Commit Graph

26 Commits

Author SHA1 Message Date
Dan Gohman
47ac0f0c7c When scheduling a block in parts, keep track of the overall
instruction index across each part. Instruction indices are used
to make live range queries, and live ranges can extend beyond
scheduling region boundaries.

Refactor the ScheduleDAGSDNodes class some more so that it
doesn't have to worry about this additional information.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@64288 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-11 04:27:20 +00:00
Dan Gohman
9e64bbb322 Factor out more code for computing register live-range informationfor
scheduling, and generalize is so that preserves state across
scheduling regions. This fixes incorrect live-range information around
terminators and labels, which are effective region boundaries.

In place of looking for terminators to anchor inter-block dependencies,
introduce special entry and exit scheduling units for this purpose.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@64254 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-10 23:27:53 +00:00
Dan Gohman
84fbac5809 Move ScheduleDAGSDNodes.h to be a private header. Front-ends
that previously included this header should include
SchedulerRegistry.h instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@63937 91177308-0d34-0410-b5e6-96231b3b80d8
2009-02-06 17:22:58 +00:00
Dan Gohman
79ce276083 Move a few containers out of ScheduleDAGInstrs::BuildSchedGraph
and into the ScheduleDAGInstrs class, so that they don't get
destructed and re-constructed for each block. This fixes a
compile-time hot spot in the post-pass scheduler.

To help facilitate this, tidy and do some minor reorganization
in the scheduler constructor functions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@62275 91177308-0d34-0410-b5e6-96231b3b80d8
2009-01-15 19:20:50 +00:00
Evan Cheng
c29a56dedb Fix PR3241: Currently EmitCopyFromReg emits a copy from the physical register to a virtual register unless it requires an expensive cross class copy. That means we are only treating "expensive to copy" register dependency as physical register dependency.
Also future proof the scheduler to handle "normal" physical register dependencies. The code is not exercised yet.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@62074 91177308-0d34-0410-b5e6-96231b3b80d8
2009-01-12 03:19:55 +00:00
Dan Gohman
c9a5b9e38b Rename BuildSchedUnits to BuildSchedGraph, and refactor the
code in ScheduleDAGSDNodes' BuildSchedGraph into separate functions.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61376 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-23 18:36:58 +00:00
Dan Gohman
3f23744df4 Fix some register-alias-related bugs in the post-RA scheduler liveness
computation code. Also, avoid adding output-depenency edges when both
defs are dead, which frequently happens with EFLAGS defs.

Compute Depth and Height lazily, and always in terms of edge latency
values. For the schedulers that don't care about latency, edge latencies
are set to 1.

Eliminate Cycle and CycleBound, and LatencyPriorityQueue's Latencies array.
These are all subsumed by the Depth and Height fields.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61073 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-16 03:25:46 +00:00
Dan Gohman
ffa391272b Make addPred and removePred return void, since the return value is not
currently used by anything.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@61066 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-16 01:00:55 +00:00
Dan Gohman
54e4c36a73 Rewrite the SDep class, and simplify some of the related code.
The Cost field is removed. It was only being used in a very limited way,
to indicate when the scheduler should attempt to protect a live register,
and it isn't really needed to do that. If we ever want the scheduler to
start inserting copies in non-prohibitive situations, we'll have to
rethink some things anyway.

A Latency field is added. Instead of giving each node a single
fixed latency, each edge can have its own latency. This will eventually
be used to model various micro-architecture properties more accurately.

The PointerIntPair class and an internal union are now used, which
reduce the overall size.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@60806 91177308-0d34-0410-b5e6-96231b3b80d8
2008-12-09 22:54:47 +00:00
Dan Gohman
942b8c8533 Check in the rest of this change. The isAntiDep flag needs to be passed
to removePred because an SUnit can both data-depend and anti-depend
on the same SUnit.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59969 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-24 17:33:52 +00:00
Dan Gohman
cddd428459 Add a flag to SDep for tracking which edges are anti-dependence edges.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59785 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-21 02:27:52 +00:00
Dan Gohman
98adea1149 Rename SDep's isSpecial to isArtificial, to make this field a little
less mysterious.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59782 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-21 02:18:56 +00:00
Dan Gohman
47d1a214a7 Change these schedulers to not emit no-ops. It turns out that
the RR scheduler actually does look at latency values, but it
doesn't use a hazard recognizer so it has no way to know when
a no-op is needed, as opposed to just stalling and incrementing
the cycle count.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59759 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-21 00:10:42 +00:00
Dan Gohman
cdb260de83 Simplify this code a little. In the fast scheduler, CreateNewSUnit
and CreateClone don't add any extra value.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59679 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-19 23:39:02 +00:00
Dan Gohman
343f0c0467 Experimental post-pass scheduling support. Post-pass scheduling
is currently off by default, and can be enabled with
-disable-post-RA-scheduler=false.

This doesn't have a significant impact on most code yet because it doesn't
yet do anything to address anti-dependencies and it doesn't attempt to
disambiguate memory references. Also, several popular targets
don't have pipeline descriptions yet.

The majority of the changes here are splitting the SelectionDAG-specific
code out of ScheduleDAG, so that ScheduleDAG can be moved to
libLLVMCodeGen.a. The interface between ScheduleDAG-using code and
the rest of the scheduling code is somewhat rough and will evolve.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59676 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-19 23:18:57 +00:00
Dan Gohman
1256f5fe76 Tidy up ScheduleNodeBottomUp methods, and make them more
consistent with ScheduleNodeTopDown methods.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59550 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-18 21:22:20 +00:00
Dan Gohman
3cc6243ddf Change SUnit's dump method to take a ScheduleDAG* instead of
a SelectionDAG*.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59488 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-18 02:06:40 +00:00
Dan Gohman
2d093f3560 Avoid using a loop in ReleasePred and ReleaseSucc methods to compute the
new CycleBound value. Instead, just update CycleBound on each call.
Also, make ReleasePred and ReleaseSucc methods more consistent accross
the various schedulers.

This also happens to make ScheduleDAGRRList's CycleBound computation
somewhat more interesting, though it still doesn't have any noticeable
effect, because no current targets that use the register-pressure
reduction scheduler provide pipeline models.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59475 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-18 00:38:59 +00:00
Dan Gohman
e93483d855 Don't bother doing latency calculations in the "fast" scheduler.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59461 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-17 19:52:36 +00:00
Dan Gohman
d23e0f81bc Remove the FlaggedNodes member from SUnit. Instead of requiring each SUnit
to carry a SmallVector of flagged nodes, just calculate the flagged nodes
dynamically when they are needed.

The local-liveness change is due to a trivial scheduling change where
the scheduler arbitrary decision differently.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59273 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-13 23:24:17 +00:00
Dan Gohman
550f5afb68 Make the Node member of SUnit private, and add accessors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59264 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-13 21:36:12 +00:00
Dan Gohman
a23b3b803e Change ScheduleDAG's DAG member from a reference to a pointer, to prepare
for the possibility of scheduling without a SelectionDAG being present.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59263 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-13 21:21:28 +00:00
Dan Gohman
9b75b37375 Change the scheduler accessor methods to accept an explicit TargetMachine
argument instead of taking the SelectionDAG's TargetMachine. This is
needed for some upcoming scheduler changes.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@59055 91177308-0d34-0410-b5e6-96231b3b80d8
2008-11-11 17:50:47 +00:00
Dan Gohman
b8cab9227a Fix command-line option printing to print two spaces where needed,
instead of requiring all "short description" strings to begin with
two spaces. This makes these strings less mysterious, and it fixes
some cases where short description strings mistakenly did not
begin with two spaces.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@57521 91177308-0d34-0410-b5e6-96231b3b80d8
2008-10-14 20:25:08 +00:00
Dan Gohman
086ec9976f Replace the LiveRegs SmallSet with a simple counter that keeps
track of the number of live registers, which is all the set was
being used for.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@56498 91177308-0d34-0410-b5e6-96231b3b80d8
2008-09-23 18:50:48 +00:00
Dan Gohman
ee2e403545 Add a new "fast" scheduler. This is currently basically just a
copy of the BURRList scheduler, but with several parts ripped
out, such as backtracking, online topological sort maintenance
(needed by backtracking), the priority queue, and Sethi-Ullman
number computation and maintenance (needed by the priority
queue). As a result of all this, it generates somewhat lower
quality code, but that's its tradeoff for running about 30%
faster than list-burr in -fast mode in many cases.

This is somewhat experimental. Moving forward, major pieces of
this can be refactored with pieces in common with
ScheduleDAGRRList.cpp.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@56307 91177308-0d34-0410-b5e6-96231b3b80d8
2008-09-18 16:26:26 +00:00