* mpn/powerpc32/*.asm: Use L(), add some measured speeds.

4 files changed, 62 insertions, 38 deletions

diff --git a/mpn/powerpc32/add_n.asm b/mpn/powerpc32/add_n.asm
index bb257f2f3..43b50101e 100644
--- a/mpn/powerpc32/add_n.asm
+++ b/mpn/powerpc32/add_n.asm
@@ -1,7 +1,11 @@
 dnl PowerPC-32 mpn_add_n -- Add two limb vectors of the same length > 0 and
 dnl store sum in a third limb vector.
+dnl
+dnl      cycles/limb
+dnl 604e:   2.67
+dnl 750:    4.5
 
-dnl Copyright 1995, 1997, 2000 Free Software Foundation, Inc.
+dnl Copyright 1995, 1997, 2000, 2002 Free Software Foundation, Inc.
 
 dnl This file is part of the GNU MP Library.
 
@@ -36,24 +40,26 @@ PROLOGUE(mpn_add_n)
 	lwz	r8,0(r4)	C load least significant s1 limb
 	lwz	r0,0(r5)	C load least significant s2 limb
 	addi	r3,r3,-4	C offset res_ptr, it's updated before it's used
-	bdz	.Lend		C If done, skip loop
-.Loop:	lwz	r9,4(r4)	C load s1 limb
+	bdz	L(end)		C If done, skip loop
+
+L(oop):	lwz	r9,4(r4)	C load s1 limb
 	lwz	r10,4(r5)	C load s2 limb
 	adde	r7,r0,r8	C add limbs with cy, set cy
 	stw	r7,4(r3)	C store result limb
-	bdz	.Lexit		C decrement CTR and exit if done
+	bdz	L(exit)		C decrement CTR and exit if done
 	lwzu	r8,8(r4)	C load s1 limb and update s1_ptr
 	lwzu	r0,8(r5)	C load s2 limb and update s2_ptr
 	adde	r7,r10,r9	C add limbs with cy, set cy
 	stwu	r7,8(r3)	C store result limb and update res_ptr
-	bdnz	.Loop		C decrement CTR and loop back
+	bdnz	L(oop)		C decrement CTR and loop back
 
-.Lend:	adde	r7,r0,r8
+L(end):	adde	r7,r0,r8
 	stw	r7,4(r3)	C store ultimate result limb
 	li	r3,0		C load cy into ...
 	addze	r3,r3		C ... return value register
 	blr
-.Lexit:	adde	r7,r10,r9
+L(exit):
+	adde	r7,r10,r9
 	stw	r7,8(r3)
 	li	r3,0		C load cy into ...
 	addze	r3,r3		C ... return value register
diff --git a/mpn/powerpc32/lshift.asm b/mpn/powerpc32/lshift.asm
index a89d1b815..7305f2b9c 100644
--- a/mpn/powerpc32/lshift.asm
+++ b/mpn/powerpc32/lshift.asm
@@ -1,6 +1,10 @@
 dnl  PowerPC-32 mpn_lshift -- Shift a number left.
+dnl
+dnl      cycles/limb
+dnl 604e:    2.0
+dnl 750:     3.0
 
-dnl Copyright 1995, 1998, 2000 Free Software Foundation, Inc.
+dnl Copyright 1995, 1998, 2000, 2002 Free Software Foundation, Inc.
 
 dnl This file is part of the GNU MP Library.
 
@@ -34,35 +38,37 @@ PROLOGUE(mpn_lshift)
 	slwi	r0,r5,2
 	add	r4,r4,r0	C make r4 point at end of s1
 	add	r7,r3,r0	C make r7 point at end of res
-	bgt	.LBIG		C branch if more than 12 limbs
+	bgt	L(BIG)		C branch if more than 12 limbs
 
 	mtctr	r5		C copy size into CTR
 	subfic	r8,r6,32
 	lwzu	r11,-4(r4)	C load first s1 limb
 	srw	r3,r11,r8	C compute function return value
-	bdz	.Lend1
+	bdz	L(end1)
 
-.Loop:	lwzu	r10,-4(r4)
+L(oop):	lwzu	r10,-4(r4)
 	slw	r9,r11,r6
 	srw	r12,r10,r8
 	or	r9,r9,r12
 	stwu	r9,-4(r7)
-	bdz	.Lend2
+	bdz	L(end2)
 	lwzu	r11,-4(r4)
 	slw	r9,r10,r6
 	srw	r12,r11,r8
 	or	r9,r9,r12
 	stwu	r9,-4(r7)
-	bdnz	.Loop
+	bdnz	L(oop)
 
-.Lend1:	slw	r0,r11,r6
+L(end1):
+	slw	r0,r11,r6
 	stw	r0,-4(r7)
 	blr
-.Lend2:	slw	r0,r10,r6
+L(end2):
+	slw	r0,r10,r6
 	stw	r0,-4(r7)
 	blr
 
-.LBIG:
+L(BIG):
 	stmw	r24,-32(r1)	C save registers we are supposed to preserve
 	lwzu	r9,-4(r4)
 	subfic	r8,r6,32
@@ -71,26 +77,26 @@ PROLOGUE(mpn_lshift)
 	addi	r5,r5,-1
 
 	andi.	r10,r5,3	C count for spill loop
-	beq	.Le
+	beq	L(e)
 	mtctr	r10
 	lwzu	r28,-4(r4)
-	bdz	.Lxe0
+	bdz	L(xe0)
 
-.Loop0:	slw	r12,r28,r6
+L(oop0):	slw	r12,r28,r6
 	srw	r24,r28,r8
 	lwzu	r28,-4(r4)
 	or	r24,r0,r24
 	stwu	r24,-4(r7)
 	mr	r0,r12
-	bdnz	.Loop0		C taken at most once!
+	bdnz	L(oop0)		C taken at most once!
 
-.Lxe0:	slw	r12,r28,r6
+L(xe0):	slw	r12,r28,r6
 	srw	r24,r28,r8
 	or	r24,r0,r24
 	stwu	r24,-4(r7)
 	mr	r0,r12
 
-.Le:	srwi	r5,r5,2		C count for unrolled loop
+L(e):	srwi	r5,r5,2		C count for unrolled loop
 	addi	r5,r5,-1
 	mtctr	r5
 	lwz	r28,-4(r4)
@@ -98,7 +104,7 @@ PROLOGUE(mpn_lshift)
 	lwz	r30,-12(r4)
 	lwzu	r31,-16(r4)
 
-.LoopU:	slw	r9,r28,r6
+L(oopU):	slw	r9,r28,r6
 	srw	r24,r28,r8
 	lwz	r28,-4(r4)
 	slw	r10,r29,r6
@@ -119,7 +125,7 @@ PROLOGUE(mpn_lshift)
 	or	r27,r11,r27
 	stwu	r27,-16(r7)
 	mr	r0,r12
-	bdnz	.LoopU
+	bdnz	L(oopU)
 
 	slw	r9,r28,r6
 	srw	r24,r28,r8
diff --git a/mpn/powerpc32/rshift.asm b/mpn/powerpc32/rshift.asm
index 62627fecd..f32db73b5 100644
--- a/mpn/powerpc32/rshift.asm
+++ b/mpn/powerpc32/rshift.asm
@@ -1,6 +1,10 @@
 dnl PowerPC-32 mpn_rshift -- Shift a number right.
+dnl
+dnl      cycles/limb
+dnl 604e:    3.3
+dnl 750:     4.0
 
-dnl Copyright 1995, 2000 Free Software Foundation, Inc.
+dnl Copyright 1995, 2000, 2002 Free Software Foundation, Inc.
 
 dnl This file is part of the GNU MP Library.
 
@@ -35,26 +39,28 @@ PROLOGUE(mpn_rshift)
 	subfic	r8,r6,32
 	lwz	r11,0(r4)	C load first s1 limb
 	slw	r3,r11,r8	C compute function return value
-	bdz	.Lend1
+	bdz	L(end1)
 
-.Loop:	lwzu	r10,4(r4)
+L(oop):	lwzu	r10,4(r4)
 	srw	r9,r11,r6
 	slw	r12,r10,r8
 	or	r9,r9,r12
 	stwu	r9,4(r7)
-	bdz	.Lend2
+	bdz	L(end2)
 	lwzu	r11,4(r4)
 	srw	r9,r10,r6
 	slw	r12,r11,r8
 	or	r9,r9,r12
 	stwu	r9,4(r7)
-	bdnz	.Loop
+	bdnz	L(oop)
 
-.Lend1:	srw	r0,r11,r6
+L(end1):
+	srw	r0,r11,r6
 	stw	r0,4(r7)
 	blr
 
-.Lend2:	srw	r0,r10,r6
+L(end2):
+	srw	r0,r10,r6
 	stw	r0,4(r7)
 	blr
 EPILOGUE(mpn_rshift)
diff --git a/mpn/powerpc32/sub_n.asm b/mpn/powerpc32/sub_n.asm
index 354a86510..4993c5bff 100644
--- a/mpn/powerpc32/sub_n.asm
+++ b/mpn/powerpc32/sub_n.asm
@@ -1,7 +1,11 @@
 dnl PowerPC-32 mpn_sub_n -- Subtract two limb vectors of the same length > 0
 dnl and store difference in a third limb vector.
+dnl
+dnl      cycles/limb
+dnl 604e:   2.67
+dnl 750:    4.5
 
-dnl Copyright 1995, 1997, 2000 Free Software Foundation, Inc.
+dnl Copyright 1995, 1997, 2000, 2002 Free Software Foundation, Inc.
 
 dnl This file is part of the GNU MP Library.
 
@@ -36,24 +40,26 @@ PROLOGUE(mpn_sub_n)
 	lwz	r8,0(r4)	C load least significant s1 limb
 	lwz	r0,0(r5)	C load least significant s2 limb
 	addi	r3,r3,-4	C offset res_ptr, it's updated before it's used
-	bdz	.Lend		C If done, skip loop
-.Loop:	lwz	r9,4(r4)	C load s1 limb
+	bdz	L(end)		C If done, skip loop
+L(oop):	lwz	r9,4(r4)	C load s1 limb
 	lwz	r10,4(r5)	C load s2 limb
 	subfe	r7,r0,r8	C subtract limbs with cy, set cy
 	stw	r7,4(r3)	C store result limb
-	bdz	.Lexit		C decrement CTR and exit if done
+	bdz	L(exit)		C decrement CTR and exit if done
 	lwzu	r8,8(r4)	C load s1 limb and update s1_ptr
 	lwzu	r0,8(r5)	C load s2 limb and update s2_ptr
 	subfe	r7,r10,r9	C subtract limbs with cy, set cy
 	stwu	r7,8(r3)	C store result limb and update res_ptr
-	bdnz	.Loop		C decrement CTR and loop back
+	bdnz	L(oop)		C decrement CTR and loop back
 
-.Lend:	subfe	r7,r0,r8
+L(end):
+	subfe	r7,r0,r8
 	stw	r7,4(r3)	C store ultimate result limb
 	subfe	r3,r0,r0	C load !cy into ...
 	subfic	r3,r3,0		C ... return value register
 	blr
-.Lexit:	subfe	r7,r10,r9
+L(exit):
+	subfe	r7,r10,r9
 	stw	r7,8(r3)
 	subfe	r3,r0,r0	C load !cy into ...
 	subfic	r3,r3,0		C ... return value register