unsplatable values into memset_pattern16 when it is available
(recent darwins).  This transforms lots of strided loop stores
of ints for example, like 5 in vpr:

  Formed memset:   call void @memset_pattern16(i8* %4, i8* getelementptr inbounds ([16 x i8]* @.memset_pattern9, i32 0, i32 0), i64 %tmp25)
    from store to: {%3,+,4}<%11> at:   store i32 3, i32* %scevgep, align 4, !tbaa !4

llvm-svn: 126040

to hack on memset, memcpy etc.

llvm-svn: 125974

llvm-svn: 125563

when safe.

The testcase is basically this nested loop:
void foo(char *X) {
  for (int i = 0; i != 100; ++i) 
    for (int j = 0; j != 100; ++j)
      X[j+i*100] = 0;
}

which gets turned into a single memset now.  clang -O3 doesn't optimize
this yet though due to a phase ordering issue I haven't analyzed yet.

llvm-svn: 122806

instruction *after* the store.  The store will always be deleted
if the transformation kicks in, so we'd do an N^2 scan of every
loop block.  Whoops.

llvm-svn: 122805

stop setting NSW: signed overflow is possible.  Thanks to Dan
for pointing these out.

llvm-svn: 122790

llvm-svn: 122788

llvm-svn: 122720

Teach it to CSE the rest of the non-side-effecting instructions.

llvm-svn: 122716

sure that the loop we're promoting into a memcpy doesn't mutate the input
of the memcpy.  Before we were just checking that the dest of the memcpy
wasn't mod/ref'd by the loop.

llvm-svn: 122712

mess with it.  We'd rather peel/unroll it than convert all of its 
stores into memsets.

llvm-svn: 122711

blocks in a loop, instead of just the header block.  This makes it more
aggressive, able to handle Duncan's Ada examples.

llvm-svn: 122704

llvm-svn: 122701

header for now for memset/memcpy opportunities.  It turns out that loop-rotate
is successfully rotating loops, but *DOESN'T MERGE THE BLOCKS*, turning "for 
loops" into 2 basic block loops that loop-idiom was ignoring.

With this fix, we form many *many* more memcpy and memsets than before, including
on the "history" loops in the viterbi benchmark, which look like this:

        for (j=0; j<MAX_history; ++j) {
          history_new[i][j+1] = history[2*i][j];
        }

Transforming these loops into memcpy's speeds up the viterbi benchmark from
11.98s to 3.55s on my machine.  Woo.

llvm-svn: 122685

llvm-svn: 122683

llvm-svn: 122682

llvm-svn: 122678

new testcase.

llvm-svn: 122662

aggressively.  In practice, this doesn't help anything though,
see the todo.

llvm-svn: 122660

should be correct now.

llvm-svn: 122659

check for "multiple of a byte" in size to make it clear that the
>> 3 below is safe.

llvm-svn: 122604

llvm-svn: 122593

llvm-svn: 122585

llvm-svn: 122574

memsets.  This is still missing one important validity check, but this is enough
to compile stuff like this:

void test0(std::vector<char> &X) {
  for (std::vector<char>::iterator I = X.begin(), E = X.end(); I != E; ++I)
    *I = 0;
}

void test1(std::vector<int> &X) {
  for (long i = 0, e = X.size(); i != e; ++i)
    X[i] = 0x01010101;
}

With:
 $ clang t.cpp -S -o - -O2 -emit-llvm | opt -loop-idiom | opt -O3 | llc 

to:

__Z5test0RSt6vectorIcSaIcEE:            ## @_Z5test0RSt6vectorIcSaIcEE
## BB#0:                                ## %entry
	subq	$8, %rsp
	movq	(%rdi), %rax
	movq	8(%rdi), %rsi
	cmpq	%rsi, %rax
	je	LBB0_2
## BB#1:                                ## %bb.nph
	subq	%rax, %rsi
	movq	%rax, %rdi
	callq	___bzero
LBB0_2:                                 ## %for.end
	addq	$8, %rsp
	ret
...
__Z5test1RSt6vectorIiSaIiEE:            ## @_Z5test1RSt6vectorIiSaIiEE
## BB#0:                                ## %entry
	subq	$8, %rsp
	movq	(%rdi), %rax
	movq	8(%rdi), %rdx
	subq	%rax, %rdx
	cmpq	$4, %rdx
	jb	LBB1_2
## BB#1:                                ## %for.body.preheader
	andq	$-4, %rdx
	movl	$1, %esi
	movq	%rax, %rdi
	callq	_memset
LBB1_2:                                 ## %for.end
	addq	$8, %rsp
	ret

llvm-svn: 122573

llvm-svn: 122567

llvm-svn: 122563