), assignment.getrValue2());
-                     // Remove the assignment
-                     stmtIt.remove();
-                  }
+                     boolean handled = false;
+                     if(castValue.getValue() instanceof VariableRef && ((VariableRef)castValue.getValue()).isIntermediate()) {
+                        if(mode==null || mode==Mode.VARS ) {
+                           mode = Mode.VARS;
+                           Collection<Integer> varUseStatements = getProgram().getVariableReferenceInfos().getVarUseStatements((VariableRef) castValue.getValue());
+                           if(varUseStatements.size() == 1 && assignment.getIndex().equals(varUseStatements.iterator().next())) {
+                              // Cast variable is only ever used in the cast
+                              getLog().append("Inlining Noop Cast " + assignment.toString(getProgram(), false) + " keeping " + assignment.getlValue());
+                              // 1. Inline the cast
+                              replace1.put((VariableRef) assignment.getlValue(), castValue);
+                              // 2. Rename the cast variable (since we like the assignment variable better)
+                              replace2.put((SymbolRef) castValue.getValue(), assignment.getlValue());
+                              // 3. Delete the cast variable
+                              delete.add((SymbolRef) castValue.getValue());
+                              // Change the type of the assignment variable
+                              Variable assignmentVar = getScope().getVariable((VariableRef) assignment.getlValue());
+                              Variable castVar = getScope().getVariable((VariableRef) castValue.getValue());
+                              assignmentVar.setType(castVar.getType());
+                              // Remove the assignment
+                              stmtIt.remove();
+                              handled = true;
+                           }
+                        }
+                     }
 
+                     if(!handled) {
+                        if(mode==null || mode==Mode.EXPR) {
+                           mode = Mode.EXPR;
+                           getLog().append("Inlining Noop Cast " + assignment.toString(getProgram(), false) + " keeping " + castValue.getValue());
+                           // 1. Inline the cast
+                           replace1.put((VariableRef) assignment.getlValue(), assignment.getrValue2());
+                           // 2. Delete the assignment variable
+                           delete.add((VariableRef) assignment.getlValue());
+                           // Remove the assignment
+                           stmtIt.remove();
+                        }
+                     }
+                  }
                }
             }
          }
       }
-      replaceVariables(castAliasses);
-      deleteSymbols(getScope(), castAliasses.keySet());
-      return (castAliasses.size() > 0);
+
+      if(replace1.size()>0) {
+         // 1. Perform first replace
+         replaceVariables(replace1);
+         // 2. Perform second replace
+         replaceVariables(replace2);
+         // 3. Delete unused symbols
+         deleteSymbols(getScope(), delete);
+      }
+
+      return (replace1.size() > 0);
    }
 
 
diff --git a/src/main/java/dk/camelot64/kickc/passes/PassNSimplifyExpressionWithZero.java b/src/main/java/dk/camelot64/kickc/passes/PassNSimplifyExpressionWithZero.java
index c59ea0bd1..5bd35c0ed 100644
--- a/src/main/java/dk/camelot64/kickc/passes/PassNSimplifyExpressionWithZero.java
+++ b/src/main/java/dk/camelot64/kickc/passes/PassNSimplifyExpressionWithZero.java
@@ -1,13 +1,12 @@
 package dk.camelot64.kickc.passes;
 
+import dk.camelot64.kickc.model.ConstantNotLiteral;
 import dk.camelot64.kickc.model.Program;
 import dk.camelot64.kickc.model.iterator.ProgramExpressionBinary;
 import dk.camelot64.kickc.model.iterator.ProgramExpressionIterator;
 import dk.camelot64.kickc.model.operators.Operator;
 import dk.camelot64.kickc.model.operators.Operators;
-import dk.camelot64.kickc.model.values.ConstantInteger;
-import dk.camelot64.kickc.model.values.PointerDereferenceSimple;
-import dk.camelot64.kickc.model.values.RValue;
+import dk.camelot64.kickc.model.values.*;
 
 import java.util.concurrent.atomic.AtomicBoolean;
 
@@ -31,19 +30,19 @@ public class PassNSimplifyExpressionWithZero extends Pass2SsaOptimization {
             RValue right = binary.getRight();
             Operator operator = programExpression.getOperator();
             if(Operators.PLUS.equals(operator) || Operators.MINUS.equals(operator) || Operators.BOOL_OR.equals(operator) || Operators.BOOL_XOR.equals(operator)) {
-               if(left instanceof ConstantInteger && ((ConstantInteger) left).getInteger() == 0) {
-                  getLog().append("Simplifying expression containing zero " + binary.getRight().toString()+ " in "+ (currentStmt==null?"":currentStmt.toString(getProgram(), false)));
+               if(isZero(left)) {
+                  getLog().append("Simplifying expression containing zero " + binary.getRight().toString() + " in " + (currentStmt == null ? "" : currentStmt.toString(getProgram(), false)));
                   if(programExpression instanceof ProgramExpressionBinary.ProgramExpressionBinaryPointerDereferenceIndexed) {
                      programExpression.set(new PointerDereferenceSimple(binary.getRight()));
-                  }  else {
+                  } else {
                      programExpression.set(binary.getRight());
                   }
                   modified.set(true);
-               } else if(right instanceof ConstantInteger && ((ConstantInteger) right).getInteger() == 0) {
-                  getLog().append("Simplifying expression containing zero " + binary.getLeft().toString()+ " in "+ (currentStmt==null?"":currentStmt.toString(getProgram(), false)));
+               } else if(isZero(right)) {
+                  getLog().append("Simplifying expression containing zero " + binary.getLeft().toString() + " in " + (currentStmt == null ? "" : currentStmt.toString(getProgram(), false)));
                   if(programExpression instanceof ProgramExpressionBinary.ProgramExpressionBinaryPointerDereferenceIndexed) {
                      programExpression.set(new PointerDereferenceSimple(binary.getLeft()));
-                  }  else {
+                  } else {
                      programExpression.set(binary.getLeft());
                   }
                   modified.set(true);
@@ -55,4 +54,25 @@ public class PassNSimplifyExpressionWithZero extends Pass2SsaOptimization {
       return modified.get();
    }
 
+   /**
+    * Determine if an RValue is effectively a zero
+    *
+    * @param left
+    * @return
+    */
+   public boolean isZero(RValue left) {
+      if(left instanceof ConstantInteger && ((ConstantInteger) left).getInteger() == 0) {
+         return true;
+      } else if(left instanceof ConstantValue) {
+         try {
+            ConstantLiteral literal = ((ConstantValue) left).calculateLiteral(getProgram().getScope());
+            if(literal instanceof ConstantInteger && ((ConstantInteger) literal).getInteger() == 0L)
+               return true;
+         } catch(ConstantNotLiteral e) {
+            // Ignore
+         }
+      }
+      return false;
+   }
+
 }
diff --git a/src/test/ref/bresenhamarr.asm b/src/test/ref/bresenhamarr.asm
index 27134b6fd..79f3d5d74 100644
--- a/src/test/ref/bresenhamarr.asm
+++ b/src/test/ref/bresenhamarr.asm
@@ -8,15 +8,13 @@ main: {
     .const y0 = 0
     .const x1 = $27
     .const y1 = $18
-    .const xd = x1-x0
-    .const yd = y1-y0
     .label x = 4
     .label idx = 2
     .label y = 5
     .label _16 = 6
     lda #y0
     sta y
-    ldx #yd/2
+    ldx #y1/2
     lda #x0
     sta x
     lda #0
@@ -39,8 +37,8 @@ main: {
     inc idx+1
   !:
     txa
-    axs #-[yd]
-    cpx #xd
+    axs #-[y1]
+    cpx #x1
     bcc b2
     beq b2
     inc y
@@ -52,7 +50,7 @@ main: {
     inc idx+1
   !:
     txa
-    axs #xd
+    axs #x1
   b2:
     lda x
     cmp #x1+1
diff --git a/src/test/ref/c64dtv-gfxexplorer.asm b/src/test/ref/c64dtv-gfxexplorer.asm
index 18ca5b0bb..83f245999 100644
--- a/src/test/ref/c64dtv-gfxexplorer.asm
+++ b/src/test/ref/c64dtv-gfxexplorer.asm
@@ -166,12 +166,10 @@ gfx_mode: {
     .label _36 = 9
     .label _40 = 3
     .label _42 = 3
-    .label _52 = 3
     .label _53 = 3
     .label _54 = 3
     .label _55 = 3
     .label _56 = 2
-    .label _57 = 3
     .label _58 = 3
     .label _59 = 3
     .label plane_a = 9
@@ -1150,13 +1148,12 @@ form_render_values: {
 form_field_ptr: {
     .label line = 3
     .label x = $13
-    .label _2 = 3
     lda form_fields_y,x
     tay
     lda form_line_hi,y
-    sta _2+1
+    sta line+1
     lda form_line_lo,y
-    sta _2
+    sta line
     lda form_fields_x,x
     sta x
     rts
@@ -1534,7 +1531,6 @@ gfx_init_plane_fill: {
     .label _1 = 3
     .label _4 = 3
     .label _5 = 3
-    .label _6 = 3
     .label gfxb = 3
     .label by = 7
     .label plane_addr = 9
@@ -1572,12 +1568,12 @@ gfx_init_plane_fill: {
     and #>$3fff
     sta _5+1
     clc
-    lda _6
+    lda gfxb
     adc #<$4000
-    sta _6
-    lda _6+1
+    sta gfxb
+    lda gfxb+1
     adc #>$4000
-    sta _6+1
+    sta gfxb+1
     lda #0
     sta by
   b1:
@@ -2072,7 +2068,6 @@ bitmap_line_xdyi: {
 }
 // bitmap_plot(byte register(X) x, byte register(Y) y)
 bitmap_plot: {
-    .label _0 = 3
     .label plotter_x = 3
     .label plotter_y = 5
     .label plotter = 3
@@ -2084,13 +2079,13 @@ bitmap_plot: {
     sta plotter_y+1
     lda bitmap_plot_ylo,y
     sta plotter_y
-    lda _0
+    lda plotter
     clc
     adc plotter_y
-    sta _0
-    lda _0+1
+    sta plotter
+    lda plotter+1
     adc plotter_y+1
-    sta _0+1
+    sta plotter+1
     lda bitmap_plot_bit,x
     ldy #0
     ora (plotter),y
@@ -2204,11 +2199,10 @@ bitmap_line_ydxd: {
 bitmap_clear: {
     .label bitmap = 3
     .label y = 2
-    .label _3 = 3
     lda bitmap_plot_xlo
-    sta _3
+    sta bitmap
     lda bitmap_plot_xhi
-    sta _3+1
+    sta bitmap+1
     lda #0
     sta y
   b1:
diff --git a/src/test/ref/c64dtv-gfxmodes.asm b/src/test/ref/c64dtv-gfxmodes.asm
index 429e6c1c9..01688cd56 100644
--- a/src/test/ref/c64dtv-gfxmodes.asm
+++ b/src/test/ref/c64dtv-gfxmodes.asm
@@ -1685,7 +1685,6 @@ bitmap_line_xdyi: {
 }
 // bitmap_plot(byte register(X) x, byte register(Y) y)
 bitmap_plot: {
-    .label _0 = 2
     .label plotter_x = 2
     .label plotter_y = 5
     .label plotter = 2
@@ -1697,13 +1696,13 @@ bitmap_plot: {
     sta plotter_y+1
     lda bitmap_plot_ylo,y
     sta plotter_y
-    lda _0
+    lda plotter
     clc
     adc plotter_y
-    sta _0
-    lda _0+1
+    sta plotter
+    lda plotter+1
     adc plotter_y+1
-    sta _0+1
+    sta plotter+1
     lda bitmap_plot_bit,x
     ldy #0
     ora (plotter),y
@@ -1817,11 +1816,10 @@ bitmap_line_ydxd: {
 bitmap_clear: {
     .label bitmap = 2
     .label y = 4
-    .label _3 = 2
     lda bitmap_plot_xlo
-    sta _3
+    sta bitmap
     lda bitmap_plot_xhi
-    sta _3+1
+    sta bitmap+1
     lda #0
     sta y
   b1:
diff --git a/src/test/ref/complex/tetris/tetris.asm b/src/test/ref/complex/tetris/tetris.asm
index ce2a47921..effd08555 100644
--- a/src/test/ref/complex/tetris/tetris.asm
+++ b/src/test/ref/complex/tetris/tetris.asm
@@ -193,9 +193,10 @@ main: {
     ldx current_ypos
     lda current_xpos
     sta current_xpos_119
-    lda current_piece_gfx
+    ldy play_spawn_current._7
+    lda PIECES,y
     sta current_piece_gfx_112
-    lda current_piece_gfx+1
+    lda PIECES+1,y
     sta current_piece_gfx_112+1
     lda current_piece_char
     sta current_piece_char_100
@@ -205,10 +206,15 @@ main: {
     ldx play_spawn_current.piece_idx
     lda #$20
     jsr render_next
-    lda current_piece_gfx
+    ldy play_spawn_current._7
+    lda PIECES,y
     sta current_piece
-    lda current_piece_gfx+1
+    lda PIECES+1,y
     sta current_piece+1
+    lda PIECES,y
+    sta current_piece_gfx
+    lda PIECES+1,y
+    sta current_piece_gfx+1
     lda #0
     sta level_bcd
     sta level
@@ -390,8 +396,8 @@ render_bcd: {
 render_next: {
     // Find the screen area
     .const next_area_offset = $28*$c+$18+4
-    .label next_piece_gfx = 5
     .label next_piece_char = $a
+    .label next_piece_gfx = 5
     .label screen_next_area = 7
     .label l = 9
     cmp #0
@@ -409,14 +415,13 @@ render_next: {
   b2:
     txa
     asl
-    // Render the next piece
     tay
+    lda PIECES_NEXT_CHARS,x
+    sta next_piece_char
     lda PIECES,y
     sta next_piece_gfx
     lda PIECES+1,y
     sta next_piece_gfx+1
-    lda PIECES_NEXT_CHARS,x
-    sta next_piece_char
     lda #0
     sta l
   b3:
@@ -817,10 +822,15 @@ play_move_down: {
     tax
     jsr play_update_score
     jsr play_spawn_current
-    lda current_piece_gfx
+    ldy play_spawn_current._7
+    lda PIECES,y
     sta current_piece
-    lda current_piece_gfx+1
+    lda PIECES+1,y
     sta current_piece+1
+    lda PIECES,y
+    sta current_piece_gfx
+    lda PIECES+1,y
+    sta current_piece_gfx+1
     lda #0
     sta current_orientation
   b11:
@@ -838,16 +848,13 @@ play_move_down: {
 // Spawn a new piece
 // Moves the next piece into the current and spawns a new next piece
 play_spawn_current: {
+    .label _7 = 4
     .label piece_idx = $21
     // Move next piece into current
     ldx next_piece_idx
     txa
     asl
-    tay
-    lda PIECES,y
-    sta current_piece_gfx
-    lda PIECES+1,y
-    sta current_piece_gfx+1
+    sta _7
     lda PIECES_CHARS,x
     sta current_piece_char
     lda PIECES_START_X,x
@@ -858,9 +865,10 @@ play_spawn_current: {
     sta play_collision.xpos
     lda current_ypos
     sta play_collision.ypos
-    lda current_piece_gfx
+    ldy _7
+    lda PIECES,y
     sta current_piece_100
-    lda current_piece_gfx+1
+    lda PIECES+1,y
     sta current_piece_100+1
     ldx #0
     jsr play_collision
diff --git a/src/test/ref/constants.asm b/src/test/ref/constants.asm
index 7d603723c..ff7cb9046 100644
--- a/src/test/ref/constants.asm
+++ b/src/test/ref/constants.asm
@@ -17,7 +17,7 @@ main: {
 // Test different signed byte constants
 test_sbytes: {
     .const bb = 0
-    .const bc = bb+2
+    .const bc = 2
     .const bf = $ff&-$7f-$7f
     .const bd = bc-4
     .const be = -bd
@@ -147,7 +147,7 @@ print_ln: {
 // Test different byte constants
 test_bytes: {
     .const bb = 0
-    .const bc = bb+2
+    .const bc = 2
     .const bd = bc-4
     lda #<$400
     sta print_line_cursor
diff --git a/src/test/ref/deep-nesting.asm b/src/test/ref/deep-nesting.asm
index 0da9881d9..8637ea14c 100644
--- a/src/test/ref/deep-nesting.asm
+++ b/src/test/ref/deep-nesting.asm
@@ -10,7 +10,6 @@ main: {
     rts
 }
 f1: {
-    .label x = 0
     .label return = f2.return+1
     jsr f2
     rts
@@ -501,7 +500,7 @@ f98: {
     rts
 }
 f99: {
-    .label return = f1.x+1
+    .label return = 1
     jsr f100
     rts
 }
diff --git a/src/test/ref/examples/3d/3d.asm b/src/test/ref/examples/3d/3d.asm
index c555981f4..996c1d34e 100644
--- a/src/test/ref/examples/3d/3d.asm
+++ b/src/test/ref/examples/3d/3d.asm
@@ -97,7 +97,7 @@ anim: {
     cmp RASTER
     bne b4
     inc BORDERCOL
-    ldy sx
+    ldx sx
     jsr calculate_matrix
     jsr store_matrix
     lda #0
@@ -129,13 +129,13 @@ anim: {
     tya
     asl
     tax
-    lda xp
+    lda #$80
     clc
-    adc #$80
+    adc xp
     sta SPRITES_XPOS,x
-    lda yp
+    lda #$80
     clc
-    adc #$80
+    adc yp
     sta SPRITES_YPOS,x
     inc i
     lda #8
@@ -335,7 +335,8 @@ print_sbyte_at: {
     .label at = 6
     cpx #0
     bmi b1
-    ldy #' '
+    lda #' '
+    sta print_char_at.ch
     jsr print_char_at
   b2:
     inc print_byte_at.at
@@ -345,7 +346,8 @@ print_sbyte_at: {
     jsr print_byte_at
     rts
   b1:
-    ldy #'-'
+    lda #'-'
+    sta print_char_at.ch
     jsr print_char_at
     txa
     eor #$ff
@@ -355,16 +357,17 @@ print_sbyte_at: {
     jmp b2
 }
 // Print a single char
-// print_char_at(byte register(Y) ch, byte* zeropage(6) at)
+// print_char_at(byte zeropage(8) ch, byte* zeropage(6) at)
 print_char_at: {
     .label at = 6
-    tya
+    .label ch = 8
+    lda ch
     ldy #0
     sta (at),y
     rts
 }
 // Print a byte as HEX at a specific position
-// print_byte_at(byte register(X) b, byte* zeropage(6) at)
+// print_byte_at(byte* zeropage(6) at)
 print_byte_at: {
     .label at = 6
     txa
@@ -374,7 +377,7 @@ print_byte_at: {
     lsr
     tay
     lda print_hextab,y
-    tay
+    sta print_char_at.ch
     jsr print_char_at
     lda #$f
     axs #0
@@ -382,7 +385,8 @@ print_byte_at: {
     bne !+
     inc print_char_at.at+1
   !:
-    ldy print_hextab,x
+    lda print_hextab,x
+    sta print_char_at.ch
     jsr print_char_at
     rts
 }
@@ -390,10 +394,10 @@ print_byte_at: {
 // The rotation matrix is prepared by calling prepare_matrix() 
 // The passed points must be in the interval [-$3f;$3f].
 // Implemented in assembler to utilize seriously fast multiplication 
-// rotate_matrix(signed byte register(X) x, signed byte zeropage(5) y, signed byte zeropage($a) z)
+// rotate_matrix(signed byte register(X) x, signed byte zeropage(5) y, signed byte zeropage(8) z)
 rotate_matrix: {
     .label y = 5
-    .label z = $a
+    .label z = 8
     txa
     sta xr
     lda y
@@ -533,62 +537,61 @@ store_matrix: {
 // Prepare the 3x3 rotation matrix into rotation_matrix[]
 // Angles sx, sy, sz are based on 2*PI=$100 
 // Method described in C= Hacking Magazine Issue 8. http://www.ffd2.com/fridge/chacking/c=hacking8.txt
-// calculate_matrix(signed byte register(Y) sx, signed byte zeropage(3) sy)
+// calculate_matrix(signed byte register(X) sx, signed byte zeropage(3) sy)
 calculate_matrix: {
     .label sy = 3
     .label t1 = 4
-    .label t3 = 5
-    .label t4 = $a
-    .label t5 = $b
-    .label t6 = $c
-    .label t7 = $d
-    .label t8 = $e
-    .label t9 = $f
-    .label t10 = $10
-    lax sy
-    axs #sz
-    stx t1
-    ldx sy
-    sty t3
-    tya
-    sec
-    sbc #sz
-    sta t4
+    .label t2 = 5
+    .label t3 = 8
+    .label t4 = $b
+    .label t5 = $c
+    .label t6 = $d
+    .label t7 = $e
+    .label t8 = $f
+    .label t9 = $10
+    lda sy
+    sta t1
+    lda sy
+    sta t2
+    stx t3
+    stx t4
     txa
-    sty $ff
     clc
-    adc $ff
+    adc t2
     sta t5
-    tya
+    txa
     sec
     sbc t1
     sta t6
-    tya
+    txa
     clc
     adc t1
     sta t7
     txa
-    sty $ff
+    eor #$ff
     sec
-    sbc $ff
+    adc t2
     sta t8
-    tya
+    txa
     eor #$ff
     sec
     adc sy
     sta t9
-    tya
+    txa
     clc
     adc sy
-    sta t10
-    lda COSH,x
+    tax
     ldy t1
+    lda COSH,y
+    ldy t2
     clc
     adc COSH,y
     sta rotation_matrix
+    ldy t1
     lda SINH,y
+    ldy t2
     sec
-    sbc SINH,x
+    sbc SINH,y
     sta rotation_matrix+1
     ldy sy
     clc
@@ -633,9 +636,8 @@ calculate_matrix: {
     sta rotation_matrix+4
     ldy t9
     lda SINH,y
-    ldy t10
     sec
-    sbc SINH,y
+    sbc SINH,x
     sta rotation_matrix+5
     ldy t4
     lda COSH,y
@@ -673,9 +675,8 @@ calculate_matrix: {
     sec
     sbc COSQ,y
     sta rotation_matrix+7
+    lda COSH,x
     ldy t9
-    lda COSH,y
-    ldy t10
     clc
     adc COSH,y
     sta rotation_matrix+8
@@ -954,9 +955,9 @@ debug_print_init: {
     str11: .text "yp@"
 }
 // Print a string at a specific screen position
-// print_str_at(byte* zeropage(6) str, byte* zeropage(8) at)
+// print_str_at(byte* zeropage(6) str, byte* zeropage(9) at)
 print_str_at: {
-    .label at = 8
+    .label at = 9
     .label str = 6
   b1:
     ldy #0
diff --git a/src/test/ref/examples/bresenham/bitmap-bresenham.asm b/src/test/ref/examples/bresenham/bitmap-bresenham.asm
index 5c7a99cac..2f2268389 100644
--- a/src/test/ref/examples/bresenham/bitmap-bresenham.asm
+++ b/src/test/ref/examples/bresenham/bitmap-bresenham.asm
@@ -184,7 +184,6 @@ bitmap_line_xdyi: {
 }
 // bitmap_plot(byte register(X) x, byte register(Y) y)
 bitmap_plot: {
-    .label _0 = 9
     .label plotter_x = 9
     .label plotter_y = $b
     .label plotter = 9
@@ -196,13 +195,13 @@ bitmap_plot: {
     sta plotter_y+1
     lda bitmap_plot_ylo,y
     sta plotter_y
-    lda _0
+    lda plotter
     clc
     adc plotter_y
-    sta _0
-    lda _0+1
+    sta plotter
+    lda plotter+1
     adc plotter_y+1
-    sta _0+1
+    sta plotter+1
     lda bitmap_plot_bit,x
     ldy #0
     ora (plotter),y
@@ -336,11 +335,10 @@ init_screen: {
 bitmap_clear: {
     .label bitmap = 9
     .label y = 2
-    .label _3 = 9
     lda bitmap_plot_xlo
-    sta _3
+    sta bitmap
     lda bitmap_plot_xhi
-    sta _3+1
+    sta bitmap+1
     lda #0
     sta y
   b1:
diff --git a/src/test/ref/examples/fastmultiply/fastmultiply8.asm b/src/test/ref/examples/fastmultiply/fastmultiply8.asm
index b01b2b3b8..1ca0be223 100644
--- a/src/test/ref/examples/fastmultiply/fastmultiply8.asm
+++ b/src/test/ref/examples/fastmultiply/fastmultiply8.asm
@@ -145,11 +145,10 @@ print_char_at: {
     rts
 }
 // Print a byte as HEX at a specific position
-// print_byte_at(byte zeropage($a) b, byte* zeropage(8) at)
+// print_byte_at(byte* zeropage(8) at)
 print_byte_at: {
-    .label b = $a
     .label at = 8
-    lda b
+    lda print_sbyte_at.b
     lsr
     lsr
     lsr
@@ -159,7 +158,7 @@ print_byte_at: {
     sta print_char_at.ch
     jsr print_char_at
     lda #$f
-    and b
+    and print_sbyte_at.b
     tay
     inc print_char_at.at
     bne !+
diff --git a/src/test/ref/examples/rotate/rotate.asm b/src/test/ref/examples/rotate/rotate.asm
index 21947b062..c28ec7109 100644
--- a/src/test/ref/examples/rotate/rotate.asm
+++ b/src/test/ref/examples/rotate/rotate.asm
@@ -27,16 +27,10 @@ main: {
     rts
 }
 anim: {
-    .label _4 = 5
-    .label _6 = 5
-    .label _9 = 5
     .label _10 = 5
-    .label _11 = 5
     .label _12 = 5
-    .label cos_a = $b
-    .label sin_a = $c
-    .label x = $d
-    .label y = $e
+    .label x = $b
+    .label y = $c
     .label xr = 7
     .label yr = 9
     .label xpos = 5
@@ -50,11 +44,6 @@ anim: {
     cmp RASTER
     bne b2
     inc BORDERCOL
-    ldy angle
-    lda COS,y
-    sta cos_a
-    lda SIN,y
-    sta sin_a
     lda #0
     sta sprite_msb
     sta i
@@ -65,25 +54,27 @@ anim: {
     // signed fixed[7.0]
     lda ys,y
     sta y
-    lda cos_a
+    ldy angle
+    lda COS,y
     jsr mulf8u_prepare
     ldy x
     jsr mulf8s_prepared
-    lda _4
+    lda mulf8s_prepared.m
     asl
     sta xr
-    lda _4+1
+    lda mulf8s_prepared.m+1
     rol
     sta xr+1
     ldy y
     jsr mulf8s_prepared
-    lda _6
+    lda mulf8s_prepared.m
     asl
     sta yr
-    lda _6+1
+    lda mulf8s_prepared.m+1
     rol
     sta yr+1
-    lda sin_a
+    ldy angle
+    lda SIN,y
     jsr mulf8u_prepare
     ldy y
     jsr mulf8s_prepared
@@ -156,44 +147,34 @@ anim: {
 // mulf8s_prepared(signed byte register(Y) b)
 mulf8s_prepared: {
     .label memA = $fd
-    .label _8 = $f
-    .label _12 = $f
     .label m = 5
-    .label return = 5
-    tya
     jsr mulf8u_prepared
     lda memA
     cmp #0
     bpl b1
     lda m+1
-    sta _8
-    tya
-    eor #$ff
+    sty $ff
     sec
-    adc _8
+    sbc $ff
     sta m+1
   b1:
     cpy #0
     bpl b2
     lda m+1
-    sta _12
-    lda memA
-    eor #$ff
     sec
-    adc _12
+    sbc memA
     sta m+1
   b2:
     rts
 }
 // Calculate fast multiply with a prepared unsigned byte to a word result
 // The prepared number is set by calling mulf8u_prepare(byte a)
-// mulf8u_prepared(byte register(A) b)
 mulf8u_prepared: {
     .label resL = $fe
     .label memB = $ff
     .label return = 5
-    sta memB
-    tax
+    sty memB
+    ldx memB
     sec
   sm1:
     lda mulf_sqr1_lo,x
diff --git a/src/test/ref/examples/scrolllogo/scrolllogo.asm b/src/test/ref/examples/scrolllogo/scrolllogo.asm
index 46983d01e..9520973bf 100644
--- a/src/test/ref/examples/scrolllogo/scrolllogo.asm
+++ b/src/test/ref/examples/scrolllogo/scrolllogo.asm
@@ -120,8 +120,8 @@ loop: {
 render_logo: {
     .label _3 = $f
     .label xpos = 9
+    .label x_char = 4
     .label logo_idx = 4
-    .label logo_start = 4
     lda xpos
     and #7
     ora #VIC_MCM
@@ -142,21 +142,20 @@ render_logo: {
     ror _3+1
     ror _3
     lda _3
-    tax
+    sta x_char
     lda xpos+1
     bmi b1
-    stx logo_start
     ldy #0
-  b3:
-    cpy logo_start
-    bne b4
+  b2:
+    cpy x_char
+    bne b3
     lda #0
     sta logo_idx
-  b6:
+  b5:
     cpy #$28
-    bne b7
+    bne b6
     rts
-  b7:
+  b6:
     lda logo_idx
     sta SCREEN,y
     lda #$28*1
@@ -181,8 +180,8 @@ render_logo: {
     sta SCREEN+$28*5,y
     iny
     inc logo_idx
-    jmp b6
-  b4:
+    jmp b5
+  b3:
     lda #0
     sta SCREEN,y
     sta SCREEN+$28*1,y
@@ -191,23 +190,23 @@ render_logo: {
     sta SCREEN+$28*4,y
     sta SCREEN+$28*5,y
     iny
-    jmp b3
+    jmp b2
   b1:
-    txa
+    lda x_char
     eor #$ff
     clc
     adc #1
     sta logo_idx
     ldy #0
-  b9:
+  b8:
     lda #$28
     cmp logo_idx
-    bne b10
-  b12:
+    bne b9
+  b11:
     cpy #$28
-    bne b13
+    bne b12
     rts
-  b13:
+  b12:
     lda #0
     sta SCREEN,y
     sta SCREEN+$28*1,y
@@ -216,8 +215,8 @@ render_logo: {
     sta SCREEN+$28*4,y
     sta SCREEN+$28*5,y
     iny
-    jmp b12
-  b10:
+    jmp b11
+  b9:
     lda logo_idx
     sta SCREEN,y
     lda #$28*1
@@ -242,7 +241,7 @@ render_logo: {
     sta SCREEN+$28*5,y
     iny
     inc logo_idx
-    jmp b9
+    jmp b8
 }
 // Generate signed word sinus table - with values in the range min-max.
 // sintab - the table to generate into
@@ -254,8 +253,7 @@ sin16s_gen2: {
     .label ampl = max-min
     .label _5 = $b
     .label _6 = $f
-    .label _8 = $f
-    .label step = $19
+    .label step = $1b
     .label sintab = 2
     .label x = 5
     .label i = 9
@@ -289,10 +287,10 @@ sin16s_gen2: {
     lda _5+3
     sta _6+1
     ldy #0
-    lda _8
+    lda _6
     sta (sintab),y
     iny
-    lda _8+1
+    lda _6+1
     sta (sintab),y
     lda #SIZEOF_SIGNED_WORD
     clc
@@ -330,13 +328,13 @@ sin16s_gen2: {
 }
 // Multiply of two signed words to a signed double word
 // Fixes offsets introduced by using unsigned multiplication
-// mul16s(signed word zeropage($f) a)
+// mul16s(signed word zeropage($17) a)
 mul16s: {
     .label _9 = $f
     .label _16 = $f
     .label m = $b
     .label return = $b
-    .label a = $f
+    .label a = $17
     lda a
     sta mul16u.a
     lda a+1
@@ -423,18 +421,17 @@ mul16u: {
 // sin16s(dword zeropage($b) x)
 sin16s: {
     .label _4 = $b
-    .label _20 = $f
     .label x = $b
-    .label return = $f
-    .label x1 = $1d
-    .label x2 = $17
-    .label x3 = $17
+    .label return = $17
+    .label x1 = $1f
+    .label x2 = $19
+    .label x3 = $19
     .label x3_6 = $f
-    .label usinx = $1f
-    .label x4 = $17
+    .label usinx = $17
+    .label x4 = $19
     .label x5 = $f
     .label x5_128 = $f
-    .label sinx = $f
+    .label sinx = $17
     .label isUpper = 4
     lda x+3
     cmp #>PI_u4f28>>$10
@@ -580,17 +577,9 @@ sin16s: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _20
-    lda usinx+1
-    sta _20+1
     sec
     lda sinx
     eor #$ff
@@ -605,15 +594,15 @@ sin16s: {
 }
 // Calculate val*val for two unsigned word values - the result is 16 selected bits of the 32-bit result.
 // The select parameter indicates how many of the highest bits of the 32-bit result to skip
-// mulu16_sel(word zeropage($17) v1, word zeropage($f) v2, byte register(X) select)
+// mulu16_sel(word zeropage($19) v1, word zeropage($f) v2, byte register(X) select)
 mulu16_sel: {
     .label _0 = $b
     .label _1 = $b
-    .label v1 = $17
+    .label v1 = $19
     .label v2 = $f
     .label return = $f
-    .label return_1 = $17
-    .label return_10 = $17
+    .label return_1 = $19
+    .label return_10 = $19
     lda v1
     sta mul16u.a
     lda v1+1
@@ -647,7 +636,7 @@ mulu16_sel: {
 div32u16u: {
     .label quotient_hi = $11
     .label quotient_lo = $f
-    .label return = $19
+    .label return = $1b
     lda #<PI2_u4f28>>$10
     sta divr16u.dividend
     lda #>PI2_u4f28>>$10
diff --git a/src/test/ref/examples/sinplotter/sine-plotter.asm b/src/test/ref/examples/sinplotter/sine-plotter.asm
index a7f77bdfb..5dfc8439a 100644
--- a/src/test/ref/examples/sinplotter/sine-plotter.asm
+++ b/src/test/ref/examples/sinplotter/sine-plotter.asm
@@ -176,7 +176,6 @@ render_sine: {
 bitmap_plot: {
     .label _1 = $10
     .label plotter = 6
-    .label plotter_1 = $10
     .label x = 4
     .label _3 = 6
     lda bitmap_plot_yhi,x
@@ -189,19 +188,19 @@ bitmap_plot: {
     lda x+1
     and #>$fff8
     sta _1+1
-    lda plotter_1
+    lda plotter
     clc
-    adc plotter
-    sta plotter_1
-    lda plotter_1+1
-    adc plotter+1
-    sta plotter_1+1
+    adc _1
+    sta plotter
+    lda plotter+1
+    adc _1+1
+    sta plotter+1
     lda x
     tay
     lda bitmap_plot_bit,y
     ldy #0
-    ora (plotter_1),y
-    sta (plotter_1),y
+    ora (plotter),y
+    sta (plotter),y
     rts
 }
 // wrap_y(signed word zeropage(6) y)
@@ -250,8 +249,7 @@ sin16s_gen2: {
     .label ampl = max-min
     .label _5 = $c
     .label _6 = 6
-    .label _8 = 6
-    .label step = $19
+    .label step = $1b
     .label sintab = 2
     .label x = 8
     .label i = 4
@@ -285,10 +283,10 @@ sin16s_gen2: {
     lda _5+3
     sta _6+1
     ldy #0
-    lda _8
+    lda _6
     sta (sintab),y
     iny
-    lda _8+1
+    lda _6+1
     sta (sintab),y
     lda #SIZEOF_SIGNED_WORD
     clc
@@ -326,13 +324,13 @@ sin16s_gen2: {
 }
 // Multiply of two signed words to a signed double word
 // Fixes offsets introduced by using unsigned multiplication
-// mul16s(signed word zeropage(6) a)
+// mul16s(signed word zeropage($17) a)
 mul16s: {
     .label _9 = 6
     .label _16 = 6
     .label m = $c
     .label return = $c
-    .label a = 6
+    .label a = $17
     lda a
     sta mul16u.a
     lda a+1
@@ -419,18 +417,17 @@ mul16u: {
 // sin16s(dword zeropage($c) x)
 sin16s: {
     .label _4 = $c
-    .label _20 = 6
     .label x = $c
-    .label return = 6
-    .label x1 = $1d
-    .label x2 = $17
-    .label x3 = $17
+    .label return = $17
+    .label x1 = $1f
+    .label x2 = $19
+    .label x3 = $19
     .label x3_6 = 6
-    .label usinx = $1f
-    .label x4 = $17
+    .label usinx = $17
+    .label x4 = $19
     .label x5 = 6
     .label x5_128 = 6
-    .label sinx = 6
+    .label sinx = $17
     .label isUpper = $16
     lda x+3
     cmp #>PI_u4f28>>$10
@@ -576,17 +573,9 @@ sin16s: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _20
-    lda usinx+1
-    sta _20+1
     sec
     lda sinx
     eor #$ff
@@ -601,15 +590,15 @@ sin16s: {
 }
 // Calculate val*val for two unsigned word values - the result is 16 selected bits of the 32-bit result.
 // The select parameter indicates how many of the highest bits of the 32-bit result to skip
-// mulu16_sel(word zeropage($17) v1, word zeropage(6) v2, byte register(X) select)
+// mulu16_sel(word zeropage($19) v1, word zeropage(6) v2, byte register(X) select)
 mulu16_sel: {
     .label _0 = $c
     .label _1 = $c
-    .label v1 = $17
+    .label v1 = $19
     .label v2 = 6
     .label return = 6
-    .label return_1 = $17
-    .label return_10 = $17
+    .label return_1 = $19
+    .label return_10 = $19
     lda v1
     sta mul16u.a
     lda v1+1
@@ -643,7 +632,7 @@ mulu16_sel: {
 div32u16u: {
     .label quotient_hi = $10
     .label quotient_lo = 6
-    .label return = $19
+    .label return = $1b
     lda #<PI2_u4f28>>$10
     sta divr16u.dividend
     lda #>PI2_u4f28>>$10
diff --git a/src/test/ref/inline-pointer-1.asm b/src/test/ref/inline-pointer-1.asm
index 3bf16934c..a3fa5bb0b 100644
--- a/src/test/ref/inline-pointer-1.asm
+++ b/src/test/ref/inline-pointer-1.asm
@@ -14,9 +14,8 @@ main: {
 // puta(byte register(A) ph, byte register(X) pl)
 puta: {
     .label screen = 2
-    .label _1 = 2
-    sta _1+1
-    stx _1
+    sta screen+1
+    stx screen
     lda #'a'
     ldy #0
     sta (screen),y
diff --git a/src/test/ref/inline-string-3.asm b/src/test/ref/inline-string-3.asm
index 1881a03a3..def27f5f2 100644
--- a/src/test/ref/inline-string-3.asm
+++ b/src/test/ref/inline-string-3.asm
@@ -8,15 +8,14 @@ main: {
     .label PTR = $9ffe
     .label SCREEN = $400
     .label ptr = 2
-    .label _6 = 2
     lda #<STRING
     sta PTR
     lda #>STRING
     sta PTR+1
     lda PTR
-    sta _6
+    sta ptr
     lda PTR+1
-    sta _6+1
+    sta ptr+1
     ldy #0
     lda (ptr),y
     sta SCREEN
diff --git a/src/test/ref/inline-word.asm b/src/test/ref/inline-word.asm
index 7b80c88b1..505c1ebc1 100644
--- a/src/test/ref/inline-word.asm
+++ b/src/test/ref/inline-word.asm
@@ -4,7 +4,6 @@
   .label SCREEN = $400
 main: {
     .label w = 3
-    .label sc = 3
     .label h = 2
     lda #0
     sta h
@@ -18,7 +17,7 @@ main: {
     stx w
     lda #'*'
     ldy #0
-    sta (sc),y
+    sta (w),y
     inx
     cpx #8
     bne b2
diff --git a/src/test/ref/line-anim.asm b/src/test/ref/line-anim.asm
index 862d9a41b..b525f2daf 100644
--- a/src/test/ref/line-anim.asm
+++ b/src/test/ref/line-anim.asm
@@ -80,38 +80,34 @@ bitmap_plot: {
     .label _1 = 7
     .label x = 3
     .label plotter = 5
-    .label plotter_1 = 7
-    .label _3 = 5
     lda bitmap_plot_yhi,x
-    sta _3+1
+    sta plotter+1
     lda bitmap_plot_ylo,x
-    sta _3
+    sta plotter
     lda x
     and #<$fff8
     sta _1
     lda x+1
     and #>$fff8
     sta _1+1
-    lda plotter_1
+    lda plotter
     clc
-    adc plotter
-    sta plotter_1
-    lda plotter_1+1
-    adc plotter+1
-    sta plotter_1+1
+    adc _1
+    sta plotter
+    lda plotter+1
+    adc _1+1
+    sta plotter+1
     lda x
     tay
     lda bitmap_plot_bit,y
     ldy #0
-    ora (plotter_1),y
-    sta (plotter_1),y
+    ora (plotter),y
+    sta (plotter),y
     rts
 }
 // Initialize the points to be animated
 // point_init(byte zeropage(2) point_idx)
 point_init: {
-    .label _0 = 9
-    .label _1 = 3
     .label _3 = 7
     .label _4 = 3
     .label _9 = 3
@@ -119,29 +115,19 @@ point_init: {
     .label _11 = 3
     .label point_idx = 2
     .label y_diff = 7
-    .label abs16s1__2 = 3
     .label abs16s1_return = 3
-    .label abs16s2__2 = 5
     .label abs16s2_return = 5
     .label x_stepf = 3
     .label x_diff = 9
     lda point_idx
     asl
-    tax
-    lda x_end,x
-    sta _0
-    lda x_end+1,x
-    sta _0+1
-    lda x_start,x
-    sta _1
-    lda x_start+1,x
-    sta _1+1
-    lda x_diff
+    tay
     sec
-    sbc _1
+    lda x_end,y
+    sbc x_start,y
     sta x_diff
-    lda x_diff+1
-    sbc _1+1
+    lda x_end+1,y
+    sbc x_start+1,y
     sta x_diff+1
     ldy point_idx
     lda y_end,y
@@ -254,22 +240,22 @@ point_init: {
     lda y_diff
     eor #$ff
     adc #0
-    sta abs16s2__2
+    sta abs16s2_return
     lda y_diff+1
     eor #$ff
     adc #0
-    sta abs16s2__2+1
+    sta abs16s2_return+1
     jmp b6
   abs16s1_b1:
     sec
     lda x_diff
     eor #$ff
     adc #0
-    sta abs16s1__2
+    sta abs16s1_return
     lda x_diff+1
     eor #$ff
     adc #0
-    sta abs16s1__2+1
+    sta abs16s1_return+1
     jmp abs16s2
 }
 // Perform division on two signed 16-bit numbers with an initial remainder.
@@ -279,9 +265,6 @@ point_init: {
 // See http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf section 6.5.5
 // divr16s(signed word zeropage(9) divisor, signed word zeropage(7) rem)
 divr16s: {
-    .label _10 = 7
-    .label _13 = 9
-    .label _18 = 3
     .label remu = 7
     .label divisoru = 9
     .label resultu = 3
@@ -311,28 +294,28 @@ divr16s: {
     rts
   b3:
     sec
-    lda _13
+    lda divisoru
     eor #$ff
     adc #0
-    sta _13
-    lda _13+1
+    sta divisoru
+    lda divisoru+1
     eor #$ff
     adc #0
-    sta _13+1
+    sta divisoru+1
     tya
     eor #1
     tay
     jmp b4
   b1:
     sec
-    lda _10
+    lda remu
     eor #$ff
     adc #0
-    sta _10
-    lda _10+1
+    sta remu
+    lda remu+1
     eor #$ff
     adc #0
-    sta _10+1
+    sta remu+1
     ldy #1
     jmp b2
 }
@@ -428,11 +411,10 @@ screen_fill: {
 bitmap_clear: {
     .label bitmap = 3
     .label y = 2
-    .label _3 = 3
     lda bitmap_plot_ylo
-    sta _3
+    sta bitmap
     lda bitmap_plot_yhi
-    sta _3+1
+    sta bitmap+1
     lda #0
     sta y
   b1:
diff --git a/src/test/ref/signed-indexed-subtract.asm b/src/test/ref/signed-indexed-subtract.asm
index 8904aed68..e82d13764 100644
--- a/src/test/ref/signed-indexed-subtract.asm
+++ b/src/test/ref/signed-indexed-subtract.asm
@@ -92,13 +92,11 @@ print_sword: {
     rts
 }
 // Print a word as HEX
-// print_word(word zeropage(4) w)
 print_word: {
-    .label w = 4
-    lda w+1
+    lda print_sword.w+1
     sta print_byte.b
     jsr print_byte
-    lda w
+    lda print_sword.w
     sta print_byte.b
     jsr print_byte
     rts
diff --git a/src/test/ref/sinusgen16.asm b/src/test/ref/sinusgen16.asm
index bdca50e5c..c2c799b3f 100644
--- a/src/test/ref/sinusgen16.asm
+++ b/src/test/ref/sinusgen16.asm
@@ -112,13 +112,11 @@ print_sword: {
     rts
 }
 // Print a word as HEX
-// print_word(word zeropage(6) w)
 print_word: {
-    .label w = 6
-    lda w+1
+    lda print_sword.w+1
     tax
     jsr print_byte
-    lda w
+    lda print_sword.w
     tax
     jsr print_byte
     rts
@@ -180,7 +178,7 @@ print_cls: {
 // sin16s_gen(signed word* zeropage(2) sintab)
 sin16s_gen: {
     .label _1 = 6
-    .label step = $19
+    .label step = $1b
     .label sintab = 2
     .label x = $a
     .label i = 4
@@ -254,17 +252,16 @@ sin16s_gen: {
 // sin16s(dword zeropage($f) x)
 sin16s: {
     .label _4 = $f
-    .label _20 = 6
     .label x = $f
     .label return = 6
-    .label x1 = $1d
-    .label x2 = 6
-    .label x3 = 6
-    .label x3_6 = 8
-    .label usinx = $1f
-    .label x4 = 6
-    .label x5 = 8
-    .label x5_128 = 8
+    .label x1 = $1f
+    .label x2 = 8
+    .label x3 = 8
+    .label x3_6 = $13
+    .label usinx = 6
+    .label x4 = 8
+    .label x5 = $13
+    .label x5_128 = $13
     .label sinx = 6
     .label isUpper = $e
     lda x+3
@@ -411,17 +408,9 @@ sin16s: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _20
-    lda usinx+1
-    sta _20+1
     sec
     lda sinx
     eor #$ff
@@ -436,15 +425,15 @@ sin16s: {
 }
 // Calculate val*val for two unsigned word values - the result is 16 selected bits of the 32-bit result.
 // The select parameter indicates how many of the highest bits of the 32-bit result to skip
-// mulu16_sel(word zeropage(6) v1, word zeropage(8) v2, byte register(X) select)
+// mulu16_sel(word zeropage(8) v1, word zeropage($13) v2, byte register(X) select)
 mulu16_sel: {
     .label _0 = $f
     .label _1 = $f
-    .label v1 = 6
-    .label v2 = 8
-    .label return = 8
-    .label return_1 = 6
-    .label return_10 = 6
+    .label v1 = 8
+    .label v2 = $13
+    .label return = $13
+    .label return_1 = 8
+    .label return_10 = 8
     lda v1
     sta mul16u.a
     lda v1+1
@@ -467,12 +456,12 @@ mulu16_sel: {
     rts
 }
 // Perform binary multiplication of two unsigned 16-bit words into a 32-bit unsigned double word
-// mul16u(word zeropage($13) a, word zeropage(8) b)
+// mul16u(word zeropage($15) a, word zeropage($13) b)
 mul16u: {
-    .label a = $13
-    .label mb = $15
+    .label a = $15
+    .label mb = $17
     .label res = $f
-    .label b = 8
+    .label b = $13
     .label return = $f
     lda b
     sta mb
@@ -524,7 +513,7 @@ mul16u: {
 div32u16u: {
     .label quotient_hi = 8
     .label quotient_lo = 6
-    .label return = $19
+    .label return = $1b
     lda #<PI2_u4f28>>$10
     sta divr16u.dividend
     lda #>PI2_u4f28>>$10
diff --git a/src/test/ref/sinusgen16b.asm b/src/test/ref/sinusgen16b.asm
index 299b22494..1d36006bf 100644
--- a/src/test/ref/sinusgen16b.asm
+++ b/src/test/ref/sinusgen16b.asm
@@ -129,13 +129,11 @@ print_sword: {
     rts
 }
 // Print a word as HEX
-// print_word(word zeropage(8) w)
 print_word: {
-    .label w = 8
-    lda w+1
+    lda print_sword.w+1
     sta print_byte.b
     jsr print_byte
-    lda w
+    lda print_sword.w
     sta print_byte.b
     jsr print_byte
     rts
@@ -198,8 +196,8 @@ print_cls: {
 // wavelength - the number of sinus points in a total sinus wavelength (the size of the table)
 // sin16s_genb(signed word* zeropage(2) sintab)
 sin16s_genb: {
-    .label _2 = 6
-    .label step = $1b
+    .label _2 = 8
+    .label step = $1d
     .label sintab = 2
     .label x = $d
     .label i = 4
@@ -268,18 +266,17 @@ sin16s_genb: {
 // result: signed word sin(x) s[0.15] - using the full range  -$7fff - $7fff
 // sin16sb(word zeropage(6) x)
 sin16sb: {
-    .label _19 = 6
     .label x = 6
-    .label return = 6
+    .label return = 8
     .label x1 = 6
-    .label x2 = 8
-    .label x3 = 8
-    .label x3_6 = $b
-    .label usinx = $1f
-    .label x4 = 8
-    .label x5 = $b
-    .label x5_128 = $b
-    .label sinx = 6
+    .label x2 = $b
+    .label x3 = $b
+    .label x3_6 = $11
+    .label usinx = 8
+    .label x4 = $b
+    .label x5 = $11
+    .label x5_128 = $11
+    .label sinx = 8
     .label isUpper = $a
     lda x+1
     cmp #>PI_u4f12
@@ -391,17 +388,9 @@ sin16sb: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _19
-    lda usinx+1
-    sta _19+1
     sec
     lda sinx
     eor #$ff
@@ -416,19 +405,19 @@ sin16sb: {
 }
 // Calculate val*val for two unsigned word values - the result is 16 selected bits of the 32-bit result.
 // The select parameter indicates how many of the highest bits of the 32-bit result to skip
-// mulu16_sel(word zeropage(8) v1, word zeropage($b) v2, byte register(X) select)
+// mulu16_sel(word zeropage($b) v1, word zeropage($11) v2, byte register(X) select)
 mulu16_sel: {
-    .label _0 = $13
-    .label _1 = $13
-    .label v1 = 8
-    .label v2 = $b
-    .label return = 8
-    .label return_11 = $b
-    .label return_14 = $b
-    .label return_16 = $b
-    .label return_17 = $b
-    .label return_18 = $b
-    .label return_20 = $b
+    .label _0 = $15
+    .label _1 = $15
+    .label v1 = $b
+    .label v2 = $11
+    .label return = $b
+    .label return_11 = $11
+    .label return_14 = $11
+    .label return_16 = $11
+    .label return_17 = $11
+    .label return_18 = $11
+    .label return_20 = $11
     lda v1
     sta mul16u.a
     lda v1+1
@@ -451,13 +440,13 @@ mulu16_sel: {
     rts
 }
 // Perform binary multiplication of two unsigned 16-bit words into a 32-bit unsigned double word
-// mul16u(word zeropage($11) a, word zeropage($b) b)
+// mul16u(word zeropage($13) a, word zeropage($11) b)
 mul16u: {
-    .label a = $11
-    .label mb = $17
-    .label res = $13
-    .label b = $b
-    .label return = $13
+    .label a = $13
+    .label mb = $19
+    .label res = $15
+    .label b = $11
+    .label return = $15
     lda b
     sta mb
     lda b+1
@@ -508,7 +497,7 @@ mul16u: {
 div32u16u: {
     .label quotient_hi = 8
     .label quotient_lo = 6
-    .label return = $1b
+    .label return = $1d
     lda #<PI2_u4f28>>$10
     sta divr16u.dividend
     lda #>PI2_u4f28>>$10
@@ -596,7 +585,7 @@ divr16u: {
 // sin16s_gen(signed word* zeropage(2) sintab)
 sin16s_gen: {
     .label _1 = 6
-    .label step = $1b
+    .label step = $1d
     .label sintab = 2
     .label x = $d
     .label i = 4
@@ -667,20 +656,19 @@ sin16s_gen: {
 // Calculate signed word sinus sin(x)
 // x: unsigned dword input u[4.28] in the interval $00000000 - PI2_u4f28
 // result: signed word sin(x) s[0.15] - using the full range  -$7fff - $7fff
-// sin16s(dword zeropage($13) x)
+// sin16s(dword zeropage($15) x)
 sin16s: {
-    .label _4 = $13
-    .label _20 = 6
-    .label x = $13
+    .label _4 = $15
+    .label x = $15
     .label return = 6
-    .label x1 = 6
-    .label x2 = 8
-    .label x3 = 8
-    .label x3_6 = $b
-    .label usinx = $1f
-    .label x4 = 8
-    .label x5 = $b
-    .label x5_128 = $b
+    .label x1 = 8
+    .label x2 = $b
+    .label x3 = $b
+    .label x3_6 = $11
+    .label usinx = 6
+    .label x4 = $b
+    .label x5 = $11
+    .label x5_128 = $11
     .label sinx = 6
     .label isUpper = $a
     lda x+3
@@ -827,17 +815,9 @@ sin16s: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _20
-    lda usinx+1
-    sta _20+1
     sec
     lda sinx
     eor #$ff
diff --git a/src/test/ref/sinusgen8.asm b/src/test/ref/sinusgen8.asm
index 5e117690a..9c3cdd67e 100644
--- a/src/test/ref/sinusgen8.asm
+++ b/src/test/ref/sinusgen8.asm
@@ -11,7 +11,6 @@
   .label print_char_cursor = 5
 main: {
     .label wavelength = $c0
-    .label _2 = 4
     .label sb = 4
     jsr sin8s_gen
     jsr print_cls
@@ -21,11 +20,9 @@ main: {
     sta print_char_cursor+1
     ldx #0
   b1:
-    lda sintabref,x
-    sta _2
     lda sintab2,x
     sec
-    sbc sb
+    sbc sintabref,x
     sta sb
     bmi b2
     lda #<str1
@@ -107,10 +104,8 @@ print_char: {
     rts
 }
 // Print a byte as HEX
-// print_byte(byte zeropage(4) b)
 print_byte: {
-    .label b = 4
-    lda b
+    lda print_sbyte.b
     lsr
     lsr
     lsr
@@ -119,7 +114,7 @@ print_byte: {
     lda print_hextab,y
     jsr print_char
     lda #$f
-    and b
+    and print_sbyte.b
     tay
     lda print_hextab,y
     jsr print_char
@@ -175,7 +170,6 @@ sin8s_gen: {
     lda x+1
     sta sin8s.x+1
     jsr sin8s
-    tya
     ldy #0
     sta (sintab),y
     inc sintab
@@ -303,17 +297,16 @@ sin8s: {
     bcc b3
     dex
   b3:
-    txa
-    tay
     lda isUpper
     cmp #0
-    beq b4
+    beq b14
     txa
     eor #$ff
     clc
     adc #1
-    tay
-  b4:
+    rts
+  b14:
+    txa
     rts
 }
 // Calculate val*val for two unsigned byte values - the result is 8 selected bits of the 16-bit result.
diff --git a/src/test/ref/sinusgen8b.asm b/src/test/ref/sinusgen8b.asm
index 67172c57e..497351206 100644
--- a/src/test/ref/sinusgen8b.asm
+++ b/src/test/ref/sinusgen8b.asm
@@ -129,10 +129,8 @@ print_char: {
     rts
 }
 // Print a byte as HEX
-// print_byte(byte zeropage(4) b)
 print_byte: {
-    .label b = 4
-    lda b
+    lda print_sbyte.b
     lsr
     lsr
     lsr
@@ -141,7 +139,7 @@ print_byte: {
     lda print_hextab,y
     jsr print_char
     lda #$f
-    and b
+    and print_sbyte.b
     tay
     lda print_hextab,y
     jsr print_char
@@ -250,17 +248,16 @@ sin16s_gen: {
 // sin16s(dword zeropage($b) x)
 sin16s: {
     .label _4 = $b
-    .label _20 = $f
     .label x = $b
     .label return = $f
-    .label x1 = $1a
-    .label x2 = $f
-    .label x3 = $f
-    .label x3_6 = $11
-    .label usinx = $20
-    .label x4 = $f
-    .label x5 = $11
-    .label x5_128 = $11
+    .label x1 = $20
+    .label x2 = $11
+    .label x3 = $11
+    .label x3_6 = $13
+    .label usinx = $f
+    .label x4 = $11
+    .label x5 = $13
+    .label x5_128 = $13
     .label sinx = $f
     .label isUpper = 4
     lda x+3
@@ -407,17 +404,9 @@ sin16s: {
     lda usinx+1
     adc x5_128+1
     sta usinx+1
-    lda usinx
-    sta sinx
-    lda usinx+1
-    sta sinx+1
     lda isUpper
     cmp #0
     beq b3
-    lda usinx
-    sta _20
-    lda usinx+1
-    sta _20+1
     sec
     lda sinx
     eor #$ff
@@ -432,15 +421,15 @@ sin16s: {
 }
 // Calculate val*val for two unsigned word values - the result is 16 selected bits of the 32-bit result.
 // The select parameter indicates how many of the highest bits of the 32-bit result to skip
-// mulu16_sel(word zeropage($f) v1, word zeropage($11) v2, byte register(X) select)
+// mulu16_sel(word zeropage($11) v1, word zeropage($13) v2, byte register(X) select)
 mulu16_sel: {
     .label _0 = $b
     .label _1 = $b
-    .label v1 = $f
-    .label v2 = $11
-    .label return = $11
-    .label return_1 = $f
-    .label return_10 = $f
+    .label v1 = $11
+    .label v2 = $13
+    .label return = $13
+    .label return_1 = $11
+    .label return_10 = $11
     lda v1
     sta mul16u.a
     lda v1+1
@@ -463,12 +452,12 @@ mulu16_sel: {
     rts
 }
 // Perform binary multiplication of two unsigned 16-bit words into a 32-bit unsigned double word
-// mul16u(word zeropage($13) a, word zeropage($11) b)
+// mul16u(word zeropage($15) a, word zeropage($13) b)
 mul16u: {
-    .label a = $13
-    .label mb = $15
+    .label a = $15
+    .label mb = $17
     .label res = $b
-    .label b = $11
+    .label b = $13
     .label return = $b
     lda b
     sta mb
@@ -629,7 +618,6 @@ sin8s_gen: {
     lda x+1
     sta sin8s.x+1
     jsr sin8s
-    tya
     ldy #0
     sta (sintab),y
     inc sintab
@@ -757,26 +745,25 @@ sin8s: {
     bcc b3
     dex
   b3:
-    txa
-    tay
     lda isUpper
     cmp #0
-    beq b4
+    beq b14
     txa
     eor #$ff
     clc
     adc #1
-    tay
-  b4:
+    rts
+  b14:
+    txa
     rts
 }
 // Calculate val*val for two unsigned byte values - the result is 8 selected bits of the 16-bit result.
 // The select parameter indicates how many of the highest bits of the 16-bit result to skip
-// mulu8_sel(byte register(X) v1, byte register(Y) v2, byte zeropage($19) select)
+// mulu8_sel(byte register(X) v1, byte register(Y) v2, byte zeropage($1b) select)
 mulu8_sel: {
     .label _0 = $13
     .label _1 = $13
-    .label select = $19
+    .label select = $1b
     tya
     jsr mul8u
     ldy select
@@ -793,7 +780,7 @@ mulu8_sel: {
 // Perform binary multiplication of two unsigned 8-bit bytes into a 16-bit unsigned word
 // mul8u(byte register(X) a, byte register(A) b)
 mul8u: {
-    .label mb = $1a
+    .label mb = $15
     .label res = $13
     .label return = $13
     sta mb
diff --git a/src/test/ref/sinusgenscale8.asm b/src/test/ref/sinusgenscale8.asm
index ba40b7df7..fea8b38e9 100644
--- a/src/test/ref/sinusgenscale8.asm
+++ b/src/test/ref/sinusgenscale8.asm
@@ -7,7 +7,7 @@
   .const PI_u4f12 = $3244
   // PI/2 in u[4.12] format
   .const PI_HALF_u4f12 = $1922
-  .label print_char_cursor = $d
+  .label print_char_cursor = $f
   .label print_line_cursor = 8
 main: {
     .label tabsize = $14
@@ -30,7 +30,6 @@ sin8u_table: {
     .const mid = sum/2+1
     .label step = $12
     .label sinx = $11
-    .label sinx_sc = $f
     .label sintab = 4
     .label x = 2
     .label i = 6
@@ -103,9 +102,10 @@ sin8u_table: {
     lda x+1
     sta sin8s.x+1
     jsr sin8s
-    sty sinx
+    sta sinx
+    tay
     jsr mul8su
-    lda sinx_sc+1
+    lda mul8su.m+1
     tax
     axs #-[mid]
     txa
@@ -142,10 +142,6 @@ sin8u_table: {
     lda #>str7
     sta print_str.str+1
     jsr print_str
-    lda sinx_sc
-    sta print_sword.w
-    lda sinx_sc+1
-    sta print_sword.w+1
     jsr print_sword
     lda #<str8
     sta print_str.str
@@ -264,10 +260,10 @@ print_str: {
     jmp b1
 }
 // Print a signed word as HEX
-// print_sword(signed word zeropage($b) w)
+// print_sword(signed word zeropage($d) w)
 print_sword: {
-    .label w = $b
-    lda w+1
+    .label w = $d
+    lda mul8su.m+1
     bpl b1
     lda #'-'
     jsr print_char
@@ -281,6 +277,10 @@ print_sword: {
     adc #0
     sta w+1
   b1:
+    lda w
+    sta print_word.w
+    lda w+1
+    sta print_word.w+1
     jsr print_word
     rts
 }
@@ -322,8 +322,7 @@ print_sbyte: {
 // mul8su(signed byte register(Y) a)
 mul8su: {
     .const b = sin8u_table.amplitude+1
-    .label m = $f
-    .label return = $f
+    .label m = $d
     tya
     tax
     lda #b
@@ -344,8 +343,8 @@ mul8su: {
 // mul8u(byte register(X) a, byte register(A) b)
 mul8u: {
     .label mb = $b
-    .label res = $f
-    .label return = $f
+    .label res = $d
+    .label return = $d
     lda #0
     sta res
     sta res+1
@@ -473,25 +472,24 @@ sin8s: {
     bcc b3
     dex
   b3:
-    txa
-    tay
     lda isUpper
     cmp #0
-    beq b4
+    beq b14
     txa
     eor #$ff
     clc
     adc #1
-    tay
-  b4:
+    rts
+  b14:
+    txa
     rts
 }
 // Calculate val*val for two unsigned byte values - the result is 8 selected bits of the 16-bit result.
 // The select parameter indicates how many of the highest bits of the 16-bit result to skip
 // mulu8_sel(byte register(X) v1, byte register(Y) v2, byte zeropage($11) select)
 mulu8_sel: {
-    .label _0 = $f
-    .label _1 = $f
+    .label _0 = $d
+    .label _1 = $d
     .label select = $11
     tya
     sta mul8u.mb
diff --git a/src/test/ref/sizeof-types.asm b/src/test/ref/sizeof-types.asm
index 24bc5e8b9..131b64694 100644
--- a/src/test/ref/sizeof-types.asm
+++ b/src/test/ref/sizeof-types.asm
@@ -2,7 +2,6 @@
 .pc = $801 "Basic"
 :BasicUpstart(main)
 .pc = $80d "Program"
-  .const SIZEOF_VOID = 0
   .const SIZEOF_BYTE = 1
   .const SIZEOF_SIGNED_BYTE = 1
   .const SIZEOF_BOOL = 1
@@ -13,7 +12,7 @@
   .const SIZEOF_SIGNED_DWORD = 4
   .label SCREEN = $400
 main: {
-    lda #'0'+SIZEOF_VOID
+    lda #'0'
     sta SCREEN
     lda #'0'+SIZEOF_BYTE
     sta SCREEN+2
diff --git a/src/test/ref/struct-4.asm b/src/test/ref/struct-4.asm
index d8901027f..13f78da1c 100644
--- a/src/test/ref/struct-4.asm
+++ b/src/test/ref/struct-4.asm
@@ -2,7 +2,6 @@
 .pc = $801 "Basic"
 :BasicUpstart(main)
 .pc = $80d "Program"
-  .const OFFS_X = 0
   .const OFFS_Y = 1
 main: {
     .label SCREEN = $400
@@ -12,7 +11,7 @@ main: {
     asl
     tay
     txa
-    sta points+OFFS_X,y
+    sta points,y
     txa
     clc
     adc #4
@@ -26,7 +25,7 @@ main: {
     tya
     asl
     tax
-    lda points+OFFS_X,x
+    lda points,x
     sta SCREEN,y
     // SCREEN[i] = points[i].x;
     lda points+OFFS_Y,x
diff --git a/src/test/ref/struct-5.asm b/src/test/ref/struct-5.asm
index 3bbbba092..ed80f73d8 100644
--- a/src/test/ref/struct-5.asm
+++ b/src/test/ref/struct-5.asm
@@ -2,39 +2,31 @@
 .pc = $801 "Basic"
 :BasicUpstart(main)
 .pc = $80d "Program"
-  .const OFFS_X = 0
   .const OFFS_Y = 1
 main: {
     .label SCREEN = $400
-    .label _1 = 4
-    .label _3 = 2
-    .label _8 = 4
-    .label _11 = 2
     .label point_i = 2
     .label point_i1 = 2
     ldx #0
   b1:
     txa
     asl
+    tay
+    tya
     clc
     adc #<points
     sta point_i
     lda #>points
     adc #0
     sta point_i+1
-    lda point_i
-    sta _1
-    lda point_i+1
-    sta _1+1
     txa
-    ldy #OFFS_X
-    sta (_1),y
+    sta points,y
     txa
     clc
     adc #4
     // points[i].x = i;
     ldy #OFFS_Y
-    sta (_3),y
+    sta (point_i),y
     inx
     cpx #4
     bne b1
@@ -42,22 +34,19 @@ main: {
   b2:
     txa
     asl
+    tay
+    tya
     clc
     adc #<points
     sta point_i1
     lda #>points
     adc #0
     sta point_i1+1
-    lda point_i1
-    sta _8
-    lda point_i1+1
-    sta _8+1
-    ldy #OFFS_X
-    lda (_8),y
+    lda points,y
     sta SCREEN,x
     // SCREEN[i] = points[i].x;
     ldy #OFFS_Y
-    lda (_11),y
+    lda (point_i1),y
     sta SCREEN+$28,x
     inx
     cpx #4
diff --git a/src/test/ref/test-comparisons-sword.asm b/src/test/ref/test-comparisons-sword.asm
index 243347b6b..e6f0e4056 100644
--- a/src/test/ref/test-comparisons-sword.asm
+++ b/src/test/ref/test-comparisons-sword.asm
@@ -324,13 +324,11 @@ print_sword: {
     rts
 }
 // Print a word as HEX
-// print_word(word zeropage($a) w)
 print_word: {
-    .label w = $a
-    lda w+1
+    lda print_sword.w+1
     sta print_byte.b
     jsr print_byte
-    lda w
+    lda print_sword.w
     sta print_byte.b
     jsr print_byte
     rts
diff --git a/src/test/ref/test-division.asm b/src/test/ref/test-division.asm
index cd75806a1..8536a6ee8 100644
--- a/src/test/ref/test-division.asm
+++ b/src/test/ref/test-division.asm
@@ -216,14 +216,12 @@ div16s: {
 divr16s: {
     .label _8 = 8
     .label _13 = $a
-    .label _16 = $e
-    .label _18 = $c
-    .label dividendu = 8
-    .label divisoru = $a
     .label resultu = $c
     .label return = $c
     .label dividend = 8
     .label divisor = $a
+    .label dividendu = 8
+    .label divisoru = $a
     lda dividend+1
     bmi b1
     ldy #0
diff --git a/src/test/ref/test-multiply-16bit.asm b/src/test/ref/test-multiply-16bit.asm
index 4b1b9e166..3fe991b14 100644
--- a/src/test/ref/test-multiply-16bit.asm
+++ b/src/test/ref/test-multiply-16bit.asm
@@ -340,9 +340,7 @@ print_sword: {
 // mulf16s(signed word zeropage(3) a, signed word zeropage(5) b)
 mulf16s: {
     .label _9 = 9
-    .label _10 = $15
     .label _13 = 9
-    .label _14 = $15
     .label _16 = 9
     .label _17 = 9
     .label m = $11
@@ -364,16 +362,12 @@ mulf16s: {
     sta _9
     lda m+3
     sta _9+1
-    lda b
-    sta _10
-    lda b+1
-    sta _10+1
     lda _16
     sec
-    sbc _10
+    sbc b
     sta _16
     lda _16+1
-    sbc _10+1
+    sbc b+1
     sta _16+1
     lda _16
     sta m+2
@@ -386,16 +380,12 @@ mulf16s: {
     sta _13
     lda m+3
     sta _13+1
-    lda a
-    sta _14
-    lda a+1
-    sta _14+1
     lda _17
     sec
-    sbc _14
+    sbc a
     sta _17
     lda _17+1
-    sbc _14+1
+    sbc a+1
     sta _17+1
     lda _17
     sta m+2
@@ -529,30 +519,24 @@ mulf16u: {
 // mul16s(signed word zeropage(3) a, signed word zeropage(5) b)
 mul16s: {
     .label _9 = 9
-    .label _10 = $15
     .label _13 = 9
-    .label _14 = $15
     .label _16 = 9
     .label _17 = 9
     .label m = $19
     .label return = $19
     .label a = 3
     .label b = 5
-    lda a
-    sta mul16u.a
-    lda a+1
-    sta mul16u.a+1
     lda b
-    sta mul16u.b
-    lda b+1
-    sta mul16u.b+1
-    lda mul16u.b
     sta mul16u.mb
-    lda mul16u.b+1
+    lda b+1
     sta mul16u.mb+1
     lda #0
     sta mul16u.mb+2
     sta mul16u.mb+3
+    lda a
+    sta mul16u.a
+    lda a+1
+    sta mul16u.a+1
     jsr mul16u
     lda a+1
     bpl b1
@@ -560,16 +544,12 @@ mul16s: {
     sta _9
     lda m+3
     sta _9+1
-    lda b
-    sta _10
-    lda b+1
-    sta _10+1
     lda _16
     sec
-    sbc _10
+    sbc b
     sta _16
     lda _16+1
-    sbc _10+1
+    sbc b+1
     sta _16+1
     lda _16
     sta m+2
@@ -582,16 +562,12 @@ mul16s: {
     sta _13
     lda m+3
     sta _13+1
-    lda a
-    sta _14
-    lda a+1
-    sta _14+1
     lda _17
     sec
-    sbc _14
+    sbc a
     sta _17
     lda _17+1
-    sbc _14+1
+    sbc a+1
     sta _17+1
     lda _17
     sta m+2
@@ -601,14 +577,13 @@ mul16s: {
     rts
 }
 // Perform binary multiplication of two unsigned 16-bit words into a 32-bit unsigned double word
-// mul16u(word zeropage($1d) a, word zeropage(9) b)
+// mul16u(word zeropage(9) a, word zeropage($17) b)
 mul16u: {
     .label mb = $11
-    .label a = $1d
+    .label a = 9
     .label res = $19
-    .label b = 9
     .label return = $19
-    .label b_1 = $17
+    .label b = $17
     lda #0
     sta res
     sta res+1
@@ -796,9 +771,9 @@ mul16u_compare: {
     sta mul16u.a
     lda a+1
     sta mul16u.a+1
-    lda mul16u.b_1
+    lda mul16u.b
     sta mul16u.mb
-    lda mul16u.b_1+1
+    lda mul16u.b+1
     sta mul16u.mb+1
     lda #0
     sta mul16u.mb+2
diff --git a/src/test/ref/test-multiply-8bit.asm b/src/test/ref/test-multiply-8bit.asm
index 195dda466..eacb3dd67 100644
--- a/src/test/ref/test-multiply-8bit.asm
+++ b/src/test/ref/test-multiply-8bit.asm
@@ -19,8 +19,6 @@ main: {
 // Perform all possible signed byte multiplications (slow and fast) and compare the results
 mul8s_compare: {
     .label ms = 8
-    .label mf = $e
-    .label mn = $c
     .label b = 3
     .label a = 2
     lda #-$80
@@ -37,10 +35,10 @@ mul8s_compare: {
     ldy b
     jsr mul8s
     lda ms
-    cmp mf
+    cmp mulf8s_prepared.m
     bne !+
     lda ms+1
-    cmp mf+1
+    cmp mulf8s_prepared.m+1
     beq b6
   !:
     ldx #0
@@ -49,10 +47,10 @@ mul8s_compare: {
     ldx #1
   b3:
     lda ms
-    cmp mn
+    cmp mul8s.m
     bne !+
     lda ms+1
-    cmp mn+1
+    cmp mul8s.m+1
     beq b4
   !:
     ldx #0
@@ -129,12 +127,10 @@ print_str: {
   !:
     jmp b1
 }
-// mul8s_error(signed byte register(X) a, signed byte zeropage(3) b, signed word zeropage(8) ms, signed word zeropage($c) mn, signed word zeropage($e) mf)
+// mul8s_error(signed byte register(X) a, signed byte zeropage(3) b, signed word zeropage(8) ms)
 mul8s_error: {
     .label b = 3
     .label ms = 8
-    .label mn = $c
-    .label mf = $e
     lda print_line_cursor
     sta print_char_cursor
     lda print_line_cursor+1
@@ -163,9 +159,9 @@ mul8s_error: {
     lda #>str3
     sta print_str.str+1
     jsr print_str
-    lda mn
+    lda mul8s.m
     sta print_sword.w
-    lda mn+1
+    lda mul8s.m+1
     sta print_sword.w+1
     jsr print_sword
     lda #<str4
@@ -173,9 +169,9 @@ mul8s_error: {
     lda #>str4
     sta print_str.str+1
     jsr print_str
-    lda mf
+    lda mulf8s_prepared.m
     sta print_sword.w
-    lda mf+1
+    lda mulf8s_prepared.m+1
     sta print_sword.w+1
     jsr print_sword
     jsr print_ln
@@ -267,36 +263,28 @@ print_sbyte: {
 // Fixes offsets introduced by using unsigned multiplication
 // mul8s(signed byte zeropage(2) a, signed byte register(Y) b)
 mul8s: {
-    .label _9 = $10
-    .label _13 = $10
     .label m = $c
-    .label return = $c
     .label a = 2
-    ldx a
     tya
     sta mul8u.mb
     lda #0
     sta mul8u.mb+1
+    ldx a
     jsr mul8u
     lda a
     cmp #0
     bpl b1
     lda m+1
-    sta _9
-    tya
-    eor #$ff
+    sty $ff
     sec
-    adc _9
+    sbc $ff
     sta m+1
   b1:
     cpy #0
     bpl b2
     lda m+1
-    sta _13
-    lda a
-    eor #$ff
     sec
-    adc _13
+    sbc a
     sta m+1
   b2:
     rts
@@ -337,44 +325,34 @@ mul8u: {
 // Fast multiply two signed bytes to a word result
 // mulf8s(signed byte register(A) a, signed byte register(X) b)
 mulf8s: {
-    .label return = $e
     jsr mulf8u_prepare
-    txa
-    tay
+    stx mulf8s_prepared.b
     jsr mulf8s_prepared
     rts
 }
 // Calculate fast multiply with a prepared unsigned byte to a word result
 // The prepared number is set by calling mulf8s_prepare(byte a)
-// mulf8s_prepared(signed byte register(Y) b)
+// mulf8s_prepared(signed byte zeropage(3) b)
 mulf8s_prepared: {
     .label memA = $fd
-    .label _8 = $10
-    .label _12 = $10
     .label m = $e
-    .label return = $e
-    tya
-    tax
+    .label b = 3
+    ldx b
     jsr mulf8u_prepared
     lda memA
     cmp #0
     bpl b1
     lda m+1
-    sta _8
-    tya
-    eor #$ff
     sec
-    adc _8
+    sbc b
     sta m+1
   b1:
-    cpy #0
+    lda b
+    cmp #0
     bpl b2
     lda m+1
-    sta _12
-    lda memA
-    eor #$ff
     sec
-    adc _12
+    sbc memA
     sta m+1
   b2:
     rts
diff --git a/src/test/ref/test-signed-word-minus-byte.asm b/src/test/ref/test-signed-word-minus-byte.asm
index 809f17757..24426bf68 100644
--- a/src/test/ref/test-signed-word-minus-byte.asm
+++ b/src/test/ref/test-signed-word-minus-byte.asm
@@ -102,13 +102,11 @@ print_sword: {
     rts
 }
 // Print a word as HEX
-// print_word(word zeropage(6) w)
 print_word: {
-    .label w = 6
-    lda w+1
+    lda print_sword.w+1
     sta print_byte.b
     jsr print_byte
-    lda w
+    lda print_sword.w
     sta print_byte.b
     jsr print_byte
     rts
diff --git a/src/test/ref/useuninitialized.asm b/src/test/ref/useuninitialized.asm
index 24be884ea..1a1ab8cff 100644
--- a/src/test/ref/useuninitialized.asm
+++ b/src/test/ref/useuninitialized.asm
@@ -2,10 +2,11 @@
 .pc = $801 "Basic"
 :BasicUpstart(main)
 .pc = $80d "Program"
+  .const b = 3
   .const s = 1
 main: {
     .label screen = $400
-    lda #3
+    lda #b
     sta screen
     lda #s
     sta screen+1