No global value numbering / duplicate removal

sums.c

#include <stdio.h>

void b01_sums(int size, int* A)
{
  int sum0 = 0;
  int sum1 = 0;
  for (int i = 0; i != size; ++i)
  {
    sum0 = sum0 + A[i];
    sum1 = sum1 + A[i];
    printf("sum0: %d\nsum1: %d\n", sum0, sum1);
  }
}

sums.ll

; generated with
; gcc sums.c -c -O3 -save-temps -emit-llvm
; opt sums.bc -gvn -S -O3

; note that there are still two loop accumulators (sum0 and sum1)

define void @b01_sums(i32 %size, i32* nocapture readonly %A) #0 {
  %1 = icmp eq i32 %size, 0
  br i1 %1, label %._crit_edge, label %.lr.ph.preheader

.lr.ph.preheader:                                 ; preds = %0
  %2 = add i32 %size, -1
  br label %.lr.ph

.lr.ph:                                           ; preds = %.lr.ph.preheader, %.lr.ph
  %indvars.iv = phi i64 [ %indvars.iv.next, %.lr.ph ], [ 0, %.lr.ph.preheader ]

  ; two sum accumulators, both the same
  %sum1.02 = phi i32 [ %6, %.lr.ph ], [ 0, %.lr.ph.preheader ]
  %sum0.01 = phi i32 [ %5, %.lr.ph ], [ 0, %.lr.ph.preheader ]
  %3 = getelementptr inbounds i32* %A, i64 %indvars.iv
  %4 = load i32* %3, align 4, !tbaa !1
  %5 = add nsw i32 %4, %sum0.01
  %6 = add nsw i32 %4, %sum1.02

  %7 = tail call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([19 x i8]* @.str, i64 0, i64 0), i32 %5, i32 %6) #2
  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
  %lftr.wideiv1 = trunc i64 %indvars.iv to i32
  %exitcond2 = icmp eq i32 %lftr.wideiv1, %2
  br i1 %exitcond2, label %._crit_edge.loopexit, label %.lr.ph

._crit_edge.loopexit:                             ; preds = %.lr.ph
  br label %._crit_edge

._crit_edge:                                      ; preds = %._crit_edge.loopexit, %0
  ret void
}

sums.s

# generated with:
# gcc sums.c -c -O3 -save-temps

	# r15  = A[i]
	# r14d = size
	testl	%r14d, %r14d
	je	LBB0_3
## BB#1:
  # r13d = sum0 = 0
	xorl	%r13d, %r13d
  # literal printf string
	leaq	L_.str(%rip), %r12
	# ebx = sum1 = 0
	xorl	%ebx, %ebx

# start loop
LBB0_2:                                 ## %.lr.ph
                                        ## =>This Inner Loop Header: Depth=1

  # A[i] (only once)
	movl	(%r15), %eax
	# add sum0 and sum1
	addl	%eax, %ebx
	addl	%eax, %r13d

  # printf call
	xorl	%eax, %eax
	movq	%r12, %rdi
	movl	%ebx, %esi
	movl	%r13d, %edx
	callq	_printf
	
	# increment etc
	addq	$4, %r15
	decl	%r14d
	jne	LBB0_2
LBB0_3:                                 ## %._crit_edge