[llvm-commits] PATCH: Fix very slow algorithm in RescheduleMIBelowKill

[Chandler Carruth]

Hello,

This patch removes an badly scaling code path from RescheduleMIBelowKill in
the twoaddr pass.

The problem is essentially that we walk the instructions of the basic
blocks of a function, and call RescheduleMIBelowKill for a large number of
them. It in turn calls findLocalKill, which then walks *the entire* use/def
chain for the register for the function. For functions with a very large
number of register uses, this starts to scale very poorly. =]

There is a better algorithm to use here. Right after we do findLocalKill,
we walk the instructions linearly from the original instruction to the
kill, checking if dependencies have been violated. This already has a cap
to prevent n^2 complexity on very large basic blocks.

If we fuse the search for the kill with the checking for dependencies we
can share the single linear walk and the single cap to prevent n^2 behavior.

This requires a bit of reshuffling, and I've tried to start pulling out
helper functions here. However, there is a lot more refactoring that could
be done here, and in particular, we should factor this enough to allow
RescheduleKillAboveMI to use the same algorithm and share most of the code.
I'd prefer to do that in follow-up patches. The latter function doesn't
show up as terribly hot in my profiles.

For some of the test cases from PR13225, after I applied the patch mailed
out earlier to generate basic blocks in the natural CFG order, this moves
findLocalKill (and its callers) from over 60% of the CPU profile, to below
0.25%. ;] The constant factors in this one were really really high.

The only remaining issue is MarkVirtRegAliveInBlock.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120712/b5eab129/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: twoaddr-linearize.patch
Type: application/octet-stream
Size: 7688 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20120712/b5eab129/attachment.obj>