[llvm-commits] [llvm] r44687 - in /llvm/trunk: include/llvm/CodeGen/Passes.h lib/CodeGen/LLVMTargetMachine.cpp lib/CodeGen/MachineLICM.cpp lib/Target/PowerPC/PPCInstrInfo.td

[Evan Cheng]

On Dec 10, 2007, at 1:23 PM, Bill Wendling wrote:

> On Dec 10, 2007 11:36 AM, Evan Cheng <evan.cheng at apple.com> wrote:
>> On Dec 10, 2007, at 6:47 AM, Dan Gohman wrote:
>>> And if that's all it's doing, MachineLICM's traversal could be
>>> simplified
>>> a little. Instead of visiting each loop individually, taking care to
>>> avoid
>>> revisiting to blocks within inner loops, the pass could just  
>>> traverse
>>> entire outer-most loops, which will implicitly include the blocks of
>>> any
>>> inner loops. Instructions nested deep in inner loops can then be
>>> hoisted
>>> all the way out of the outer-most loop in a single step instead of
>>> being
>>> hoisted out one loop at a time.
>>
>> I don't think lifting loop invariant from inner loop all the way out
>> of outer-most loop is a good idea. That will increase register
>> pressure in basic blocks where it is not used.
>>
> This is going to happen with the current pass, though. Each loop is
> going to see the hoisted instructions from the previous iteration and
> try to re-hoist them. Is there some heuristic we should apply to
> prevent it from hoisting instructions too far?
>

I am not sure. :-)

For innermost loops, hoisting invariants out into the preheader always  
make sense. Intuitively, hoisting invariants from inner loops out of  
the outermost loop only makes sense when all (or a lot, whatever that  
means :-) of the inner loops use it. Or at least the first inner loop  
use it.

What does the LLVM level LICM do?

Evan

> -bw
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits