Unterschied zwischen parallel + for getrennt und zusammen

--- a5.asm	2012-05-07 22:07:15.000000000 +0200
+++ ../immortal.asm	2012-05-07 22:20:47.000000000 +0200
@@ -73,2 +72,0 @@
-	xorl	%r13d, %r13d	# sum
-	.cfi_offset 13, -16
@@ -78,0 +77 @@
+	.cfi_offset 13, -16
@@ -85 +84 @@
-	movl	20(%rdi), %ebx	# <variable>.halfway, D.28120
+	movl	16(%rdi), %ebx	# <variable>.halfway, D.28120
@@ -90,0 +90 @@
+	xorl	%esi, %esi	# sum
@@ -94 +94 @@
-	movl	%ebx, %r8d	# D.28120, D.28132
+	movl	%ebx, %r9d	# D.28120, D.28132
@@ -96,6 +96,6 @@
-	movl	$1, %r10d	#, tmp83
-	cmovbe	%edx, %r8d	# m,, D.28132
-	movq	(%r12), %rsi	# <variable>.a, pretmp.315
-	movl	16(%r12), %r11d	# <variable>.digits, pretmp.317
-	leal	(%rbp,%rax), %r9d	#, ivtmp.352
-	subl	%ebp, %r10d	# D.28122, tmp83
+	movl	$1, %r11d	#, tmp83
+	cmovbe	%edx, %r9d	# m,, D.28132
+	movq	8(%r12), %rdi	# <variable>.a, pretmp.315
+	movl	20(%r12), %r13d	# <variable>.digits, pretmp.317
+	leal	(%rbp,%rax), %r10d	#, ivtmp.352
+	subl	%ebp, %r11d	# D.28122, tmp83
@@ -105,2 +105,2 @@
-	movl	%r11d, %ecx	# pretmp.317, ivtmp.341
-	incl	%r8d	# D.28134
+	movl	%r13d, %ecx	# pretmp.317, ivtmp.341
+	incl	%r9d	# D.28134
@@ -112,3 +112,3 @@
-	mov	%edx, %edi	# m, m
-	movzbl	(%rsi,%rax), %eax	#* ivtmp.341, tmp77
-	movzbl	(%rsi,%rdi), %edi	#* m, tmp76
+	mov	%edx, %r8d	# m, m
+	movzbl	(%rdi,%rax), %eax	#* ivtmp.341, tmp77
+	movzbl	(%rdi,%r8), %r8d	#* m, tmp76
@@ -116 +116 @@
-	imull	%edi, %eax	# tmp76, tmp77
+	imull	%r8d, %eax	# tmp76, tmp77
@@ -119,2 +119,2 @@
-	addq	%rax, %r13	# tmp78, sum
-	cmpl	%edx, %r8d	# m, D.28134
+	addq	%rax, %rsi	# tmp78, sum
+	cmpl	%edx, %r9d	# m, D.28134
@@ -122,2 +122,2 @@
-	addl	%ebp, %r9d	# D.28122, ivtmp.352
-	movl	%r9d, %eax	# ivtmp.352, tmp79
+	addl	%ebp, %r10d	# D.28122, ivtmp.352
+	movl	%r10d, %eax	# ivtmp.352, tmp79
@@ -127 +127 @@
-	leal	(%r10,%r9), %edx	#, m
+	leal	(%r11,%r10), %edx	#, m
@@ -129,2 +129,2 @@
-	movl	%edx, %r8d	# m, D.28132
-	cmovbe	%ebx, %r8d	# D.28120,, D.28132
+	movl	%edx, %r9d	# m, D.28132
+	cmovbe	%ebx, %r9d	# D.28120,, D.28132
@@ -135,2 +135 @@
-	call	GOMP_barrier	#
-	lock addq	%r13, 8(%r12)	# sum,
+	lock addq	%rsi, (%r12)	# sum,* .omp_data_i
@@ -1162,4 +1161,3 @@
-	movq	%r13, 4176(%rsp)	# a, .omp_data_o.176.a
-	movl	%edi, 4192(%rsp)	#, .omp_data_o.176.digits
-	movl	%edi, %eax	#, tmp630
-	movq	$0, 4184(%rsp)	#, .omp_data_o.176.sum
+	movq	%r13, 4184(%rsp)	# a, .omp_data_o.176.a
+	movl	36(%rsp), %eax	# %sfp, tmp630
+	movl	%edi, 4196(%rsp)	#, .omp_data_o.176.digits
@@ -1168 +1166,2 @@
-	movl	%eax, 4196(%rsp)	# tmp630, .omp_data_o.176.halfway
+	movl	%eax, 4192(%rsp)	# tmp630, .omp_data_o.176.halfway
+	movq	$0, 4176(%rsp)	#, .omp_data_o.176.sum
@@ -1173,2 +1172,2 @@
-	movl	4192(%rsp), %eax	# .omp_data_o.176.digits,
-	movq	4176(%rsp), %r13	# .omp_data_o.176.a, a
+	movl	4196(%rsp), %eax	# .omp_data_o.176.digits,
+	movq	4184(%rsp), %r13	# .omp_data_o.176.a, a
@@ -1176 +1175 @@
-	movq	4184(%rsp), %rax	# .omp_data_o.176.sum, .omp_data_o.176.sum
+	movq	4176(%rsp), %rax	# .omp_data_o.176.sum, .omp_data_o.176.sum