diff --git a/CONFIG.md b/CONFIG.md
index 5d09de23..8310b8d8 100644
--- a/CONFIG.md
+++ b/CONFIG.md
@@ -51,7 +51,7 @@ Choose if your CPU supports the Western Design Center 65C02 instruction set, lik
 * CSG 65CE02 
 * CSG 4510 - used in the Commodore 65 prototypes
 
-It enables some speed/size code optimizations. Warning - this is untested!
+It enables some speed/size code optimizations.
 
 ### `CONFIG_CPU_WDC_65816`
 
@@ -60,7 +60,7 @@ Choose if your CPU supports the 16-bit Western Design Center 65816 instruction s
 * WDC 65C816 - used in the Flash 8 accelerator
 * WDC 65C816S - used in the SuperCPU accelerator
 
-It enables some speed/size code optimizations. Warning - this is untested!
+It enables some speed/size code optimizations.
 
 ## Memory model
 
diff --git a/c64/aliases/,aliases_cpu.s b/c64/aliases/,aliases_cpu.s
new file mode 100644
index 00000000..45f4727b
--- /dev/null
+++ b/c64/aliases/,aliases_cpu.s
@@ -0,0 +1,45 @@
+
+//
+// Provide pseudocommands to make CPU optimizations easier
+//
+
+
+.pseudocommand _phx
+{
+#if CONFIG_CPU_WDC_65C02 || CONFIG_CPU_WDC_65816
+	phx
+#else
+	txa
+	pha
+#endif	
+}
+
+.pseudocommand _phy
+{
+#if CONFIG_CPU_WDC_65C02 || CONFIG_CPU_WDC_65816
+	phy
+#else
+	tya
+	pha
+#endif	
+}
+
+.pseudocommand _plx
+{
+#if CONFIG_CPU_WDC_65C02 || CONFIG_CPU_WDC_65816
+	plx
+#else
+	pla
+	tax
+#endif	
+}
+
+.pseudocommand _ply
+{
+#if CONFIG_CPU_WDC_65C02 || CONFIG_CPU_WDC_65816
+	ply
+#else
+	pla
+	tay
+#endif	
+}
diff --git a/c64/basic/ab1e.strout.s b/c64/basic/ab1e.strout.s
index 165af8b1..0526a6b1 100644
--- a/c64/basic/ab1e.strout.s
+++ b/c64/basic/ab1e.strout.s
@@ -7,9 +7,8 @@ STROUT:
 	// Setup pointer
 	sty FRESPC+1
 	sta FRESPC+0
-
-	txa	
-	pha
+	
+	_phx
 
 	// Get offset ready
 	ldy #$00
@@ -29,6 +28,6 @@ STROUT:
 	iny
 	bne !-
 !:
-	pla
+	pla // XXX can we use _plx here?
 	tax
 	rts
diff --git a/c64/basic/basic_do_error.s b/c64/basic/basic_do_error.s
index 6266b0ef..984e71eb 100644
--- a/c64/basic/basic_do_error.s
+++ b/c64/basic/basic_do_error.s
@@ -2,8 +2,7 @@
 
 basic_do_error:
 	// Save error number
-	txa
-	pha
+	_phx
 
 	// Print ? at start
 	jsr print_return
@@ -11,8 +10,7 @@ basic_do_error:
 	jsr JCHROUT
 
 	// Print main part of error message
-	pla
-	tax
+	_plx
 	jsr print_packed_message
 
 	// Print ERROR at end
diff --git a/c64/basic/cmd_list.s b/c64/basic/cmd_list.s
index 08afa852..190c6ca5 100644
--- a/c64/basic/cmd_list.s
+++ b/c64/basic/cmd_list.s
@@ -98,15 +98,9 @@ list_not_quote:
 	// Display a token
 
 	// Save registers
-#if CONFIG_CPU_MOS_6502
 	tax
 	pha
-	tya
-	pha
-#else
-	phx
-	phy
-#endif
+	_phy
 
 	// Get pointer to compressed keyword list
 	lda #<packed_keywords
@@ -124,13 +118,7 @@ list_not_quote:
 	ldy #$ff
 	jsr packed_word_search
 
-#if CONFIG_CPU_MOS_6502
-	pla
-	tay
-#else
-	ply
-#endif
-
+	_ply
 	pla
 
 	cmp #$8f
diff --git a/c64/basic/print_integer.s b/c64/basic/print_integer.s
index ff9938fd..e99383a0 100644
--- a/c64/basic/print_integer.s
+++ b/c64/basic/print_integer.s
@@ -5,8 +5,7 @@
 print_integer:
 
 	pha
-	txa
-	pha
+	_phx
 
 	// Clear temporary output
 	lda #$00
diff --git a/c64/basic/print_packed_message.s b/c64/basic/print_packed_message.s
index 7012645b..33f9e152 100644
--- a/c64/basic/print_packed_message.s
+++ b/c64/basic/print_packed_message.s
@@ -32,15 +32,13 @@ packed_message_search:
 
 found_message_in_token_stream:
 	// Now print each word in the message
-	tya
-	pha
+	_phy
 	lda packed_message_tokens,y
 	cmp #$ff
 	beq !+
 	jsr print_packed_word
 	jsr print_space
-	pla
-	tay
+	_ply
 	iny
 	bne found_message_in_token_stream
 !:
@@ -139,16 +137,14 @@ next_packed_word_char:
 	tax
 
 	// Save Y so it doesn't get clobbered by CHROUT
-	tya
-	pha
+	_phy
 
 	// X=1-14 = first 14 chars, so subtract one
 	lda packed_message_chars-1,x
 	jsr JCHROUT
 
 	// See if there is a char in the low nybl to print
-	pla
-	tay
+	_ply
 	lda (FRESPC),y
 	and #$0f
 	beq end_of_packed_word
@@ -158,8 +154,7 @@ next_packed_word_char:
 has_lo_nybl_char:
 	// We need to print the low nybl char as well
 	tax
-	tya
-	pha
+	_phy
 
 	// Skip lower nybls of $F or $0 in nybl packed
 	// bytes, as these do not encode characters
@@ -173,8 +168,7 @@ has_nybl:
 	lda packed_message_chars-1,x
 	jsr JCHROUT
 !:
-	pla
-	tay
+	_ply
 	// FALLTHROUGH
 
 no_lo_nybl_char:
@@ -204,8 +198,7 @@ next_is_literal_char:
 unpack_literal_char:
 	// Output the literal char, and ready for next char
 	iny
-	tya
-	pha
+	_phy
 	lda (FRESPC),y
 	jsr JCHROUT
 	pla
@@ -222,7 +215,6 @@ is_uncommon_char:
 
 	// We need to print the low nybl char as well
 	tax
-	tya
-	pha
+	_phy
 
 	jmp has_nybl
diff --git a/c64/basic/wedge_dos.s b/c64/basic/wedge_dos.s
index 7ede8e76..2f706df5 100644
--- a/c64/basic/wedge_dos.s
+++ b/c64/basic/wedge_dos.s
@@ -232,8 +232,7 @@ wedge_dos_clean_exit:
 wedge_dos_basic_error:
 	pha
 	jsr JCLALL
-	pla
-	tax
+	_plx
 	dex
 	jmp do_basic_error
 
diff --git a/c64/kernal/chkinout_real.s b/c64/kernal/chkinout_real.s
index c52545e3..d20dae1d 100644
--- a/c64/kernal/chkinout_real.s
+++ b/c64/kernal/chkinout_real.s
@@ -14,8 +14,7 @@
 chkin_real:
 
 	// Store Y for preservation
-	tya
-	pha
+	_phy
 
 	// Reset status
 	jsr kernalstatus_reset
@@ -59,6 +58,7 @@ chkin_set_device:
 	// FALLTROUGH
 
 // Common part for booth CHKIN and CKOUT
+// XXX add CPU-dependent optimizations here
 
 chkinout_end:
 	tax
diff --git a/c64/kernal/chrin_real.s b/c64/kernal/chrin_real.s
index 5e0a7e17..bf4f8162 100644
--- a/c64/kernal/chrin_real.s
+++ b/c64/kernal/chrin_real.s
@@ -44,8 +44,7 @@ chrin_done_fail:
 chrin_keyboard:
 
 	// Save X
-	txa
-	pha
+	_phx
 
 chrin_repeat:
 
@@ -65,8 +64,7 @@ chrin_repeat:
 	// Return carriage return and clear pending input flag
 	lda #$00
 	sta CRSW
-	pla
-	tax
+	_plx
 	clc
 	lda #$0d
 	rts
@@ -74,8 +72,7 @@ chrin_repeat:
 not_end_of_input:
 	// Return next byte of waiting input and advance index
 	tay
-	pla
-	tax
+	_plx
 	lda (LXSP),y
 	jsr screen_code_to_petscii
 	inc CRSW
@@ -126,8 +123,7 @@ read_from_keyboard:
 	sta CRSW
 	// Return first char of line
 	ldy #$00
-	pla
-	tax
+	_plx
 	lda (LXSP),y
 	jsr screen_code_to_petscii
 	clc
@@ -137,8 +133,7 @@ empty_line:
 	// For an empty line, just return the carriage return
 	// (and don't forget to actually print the carriage return, so that
 	// the cursor advances and screen scrolls as required)
-	pla
-	tax
+	_plx
 	clc
 	lda #$0d
 	rts
diff --git a/c64/kernal/clall_real.s b/c64/kernal/clall_real.s
index 2a24af72..866aba28 100644
--- a/c64/kernal/clall_real.s
+++ b/c64/kernal/clall_real.s
@@ -12,8 +12,7 @@
 clall_real:
 
 	// Store .Y register
-	tya
-	pha
+	_phy
 
 	// Original routine probably just sets LDTND to 0, but this is not really safe,
 	// so we actually close all the channels; at least IDE64 does the same for
@@ -27,8 +26,7 @@ clall_real:
 	jmp !-
 !:
 	// Restore .Y register
-	pla
-	tay
+	_ply
 
 	// 'C64 Programmers Reference Guide', page 281, claims it calls CLRCHN too
 	jsr CLRCHN
diff --git a/c64/kernal/f1ca.chrout.s b/c64/kernal/f1ca.chrout.s
index 46e187b4..4d2c02aa 100644
--- a/c64/kernal/f1ca.chrout.s
+++ b/c64/kernal/f1ca.chrout.s
@@ -17,15 +17,8 @@ CHROUT:
 	// don't get modified, in agreement with C64 PRG's
 	// description of CHROUT)
 
-#if CONFIG_CPU_MOS_6502
-	txa
-	pha
-	tya
-	pha
-#else
-	phx
-	phy
-#endif
+	_phx
+	_phy
 
 	php
 
@@ -52,16 +45,8 @@ chrout_done_fail:
 	plp
 
 	// Restore X and Y
-
-#if CONFIG_CPU_MOS_6502
-	pla
-	tay
-	pla
-	tax
-#else
-	ply
-	plx
-#endif
+	_ply
+	_plx
 
 	sec // indicate failure
 	rts
@@ -71,16 +56,8 @@ chrout_done_unknown_device:
 	plp
 
 	// Restore X and Y
-
-#if CONFIG_CPU_MOS_6502
-	pla
-	tay
-	pla
-	tax
-#else
-	ply
-	plx
-#endif
+	_ply
+	_plx
 
 	// End wioth error
 	jmp lvs_device_not_found_error
@@ -90,16 +67,8 @@ chrout_done_success:
 	plp
 
 	// Restore X and Y
-
-#if CONFIG_CPU_MOS_6502
-	pla
-	tay
-	pla
-	tax
-#else
-	ply
-	plx
-#endif
+	_ply
+	_plx
 
 	lda SCHAR
 	clc // indicate success
diff --git a/c64/kernal/iec/iec_return.s b/c64/kernal/iec/iec_return.s
index 149f781f..b343ba49 100644
--- a/c64/kernal/iec/iec_return.s
+++ b/c64/kernal/iec/iec_return.s
@@ -14,10 +14,8 @@ iec_return_success:
 	lda #$00
 	sta C3PO
 	// Restore registers
-	pla
-	tay
-	pla
-	tax
+	_ply
+	_plx
 	// Report success
 	clc
 	rts
@@ -30,9 +28,7 @@ iec_return_DEVICE_NOT_FOUND:
 	lda #$00
 	sta C3PO
 	// Restore registers
-	pla
-	tay
-	pla
-	tax
+	_ply
+	_plx
 	// Report error
 	jmp kernalerror_DEVICE_NOT_FOUND
diff --git a/c64/kernal/iec/iec_rx_byte.s b/c64/kernal/iec/iec_rx_byte.s
index 01f34846..ebcd8f12 100644
--- a/c64/kernal/iec/iec_rx_byte.s
+++ b/c64/kernal/iec/iec_rx_byte.s
@@ -7,10 +7,8 @@
 iec_rx_byte:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Timing is critical here - execute on disabled IRQs
 	// The best practice would be to do PHP first, but it seems this is not
diff --git a/c64/kernal/iec/iec_turnaround_to_listen.s b/c64/kernal/iec/iec_turnaround_to_listen.s
index fb4cd078..ebb9b867 100644
--- a/c64/kernal/iec/iec_turnaround_to_listen.s
+++ b/c64/kernal/iec/iec_turnaround_to_listen.s
@@ -5,10 +5,8 @@
 iec_turnaround_to_listen:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Timing is critical here - execute on disabled IRQs
 	php
diff --git a/c64/kernal/iec/iec_turnaround_to_talk.s b/c64/kernal/iec/iec_turnaround_to_talk.s
index f7ea9c9b..73e0905a 100644
--- a/c64/kernal/iec/iec_turnaround_to_talk.s
+++ b/c64/kernal/iec/iec_turnaround_to_talk.s
@@ -5,10 +5,8 @@
 iec_turnaround_to_talk:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Timing is critical here - execute on disabled IRQs
 	php
diff --git a/c64/kernal/iec/iec_tx_byte.s b/c64/kernal/iec/iec_tx_byte.s
index 7721cc0e..1ce56d4f 100644
--- a/c64/kernal/iec/iec_tx_byte.s
+++ b/c64/kernal/iec/iec_tx_byte.s
@@ -10,10 +10,8 @@
 iec_tx_byte:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Notify all devices that we are going to send a byte
 	// and it is going to be a data byte (released ATN)
diff --git a/c64/kernal/iec/iec_tx_command.s b/c64/kernal/iec/iec_tx_command.s
index 42c98960..0de1f8ee 100644
--- a/c64/kernal/iec/iec_tx_command.s
+++ b/c64/kernal/iec/iec_tx_command.s
@@ -10,10 +10,8 @@
 iec_tx_command:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Notify all devices that we are going to send a byte
 	// and it is going to be a command (pulled ATN)
diff --git a/c64/kernal/iec/iec_tx_command_finalize.s b/c64/kernal/iec/iec_tx_command_finalize.s
index 6f11ea03..3dc5b76f 100644
--- a/c64/kernal/iec/iec_tx_command_finalize.s
+++ b/c64/kernal/iec/iec_tx_command_finalize.s
@@ -6,10 +6,8 @@
 iec_tx_command_finalize:
 
 	// Store .X and .Y on the stack - preserve them
-	txa
-	pha
-	tya
-	pha
+	_phx
+	_phy
 
 	// Give device some time to complete command reception, release ATN,
 	// give it some time again before (possible) next command
diff --git a/c64/kernal/interrupts/ea81.return_from_interrupt.s b/c64/kernal/interrupts/ea81.return_from_interrupt.s
index 4908b8ed..47aab053 100644
--- a/c64/kernal/interrupts/ea81.return_from_interrupt.s
+++ b/c64/kernal/interrupts/ea81.return_from_interrupt.s
@@ -12,15 +12,7 @@ return_from_interrupt:
 	// Restore registers and return
 	// Sequence according to Compute's Mapping the 64 p73
 
-#if CONFIG_CPU_MOS_6502
-	pla
-	tay
-	pla
-	tax
-#else
-	ply
-	plx
-#endif
+	_ply
+	_plx
 	pla
 	rti
-
diff --git a/c64/kernal/interrupts/fe47.default_nmi_handler.s b/c64/kernal/interrupts/fe47.default_nmi_handler.s
index 97cdb1d6..fbdddfb6 100644
--- a/c64/kernal/interrupts/fe47.default_nmi_handler.s
+++ b/c64/kernal/interrupts/fe47.default_nmi_handler.s
@@ -6,15 +6,8 @@ default_nmi_handler:
 
 	// Save registers, sequence according to Compute's Mapping the Commodore 64, page 73
 	pha
-#if CONFIG_CPU_MOS_6502
-	txa
-	pha
-	tya
-	pha
-#else
-	phx
-	phy
-#endif
+	_phx
+	_phy
 
 	// XXX: RS-232 support is not implemented
 
diff --git a/c64/kernal/interrupts/hw_entry_irq.s b/c64/kernal/interrupts/hw_entry_irq.s
index 517100c4..9b16313b 100644
--- a/c64/kernal/interrupts/hw_entry_irq.s
+++ b/c64/kernal/interrupts/hw_entry_irq.s
@@ -11,15 +11,8 @@ hw_entry_irq:
 
 	// Save registers, sequence according to Compute's Mapping the Commodore 64, page 73
 	pha
-#if CONFIG_CPU_MOS_6502
-	txa
-	pha
-	tya
-	pha
-#else
-	phx
-	phy
-#endif
+	_phx
+	_phy
 
 #if CONFIG_CPU_MOS_6502 && CONFIG_BCD_SAFE_INTERRUPTS
 	cld // clear decimal flag to allow using it without disabling interrupts