a flag for now.
First off, thanks to Daniel Jasper for really pointing out the issue
here. It's been here forever (at least, I think it was there when
I first wrote this code) without getting really noticed or fixed.
The key problem is what happens when two reasonably common patterns
happen at the same time: we outline multiple cold regions of code, and
those regions in turn have diamonds or other CFGs for which we can't
just topologically lay them out. Consider some C code that looks like:
if (a1()) { if (b1()) c1(); else d1(); f1(); }
if (a2()) { if (b2()) c2(); else d2(); f2(); }
done();
Now consider the case where a1() and a2() are unlikely to be true. In
that case, we might lay out the first part of the function like:
a1, a2, done;
And then we will be out of successors in which to build the chain. We go
to find the best block to continue the chain with, which is perfectly
reasonable here, and find "b1" let's say. Laying out successors gets us
to:
a1, a2, done; b1, c1;
At this point, we will refuse to lay out the successor to c1 (f1)
because there are still un-placed predecessors of f1 and we want to try
to preserve the CFG structure. So we go get the next best block, d1.
... wait for it ...
Except that the next best block *isn't* d1. It is b2! d1 is waaay down
inside these conditionals. It is much less important than b2. Except
that this is exactly what we didn't want. If we keep going we get the
entire set of the rest of the CFG *interleaved*!!!
a1, a2, done; b1, c1; b2, c2; d1, f1; d2, f2;
So we clearly need a better strategy here. =] My current favorite
strategy is to actually try to place the block whose predecessor is
closest. This very simply ensures that we unwind these kinds of CFGs the
way that is natural and fitting, and should minimize the number of cache
lines instructions are spread across.
It also happens to be *dead simple*. It's like the datastructure was
specifically set up for this use case or something. We only push blocks
onto the work list when the last predecessor for them is placed into the
chain. So the back of the worklist *is* the nearest next block.
Unfortunately, a change like this is going to cause *soooo* many
benchmarks to swing wildly. So for now I'm adding this under a flag so
that we and others can validate that this is fixing the problems
described, that it seems possible to enable, and hopefully that it fixes
more of our problems long term.
llvm-svn: 231238
In a CFG with the edges A->B->C and A->C, B is an optional branch.
LLVM's default behavior is to lay the blocks out naturally, i.e. A, B,
C, in order to improve code locality and fallthroughs. However, if a
function contains many of those optional branches only a few of which
are taken, this leads to a lot of unnecessary icache misses. Moving B
out of line can work around this.
Review: http://reviews.llvm.org/D7719
llvm-svn: 231230
No functional changes intended.
(I plan on doing some modifications to this function and would like to
have as few unrelated changes as possible in the patch)
llvm-svn: 229649
Canonicalize access to function attributes to use the simpler API.
getAttributes().getAttribute(AttributeSet::FunctionIndex, Kind)
=> getFnAttribute(Kind)
getAttributes().hasAttribute(AttributeSet::FunctionIndex, Kind)
=> hasFnAttribute(Kind)
Also, add `Function::getFnStackAlignment()`, and canonicalize:
getAttributes().getStackAlignment(AttributeSet::FunctionIndex)
=> getFnStackAlignment()
llvm-svn: 229208
Some benchmarks have shown that this could lead to a potential
performance benefit, and so adding some flags to try to help measure the
difference.
A possible explanation. In diamond-shaped CFGs (A followed by either
B or C both followed by D), putting B and C both in between A and
D leads to the code being less dense than it could be. Always either
B or C have to be skipped increasing the chance of cache misses etc.
Moving either B or C to after D might be beneficial on average.
In the long run, but we should probably do a better job of analyzing the
basic block and branch probabilities to move the correct one of B or
C to after D. But even if we don't use this in the long run, it is
a good baseline for benchmarking.
Original patch authored by Daniel Jasper with test tweaks and a second
flag added by me.
Differential Revision: http://reviews.llvm.org/D6969
llvm-svn: 226034
The existing code provided for specifying a global loop alignment preference.
However, the preferred loop alignment might depend on the loop itself. For
recent POWER cores, loops between 5 and 8 instructions should have 32-byte
alignment (while the others are better with 16-byte alignment) so that the
entire loop will fit in one i-cache line.
To support this, getPrefLoopAlignment has been made virtual, and can be
provided with an optional MachineLoop* so the target can inspect the loop
before answering the query. The default behavior, as before, is to return the
value set with setPrefLoopAlignment. MachineBlockPlacement now queries the
target for each loop instead of only once per function. There should be no
functional change for other targets.
llvm-svn: 225117
This is to be consistent with StringSet and ultimately with the standard
library's associative container insert function.
This lead to updating SmallSet::insert to return pair<iterator, bool>,
and then to update SmallPtrSet::insert to return pair<iterator, bool>,
and then to update all the existing users of those functions...
llvm-svn: 222334
shorter/easier and have the DAG use that to do the same lookup. This
can be used in the future for TargetMachine based caching lookups from
the MachineFunction easily.
Update the MIPS subtarget switching machinery to update this pointer
at the same time it runs.
llvm-svn: 214838
string_ostream is a safe and efficient string builder that combines opaque
stack storage with a built-in ostream interface.
small_string_ostream<bytes> additionally permits an explicit stack storage size
other than the default 128 bytes to be provided. Beyond that, storage is
transferred to the heap.
This convenient class can be used in most places an
std::string+raw_string_ostream pair or SmallString<>+raw_svector_ostream pair
would previously have been used, in order to guarantee consistent access
without byte truncation.
The patch also converts much of LLVM to use the new facility. These changes
include several probable bug fixes for truncated output, a programming error
that's no longer possible with the new interface.
llvm-svn: 211749
define below all header includes in the lib/CodeGen/... tree. While the
current modules implementation doesn't check for this kind of ODR
violation yet, it is likely to grow support for it in the future. It
also removes one layer of macro pollution across all the included
headers.
Other sub-trees will follow.
llvm-svn: 206837
The motivation is to mark dump methods as used in debug builds so that they can
be called from lldb, but to not do so in release builds so that they can be
dead-stripped.
There's lots of potential follow-up work suggested in the thread
"Should dump methods be LLVM_ATTRIBUTE_USED only in debug builds?" on cfe-dev,
but everyone seems to agreen on this subset.
Macro name chosen by fair coin toss.
llvm-svn: 198456
We now only allow breaking source order if the exit block frequency is
significantly higher than the other exit block. The actual bias is
currently under a flag so the best cut-off can be found; the flag
defaults to the old behavior. The idea is to get some benchmark coverage
over different values for the flag and pick the best one.
When we require the new frequency to be at least 20% higher than the old
frequency I see a 5% speedup on zlib's deflate when compressing a random
file on x86_64/westmere. Hal reported a small speedup on Fhourstones on
a BG/Q and no regressions in the test suite.
The test case is the full long_match function from zlib's deflate. I was
reluctant to add it for previous tweaks to branch probabilities because
it's large and potentially fragile, but changed my mind since it's an
important use case and more likely to break with all the current work
going into the PGO infrastructure.
Differential Revision: http://llvm-reviews.chandlerc.com/D2202
llvm-svn: 195265
(4.58s vs 3.2s on an oldish Mac Tower).
The corresponding src is excerpted bellow. The lopp accounts for about 90% of execution time.
--------------------
cat -n test-suite/MultiSource/Benchmarks/Olden/em3d/make_graph.c
90
91 for (k=0; k<j; k++)
92 if (other_node == cur_node->to_nodes[k]) break;
The defective layout is sketched bellow, where the two branches need to swap.
------------------------------------------------------------------------
L:
...
if (cond) goto out-of-loop
goto L
While this code sequence is defective, I don't understand why it incurs 1/3 of
execution time. CPU-event-profiling indicates the poor laoyout dose not increase
in br-misprediction; it dosen't increase stall cycle at all, and it dosen't
prevent the CPU detect the loop (i.e. Loop-Stream-Detector seems to be working fine
as well)...
The root cause of the problem is that the layout pass calls AnalyzeBranch()
with basic-block which is not updated to reflect its current layout.
rdar://13966341
llvm-svn: 183174
When debugging performance regressions we often ask ourselves if the regression
that we see is due to poor isel/sched/ra or due to some micro-architetural
problem. When comparing two code sequences one good way to rule out front-end
bottlenecks (and other the issues) is to force code alignment. This pass adds
a flag that forces the alignment of all of the basic blocks in the program.
llvm-svn: 179353
This fixes some of the cycles between libCodeGen and libSelectionDAG. It's still
a complete mess but as long as the edges consist of virtual call it doesn't
cause breakage. BasicTTI did static calls and thus broke some build
configurations.
llvm-svn: 172246
directly.
This is in preparation for removing the use of the 'Attribute' class as a
collection of attributes. That will shift to the AttributeSet class instead.
llvm-svn: 171253
Sooooo many of these had incorrect or strange main module includes.
I have manually inspected all of these, and fixed the main module
include to be the nearest plausible thing I could find. If you own or
care about any of these source files, I encourage you to take some time
and check that these edits were sensible. I can't have broken anything
(I strictly added headers, and reordered them, never removed), but they
may not be the headers you'd really like to identify as containing the
API being implemented.
Many forward declarations and missing includes were added to a header
files to allow them to parse cleanly when included first. The main
module rule does in fact have its merits. =]
llvm-svn: 169131
We use the enums to query whether an Attributes object has that attribute. The
opaque layer is responsible for knowing where that specific attribute is stored.
llvm-svn: 165488
Previously, MBP essentially aligned every branch target it could. This
bloats code quite a bit, especially non-looping code which has no real
reason to prefer aligned branch targets so heavily.
As Andy said in review, it's still a bit odd to do this without a real
cost model, but this at least has much more plausible heuristics.
Fixes PR13265.
llvm-svn: 161409
We branch to the successor with higher edge weight first.
Convert from
je LBB4_8 --> to outer loop
jmp LBB4_14 --> to inner loop
to
jne LBB4_14
jmp LBB4_8
PR12750
rdar: 11393714
llvm-svn: 161018
rotation. When there is a loop backedge which is an unconditional
branch, we will end up with a branch somewhere no matter what. Try
placing this backedge in a fallthrough position above the loop header as
that will definitely remove at least one branch from the loop iteration,
where whole loop rotation may not.
I haven't seen any benchmarks where this is important but loop-blocks.ll
tests for it, and so this will be covered when I flip the default.
llvm-svn: 154812
laid out in a form with a fallthrough into the header and a fallthrough
out of the bottom. In that case, leave the loop alone because any
rotation will introduce unnecessary branches. If either side looks like
it will require an explicit branch, then the rotation won't add any, do
it to ensure the branch occurs outside of the loop (if possible) and
maximize the benefit of the fallthrough in the bottom.
llvm-svn: 154806
This is a complex change that resulted from a great deal of
experimentation with several different benchmarks. The one which proved
the most useful is included as a test case, but I don't know that it
captures all of the relevant changes, as I didn't have specific
regression tests for each, they were more the result of reasoning about
what the old algorithm would possibly do wrong. I'm also failing at the
moment to craft more targeted regression tests for these changes, if
anyone has ideas, it would be welcome.
The first big thing broken with the old algorithm is the idea that we
can take a basic block which has a loop-exiting successor and a looping
successor and use the looping successor as the layout top in order to
get that particular block to be the bottom of the loop after layout.
This happens to work in many cases, but not in all.
The second big thing broken was that we didn't try to select the exit
which fell into the nearest enclosing loop (to which we exit at all). As
a consequence, even if the rotation worked perfectly, it would result in
one of two bad layouts. Either the bottom of the loop would get
fallthrough, skipping across a nearer enclosing loop and thereby making
it discontiguous, or it would be forced to take an explicit jump over
the nearest enclosing loop to earch its successor. The point of the
rotation is to get fallthrough, so we need it to fallthrough to the
nearest loop it can.
The fix to the first issue is to actually layout the loop from the loop
header, and then rotate the loop such that the correct exiting edge can
be a fallthrough edge. This is actually much easier than I anticipated
because we can handle all the hard parts of finding a viable rotation
before we do the layout. We just store that, and then rotate after
layout is finished. No inner loops get split across the post-rotation
backedge because we check for them when selecting the rotation.
That fix exposed a latent problem with our exitting block selection --
we should allow the backedge to point into the middle of some inner-loop
chain as there is no real penalty to it, the whole point is that it
*won't* be a fallthrough edge. This may have blocked the rotation at all
in some cases, I have no idea and no test case as I've never seen it in
practice, it was just noticed by inspection.
Finally, all of these fixes, and studying the loops they produce,
highlighted another problem: in rotating loops like this, we sometimes
fail to align the destination of these backwards jumping edges. Fix this
by actually walking the backwards edges rather than relying on loopinfo.
This fixes regressions on heapsort if block placement is enabled as well
as lots of other cases where the previous logic would introduce an
abundance of unnecessary branches into the execution.
llvm-svn: 154783
the loop header has a non-loop predecessor which has been pre-fused into
its chain due to unanalyzable branches. In this case, rotating the
header into the body of the loop in order to place a loop exit at the
bottom of the loop is a Very Bad Idea as it makes the loop
non-contiguous.
I'm working on a good test case for this, but it's a bit annoynig to
craft. I should get one shortly, but I'm submitting this now so I can
begin the (lengthy) performance analysis process. An initial run of LNT
looks really, really good, but there is too much noise there for me to
trust it much.
llvm-svn: 154395
where a chain outside of the loop block-set ended up in the worklist for
scheduling as part of the contiguous loop. However, asserting the first
block in the chain is in the loop-set isn't a valid check -- we may be
forced to drag a chain into the worklist due to one block in the chain
being part of the loop even though the first block is *not* in the loop.
This occurs when we have been forced to form a chain early due to
un-analyzable branches.
No test case here as I have no idea how to even begin reducing one, and
it will be hopelessly fragile. We have to somehow end up with a loop
header of an inner loop which is a successor of a basic block with an
unanalyzable pair of branch instructions. Ow. Self-host triggers it so
it is unlikely it will regress.
This at least gets block placement back to passing selfhost and the test
suite. There are still a lot of slowdown that I don't like coming out of
block placement, although there are now also a lot of speedups. =[ I'm
seeing swings in both directions up to 10%. I'm going to try to find
time to dig into this and see if we can turn this on for 3.1 as it does
a really good job of cleaning up after some loops that degraded with the
inliner changes.
llvm-svn: 154287
Moving toward a uniform style of pass definition to allow easier target configuration.
Globally declare Pass ID.
Globally declare pass initializer.
Use INITIALIZE_PASS consistently.
Add a call to the initializer from CodeGen.cpp.
Remove redundant "createPass" functions and "getPassName" methods.
While cleaning up declarations, cleaned up comments (sorry for large diff).
llvm-svn: 150100
