[use macros to allow switching between 4-limb and 8-limb unroll
peter@cordes.ca**20080324205442
 also make it easy to move the add/lea up/down in the loop
] {
hunk ./rshift.asm 362
+
+C still need to comment/uncomment loop counters/pointer increments.  and the ALIGN(16)
+ define(`C2_UNROLL8')
+C define(`C2_8REG')
+
hunk ./rshift.asm 368
-define(n,%rdx)
-define(reg1,%r9)
-define(reg2,%rax)
-define(reg3,%r8)  		C referenced the fewest times
-define(reg4,%r11)
-define(reg5,%r12)
-define(reg6,%r13)
-define(reg7,%r14)
-define(reg8,%r15)
-C define(reg8,reg4)
+define(`n',%rdx)
+define(`reg1',%r9)
+define(`reg2',%rax)
+define(`reg3',%r8)  		C referenced the fewest times
+define(`reg4',%r11)
+C  pipeline depth is only 4, so we can avoid overlap with only 4 regs.
+ifdef(`C2_8REG', `
+ define(`reg5',%r12)
+ define(`reg6',%r13)
+ define(`reg7',%r14)
+ define(`reg8',%r15)
+	', `
+ define(reg5,reg1)
+ define(reg6,reg2)
+ define(reg7,reg3)
+ define(reg8,reg4)
+')
+
+ifdef(`C2_8REG', `
+  	push	%r12
+  	push	%r13
+  	push	%r14
+  	push	%r15
+', )
hunk ./rshift.asm 393
-C 	push	reg2
-C 	push	reg4
- 	push	%r12
- 	push	%r13
- 	push	%r14
- 	push	%r15
hunk ./rshift.asm 412
-	test	$7,	%dl
+ifdef(`C2_UNROLL8',
+	`test	$7,	%dl'
+,	`test	$3,	%dl' )
hunk ./rshift.asm 439
+define(`srcoff',0)
+define(`dstoff',0)
hunk ./rshift.asm 442
-L(c2_loop):	shrd	%cl, reg1, reg8
+L(c2_loop):
+	shrd	%cl, reg1, reg8
hunk ./rshift.asm 447
-C   	add	$32,	%rsi
+C	add	$32,	%rsi; define(`srcoff',32)
hunk ./rshift.asm 449
-	mov	reg1,	8(%rdi)
-	mov	(8-64)(%rsi),	reg5
-   	lea	64(%rsi),%rsi
+	mov	reg1,	(8-dstoff)(%rdi)
+	mov	(8-srcoff)(%rsi),	reg5
+	lea	64(%rsi),%rsi;  define(`srcoff',64)
hunk ./rshift.asm 454
-C   	add	$32,	%rdi
-	mov	reg2,	(16-0)(%rdi)
-	mov	(16-64)(%rsi),	reg6
+C	add	$32,	%rdi; define(`dstoff',32)
+	mov	reg2,	(16-dstoff)(%rdi)
+	mov	(16-srcoff)(%rsi),	reg6
hunk ./rshift.asm 459
-C   	sub	$4, n
- 	mov	reg3,	(24-0)(%rdi)
-	mov	(24-64)(%rsi),	reg7
+C	sub	$4, n
+	mov	reg3,	(24-dstoff)(%rdi)
+	mov	(24-srcoff)(%rsi),	reg7
hunk ./rshift.asm 463
-  	shrd	%cl, reg5, reg4
-  	mov	(32-64)(%rsi),	reg8
-  	mov	reg4,	(32-0)(%rdi)
-  	lea	64(%rdi),%rdi
- C	L(c2_10):
-  	shrd	%cl, reg6, reg5
-  	mov	(40-64)(%rsi),	reg1
-  	mov	reg5,	(40-64)(%rdi)
- C	L(c2_01):
-  	shrd	%cl, reg7, reg6
-  	mov	(48-64)(%rsi),	reg2
-  	mov	reg6,	(48-64)(%rdi)
- C	L(c2_00):
-  	shrd	%cl, reg8, reg7
-  	sub	$8, n
-  	mov	(56-64)(%rsi),	reg3
-  	mov	reg7,	(56-64)(%rdi)
hunk ./rshift.asm 464
+ifdef(`C2_UNROLL8', `
+	shrd	%cl, reg5, reg4
+	mov	reg4,	(32-dstoff)(%rdi)
+	lea	64(%rdi),%rdi; define(`dstoff',64)
+	mov	(32-srcoff)(%rsi),	reg8
+C	L(c2_10):
+	shrd	%cl, reg6, reg5
+	mov	reg5,	(40-dstoff)(%rdi)
+	mov	(40-srcoff)(%rsi),	reg1
+C	L(c2_01):
+	shrd	%cl, reg7, reg6
+	mov	reg6,	(48-dstoff)(%rdi)
+	mov	(48-srcoff)(%rsi),	reg2
+C	L(c2_00):
+	shrd	%cl, reg8, reg7
+	sub	$8, n
+	mov	reg7,	(56-dstoff)(%rdi)
+	mov	(56-srcoff)(%rsi),	reg3
+') C endif
hunk ./rshift.asm 490
-C  	pop	reg4
-C  	pop	reg2
-	pop	%r15
-	pop	%r14
-	pop	%r13
-	pop	%r12
+ifdef(`C2_8REG', `
+ 	pop	%r15
+ 	pop	%r14
+ 	pop	%r13
+ 	pop	%r12
+',)
}