diff --git a/internal/commands/shiftl.go b/internal/commands/shiftl.go
index 85c6a21..5ed6ca5 100644
--- a/internal/commands/shiftl.go
+++ b/internal/commands/shiftl.go
@@ -251,15 +251,36 @@ func (c *ShiftLCommand) generateByteCopy() []string {
 func (c *ShiftLCommand) generateByteShiftConst(amount uint16) []string {
 	var asm []string
 	
-	// Copy source to destination if needed
-	copyAsm := c.generateByteCopy()
-	asm = append(asm, copyAsm...)
-	
-	// Apply shift
-	for i := uint16(0); i < amount; i++ {
-		asm = append(asm, fmt.Sprintf("\tasl %s", c.destVarName))
+	// Check if same variable
+	if c.sourceIsVar && c.sourceVarName == c.destVarName {
+		// Same variable: load to accumulator, shift, store back (faster)
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+		// Shift in accumulator (faster: 2 cycles vs 5 for memory)
+		for i := uint16(0); i < amount; i++ {
+			asm = append(asm, "\tasl")
+		}
+		asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+		return asm
 	}
 	
+	// Different variables or literal source: shift in accumulator
+	if c.sourceIsVar {
+		// Variable source
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+	} else {
+		// Literal source
+		val := uint8(c.sourceValue & 0xFF)
+		asm = append(asm, fmt.Sprintf("\tlda #$%02x", val))
+	}
+	
+	// Shift in accumulator (faster: 2 cycles vs 5 for memory)
+	for i := uint16(0); i < amount; i++ {
+		asm = append(asm, "\tasl")
+	}
+	
+	// Store result
+	asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+	
 	return asm
 }
 
@@ -267,24 +288,61 @@ func (c *ShiftLCommand) generateByteShiftConst(amount uint16) []string {
 func (c *ShiftLCommand) generateByteShiftVar(ctx *compiler.CompilerContext) ([]string, error) {
 	var asm []string
 	
-	// Copy source to destination if needed
-	copyAsm := c.generateByteCopy()
-	asm = append(asm, copyAsm...)
+	// Check if same variable
+	if c.sourceIsVar && c.sourceVarName == c.destVarName {
+		// Same variable: load to accumulator, shift loop, store back (faster)
+		// Generate labels
+		loopLabel := ctx.GeneralStack.Push()
+		ctx.GeneralStack.Pop()
+		doneLabel := ctx.GeneralStack.Push()
+		ctx.GeneralStack.Pop()
+		
+		// Load value to accumulator
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+		
+		// Variable shift loop (in accumulator, 2 cycles vs 5 for memory)
+		asm = append(asm, fmt.Sprintf("\tldx %s", c.amountVarName))
+		asm = append(asm, fmt.Sprintf("\tbeq %s", doneLabel))
+		asm = append(asm, loopLabel)
+		asm = append(asm, "\tasl")
+		asm = append(asm, "\tdex")
+		asm = append(asm, fmt.Sprintf("\tbne %s", loopLabel))
+		asm = append(asm, doneLabel)
+		
+		// Store result back
+		asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+		
+		return asm, nil
+	}
 	
+	// Different variables or literal: shift in accumulator (faster)
 	// Generate labels
 	loopLabel := ctx.GeneralStack.Push()
 	ctx.GeneralStack.Pop()
 	doneLabel := ctx.GeneralStack.Push()
 	ctx.GeneralStack.Pop()
 	
-	// Variable shift loop
+	// Load source to accumulator
+	if c.sourceIsVar {
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+	} else {
+		val := uint8(c.sourceValue & 0xFF)
+		asm = append(asm, fmt.Sprintf("\tlda #$%02x", val))
+	}
+	
+	// Load shift amount to X
 	asm = append(asm, fmt.Sprintf("\tldx %s", c.amountVarName))
 	asm = append(asm, fmt.Sprintf("\tbeq %s", doneLabel))
+	
+	// Shift loop (in accumulator, 2 cycles vs 5 for memory)
 	asm = append(asm, loopLabel)
-	asm = append(asm, fmt.Sprintf("\tasl %s", c.destVarName))
+	asm = append(asm, "\tasl")
 	asm = append(asm, "\tdex")
 	asm = append(asm, fmt.Sprintf("\tbne %s", loopLabel))
+	
+	// Store result
 	asm = append(asm, doneLabel)
+	asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
 	
 	return asm, nil
 }
diff --git a/internal/commands/shiftl_test.go b/internal/commands/shiftl_test.go
index 8248069..7704036 100644
--- a/internal/commands/shiftl_test.go
+++ b/internal/commands/shiftl_test.go
@@ -366,10 +366,10 @@ func TestShiftLCommand_Generate(t *testing.T) {
 			},
 			wantLines: []string{
 				"\tlda #$01",
+				"\tasl",
+				"\tasl",
+				"\tasl",
 				"\tsta result",
-				"\tasl result",
-				"\tasl result",
-				"\tasl result",
 			},
 		},
 		{
@@ -450,14 +450,14 @@ func TestShiftLCommand_Generate(t *testing.T) {
 			},
 			wantLines: []string{
 				"\tlda value",
-				"\tsta result",
 				"\tldx shift",
 				"\tbeq _L2",
 				"_L1",
-				"\tasl result",
+				"\tasl",
 				"\tdex",
 				"\tbne _L1",
 				"_L2",
+				"\tsta result",
 			},
 		},
 		{
@@ -509,13 +509,15 @@ func TestShiftLCommand_Generate(t *testing.T) {
 				}
 			},
 			wantLines: []string{
+				"\tlda value",
 				"\tldx shift",
 				"\tbeq _L2",
 				"_L1",
-				"\tasl value",
+				"\tasl",
 				"\tdex",
 				"\tbne _L1",
 				"_L2",
+				"\tsta value",
 			},
 		},
 		{
@@ -609,11 +611,11 @@ func TestShiftLCommand_Generate(t *testing.T) {
 			},
 			wantLines: []string{
 				"\tlda wordval",
+				"\tasl",
+				"\tasl",
+				"\tasl",
+				"\tasl",
 				"\tsta result",
-				"\tasl result",
-				"\tasl result",
-				"\tasl result",
-				"\tasl result",
 			},
 		},
 		{
diff --git a/internal/commands/shiftr.go b/internal/commands/shiftr.go
index f366fc9..f62770c 100644
--- a/internal/commands/shiftr.go
+++ b/internal/commands/shiftr.go
@@ -263,15 +263,36 @@ func (c *ShiftRCommand) generateByteCopy() []string {
 func (c *ShiftRCommand) generateByteShiftConst(amount uint16) []string {
 	var asm []string
 	
-	// Copy source to destination if needed
-	copyAsm := c.generateByteCopy()
-	asm = append(asm, copyAsm...)
-	
-	// Apply shift (right shift uses LSR)
-	for i := uint16(0); i < amount; i++ {
-		asm = append(asm, fmt.Sprintf("\tlsr %s", c.destVarName))
+	// Check if same variable
+	if c.sourceIsVar && c.sourceVarName == c.destVarName {
+		// Same variable: load to accumulator, shift, store back (faster)
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+		// Shift in accumulator (faster: 2 cycles vs 5 for memory)
+		for i := uint16(0); i < amount; i++ {
+			asm = append(asm, "\tlsr")
+		}
+		asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+		return asm
 	}
 	
+	// Different variables or literal source: shift in accumulator
+	if c.sourceIsVar {
+		// Variable source
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+	} else {
+		// Literal source
+		val := uint8(c.sourceValue & 0xFF)
+		asm = append(asm, fmt.Sprintf("\tlda #$%02x", val))
+	}
+	
+	// Shift in accumulator (faster: 2 cycles vs 5 for memory)
+	for i := uint16(0); i < amount; i++ {
+		asm = append(asm, "\tlsr")
+	}
+	
+	// Store result
+	asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+	
 	return asm
 }
 
@@ -279,24 +300,61 @@ func (c *ShiftRCommand) generateByteShiftConst(amount uint16) []string {
 func (c *ShiftRCommand) generateByteShiftVar(ctx *compiler.CompilerContext) ([]string, error) {
 	var asm []string
 	
-	// Copy source to destination if needed
-	copyAsm := c.generateByteCopy()
-	asm = append(asm, copyAsm...)
+	// Check if same variable
+	if c.sourceIsVar && c.sourceVarName == c.destVarName {
+		// Same variable: load to accumulator, shift loop, store back (faster)
+		// Generate labels
+		loopLabel := ctx.GeneralStack.Push()
+		ctx.GeneralStack.Pop()
+		doneLabel := ctx.GeneralStack.Push()
+		ctx.GeneralStack.Pop()
+		
+		// Load value to accumulator
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+		
+		// Variable shift loop (in accumulator, 2 cycles vs 5 for memory)
+		asm = append(asm, fmt.Sprintf("\tldx %s", c.amountVarName))
+		asm = append(asm, fmt.Sprintf("\tbeq %s", doneLabel))
+		asm = append(asm, loopLabel)
+		asm = append(asm, "\tlsr")
+		asm = append(asm, "\tdex")
+		asm = append(asm, fmt.Sprintf("\tbne %s", loopLabel))
+		asm = append(asm, doneLabel)
+		
+		// Store result back
+		asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
+		
+		return asm, nil
+	}
 	
+	// Different variables or literal: shift in accumulator (faster)
 	// Generate labels
 	loopLabel := ctx.GeneralStack.Push()
 	ctx.GeneralStack.Pop()
 	doneLabel := ctx.GeneralStack.Push()
 	ctx.GeneralStack.Pop()
 	
-	// Variable shift loop (right shift uses LSR)
+	// Load source to accumulator
+	if c.sourceIsVar {
+		asm = append(asm, fmt.Sprintf("\tlda %s", c.sourceVarName))
+	} else {
+		val := uint8(c.sourceValue & 0xFF)
+		asm = append(asm, fmt.Sprintf("\tlda #$%02x", val))
+	}
+	
+	// Load shift amount to X
 	asm = append(asm, fmt.Sprintf("\tldx %s", c.amountVarName))
 	asm = append(asm, fmt.Sprintf("\tbeq %s", doneLabel))
+	
+	// Shift loop (in accumulator, 2 cycles vs 5 for memory)
 	asm = append(asm, loopLabel)
-	asm = append(asm, fmt.Sprintf("\tlsr %s", c.destVarName))
+	asm = append(asm, "\tlsr")
 	asm = append(asm, "\tdex")
 	asm = append(asm, fmt.Sprintf("\tbne %s", loopLabel))
+	
+	// Store result
 	asm = append(asm, doneLabel)
+	asm = append(asm, fmt.Sprintf("\tsta %s", c.destVarName))
 	
 	return asm, nil
 }
diff --git a/internal/commands/shiftr_test.go b/internal/commands/shiftr_test.go
index 15a59d0..9e3bb44 100644
--- a/internal/commands/shiftr_test.go
+++ b/internal/commands/shiftr_test.go
@@ -366,10 +366,10 @@ func TestShiftRCommand_Generate(t *testing.T) {
 			},
 			wantLines: []string{
 				"\tlda #$08",
+				"\tlsr",
+				"\tlsr",
+				"\tlsr",
 				"\tsta result",
-				"\tlsr result",
-				"\tlsr result",
-				"\tlsr result",
 			},
 		},
 		{
@@ -450,14 +450,14 @@ func TestShiftRCommand_Generate(t *testing.T) {
 			},
 			wantLines: []string{
 				"\tlda value",
-				"\tsta result",
 				"\tldx shift",
 				"\tbeq _L2",
 				"_L1",
-				"\tlsr result",
+				"\tlsr",
 				"\tdex",
 				"\tbne _L1",
 				"_L2",
+				"\tsta result",
 			},
 		},
 		{
@@ -509,13 +509,15 @@ func TestShiftRCommand_Generate(t *testing.T) {
 				}
 			},
 			wantLines: []string{
+				"\tlda value",
 				"\tldx shift",
 				"\tbeq _L2",
 				"_L1",
-				"\tlsr value",
+				"\tlsr",
 				"\tdex",
 				"\tbne _L1",
 				"_L2",
+				"\tsta value",
 			},
 		},
 		{