about summary refs log tree commit diff
path: root/sysdeps/sparc/sparc32/addmul_1.S
diff options
context:
space:
mode:
Diffstat (limited to 'sysdeps/sparc/sparc32/addmul_1.S')
-rw-r--r--sysdeps/sparc/sparc32/addmul_1.S208
1 files changed, 90 insertions, 118 deletions
diff --git a/sysdeps/sparc/sparc32/addmul_1.S b/sysdeps/sparc/sparc32/addmul_1.S
index 1028f19b54..4708393aa5 100644
--- a/sysdeps/sparc/sparc32/addmul_1.S
+++ b/sysdeps/sparc/sparc32/addmul_1.S
@@ -1,146 +1,118 @@
-! SPARC __mpn_addmul_1 -- Multiply a limb vector with a limb and add
-! the result to a second limb vector.
-!
+! SPARC v8 __mpn_addmul_1 -- Multiply a limb vector with a limb and
+! add the result to a second limb vector.
+
 ! Copyright (C) 1992-2019 Free Software Foundation, Inc.
-!
+
 ! This file is part of the GNU MP Library.
-!
+
 ! The GNU MP Library is free software; you can redistribute it and/or modify
 ! it under the terms of the GNU Lesser General Public License as published by
 ! the Free Software Foundation; either version 2.1 of the License, or (at your
 ! option) any later version.
-!
+
 ! The GNU MP Library is distributed in the hope that it will be useful, but
 ! WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
 ! or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
 ! License for more details.
-!
+
 ! You should have received a copy of the GNU Lesser General Public License
 ! along with the GNU MP Library; see the file COPYING.LIB.  If not,
 ! see <https://www.gnu.org/licenses/>.
 
 
 ! INPUT PARAMETERS
-! RES_PTR	o0
-! S1_PTR	o1
-! SIZE		o2
-! S2_LIMB	o3
+! res_ptr	o0
+! s1_ptr	o1
+! size		o2
+! s2_limb	o3
 
 #include <sysdep.h>
 
 ENTRY(__mpn_addmul_1)
-	! Make S1_PTR and RES_PTR point at the end of their blocks
-	! and put (- 4 x SIZE) in index/loop counter.
-	sll	%o2,2,%o2
-	add	%o0,%o2,%o4	! RES_PTR in o4 since o0 is retval
-	add	%o1,%o2,%o1
-	sub	%g0,%o2,%o2
+	ld	[%o1+0],%o4	! 1
+	sll	%o2,4,%g1
+	orcc	%g0,%g0,%g2
+	mov	%o7,%g4			! Save return address register
+	and	%g1,(4-1)<<4,%g1
+1:	call	2f
+	 add	%o7,3f-1b,%g3
+2:	jmp	%g3+%g1
+	 mov	%g4,%o7			! Restore return address register
 
-	cmp	%o3,0xfff
-	bgu	LOC(large)
+	.align	4
+3:
+LOC(00):
+	add	%o0,-4,%o0
+	b	LOC(loop00)		/* 4, 8, 12, ... */
+	 add	%o1,-4,%o1
+	nop
+LOC(01):
+	b	LOC(loop01)		/* 1, 5, 9, ... */
+	 nop
+	nop
+	nop
+LOC(10):
+	add	%o0,-12,%o0	/* 2, 6, 10, ... */
+	b	LOC(loop10)
+	 add	%o1,4,%o1
+	nop
+LOC(11):
+	add	%o0,-8,%o0	/* 3, 7, 11, ... */
+	b	LOC(loop11)
+	 add	%o1,-8,%o1
 	nop
 
-	ld	[%o1+%o2],%o5
-	mov	0,%o0
-	b	LOC(0)
-	 add	%o4,-4,%o4
-LOC(loop0):
-	addcc	%o5,%g1,%g1
-	ld	[%o1+%o2],%o5
-	addx	%o0,%g0,%o0
-	st	%g1,[%o4+%o2]
-LOC(0):	wr	%g0,%o3,%y
-	sra	%o5,31,%g2
-	and	%o3,%g2,%g2
-	andcc	%g1,0,%g1
-	mulscc	%g1,%o5,%g1
- 	mulscc	%g1,%o5,%g1
- 	mulscc	%g1,%o5,%g1
- 	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,%o5,%g1
-	mulscc	%g1,0,%g1
-	sra	%g1,20,%g4
-	sll	%g1,12,%g1
- 	rd	%y,%g3
-	srl	%g3,20,%g3
-	or	%g1,%g3,%g1
-
-	addcc	%g1,%o0,%g1
-	addx	%g2,%g4,%o0	! add sign-compensation and cy to hi limb
-	addcc	%o2,4,%o2	! loop counter
-	bne	LOC(loop0)
-	 ld	[%o4+%o2],%o5
-
-	addcc	%o5,%g1,%g1
-	addx	%o0,%g0,%o0
-	retl
-	st	%g1,[%o4+%o2]
-
-
-LOC(large):
-	ld	[%o1+%o2],%o5
-	mov	0,%o0
-	sra	%o3,31,%g4	! g4 = mask of ones iff S2_LIMB < 0
-	b	LOC(1)
-	 add	%o4,-4,%o4
 LOC(loop):
-	addcc	%o5,%g3,%g3
-	ld	[%o1+%o2],%o5
-	addx	%o0,%g0,%o0
-	st	%g3,[%o4+%o2]
-LOC(1):	wr	%g0,%o5,%y
-	and	%o5,%g4,%g2
-	andcc	%g0,%g0,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%o3,%g1
-	mulscc	%g1,%g0,%g1
-	rd	%y,%g3
-	addcc	%g3,%o0,%g3
-	addx	%g2,%g1,%o0
-	addcc	%o2,4,%o2
-	bne	LOC(loop)
-	 ld	[%o4+%o2],%o5
+	addcc	%g3,%g2,%g3	! 1
+	ld	[%o1+4],%o4	! 2
+	rd	%y,%g2		! 1
+	addx	%g0,%g2,%g2
+	ld	[%o0+0],%g1	! 2
+	addcc	%g1,%g3,%g3
+	st	%g3,[%o0+0]	! 1
+LOC(loop00):
+	umul	%o4,%o3,%g3	! 2
+	ld	[%o0+4],%g1	! 2
+	addxcc	%g3,%g2,%g3	! 2
+	ld	[%o1+8],%o4	! 3
+	rd	%y,%g2		! 2
+	addx	%g0,%g2,%g2
+	nop
+	addcc	%g1,%g3,%g3
+	st	%g3,[%o0+4]	! 2
+LOC(loop11):
+	umul	%o4,%o3,%g3	! 3
+	addxcc	%g3,%g2,%g3	! 3
+	ld	[%o1+12],%o4	! 4
+	rd	%y,%g2		! 3
+	add	%o1,16,%o1
+	addx	%g0,%g2,%g2
+	ld	[%o0+8],%g1	! 2
+	addcc	%g1,%g3,%g3
+	st	%g3,[%o0+8]	! 3
+LOC(loop10):
+	umul	%o4,%o3,%g3	! 4
+	addxcc	%g3,%g2,%g3	! 4
+	ld	[%o1+0],%o4	! 1
+	rd	%y,%g2		! 4
+	addx	%g0,%g2,%g2
+	ld	[%o0+12],%g1	! 2
+	addcc	%g1,%g3,%g3
+	st	%g3,[%o0+12]	! 4
+	add	%o0,16,%o0
+	addx	%g0,%g2,%g2
+LOC(loop01):
+	addcc	%o2,-4,%o2
+	bg	LOC(loop)
+	 umul	%o4,%o3,%g3	! 1
 
-	addcc	%o5,%g3,%g3
-	addx	%o0,%g0,%o0
+	addcc	%g3,%g2,%g3	! 4
+	rd	%y,%g2		! 4
+	addx	%g0,%g2,%g2
+	ld	[%o0+0],%g1	! 2
+	addcc	%g1,%g3,%g3
+	st	%g3,[%o0+0]	! 4
 	retl
-	st	%g3,[%o4+%o2]
+	 addx	%g0,%g2,%o0
 
 END(__mpn_addmul_1)