| Age | Commit message (Collapse) | Author |
|---|
|
Now that we updated to upstream cpplint, a lot of these NOLINTs are no
longer necessary.
Bug: 68951293
Change-Id: If8ed5ffe89727f313f907a214b6d8fd2a2eddbad
|
|
Cleanup errors from upstream cpplint in preparation
for moving art's cpplint fork to upstream tip-of-tree cpplint.
Test: cd art && mm
Bug: 68951293
Change-Id: I15faed4594cbcb8399850f8bdee39d42c0c5b956
|
|
Passes using local ArenaAllocator were hiding their memory
usage from the allocation counting, making it difficult to
track down where memory was used. Using ScopedArenaAllocator
reveals the memory usage.
This changes the HGraph constructor which requires a lot of
changes in tests. Refactor these tests to limit the amount
of work needed the next time we change that constructor.
Test: m test-art-host-gtest
Test: testrunner.py --host
Test: Build with kArenaAllocatorCountAllocations = true.
Bug: 64312607
Change-Id: I34939e4086b500d6e827ff3ef2211d1a421ac91a
|
|
Replace most uses of the runtime's Primitive in compiler
with a new class DataType. This prepares for introducing
new types, such as Uint8, that the runtime does not need
to know about.
Test: m test-art-host-gtest
Test: testrunner.py --host
Bug: 23964345
Change-Id: Iec2ad82454eec678fffcd8279a9746b90feb9b0c
|
|
Add explicit field initialization to default value where necessary.
Also clean up interpreter intrinsics header.
Test: m
Change-Id: I7a850ac30dcccfb523a5569fb8400b9ac892c8e5
|
|
Rationale:
Lots of code should not depend on int only. This CL generalizes
the kinds of types that can be optimized after analysis. As part
of the CL, however, a minor cleanup regarding type safety of the
stored induction var analysis results is required. This further
improved our int benchmark, and brings the long benchmark up-to-par.
Test: m test-art-host-run-test
Change-Id: I5dfb623dabf9113de90c2f6da99328dda8f8b60b
|
|
Rationale:
Based on some self-imposed "blind" testing, improved
the induction variable analysis for typical cases
that provide a bit more elaborate HIR.
Test: test-art-host
Change-Id: I6e6bbf99928c29973178fa48f3942b14bf069944
|
|
Rationale:
Information on polynomial sequences is nice to further enhance
BCE and last-value assignment. In this case, this CL enables more
loop optimizations for benchpress' Sum (80 x speedup). Also
changed rem-based geometric induction to wrap-around induction.
Test: test-art-host
Change-Id: Ie4d2659edefb814edda2c971c1f70ba400c31111
|
|
Rationale:
Information on geometric and polynomial (coming soon) sequences
are nice to have to further enhance BCE and last-value assignment.
Test: test-art-host
Change-Id: Ib5e2998c3eb1009def6fd00b82935da7c3ba7c6e
|
|
Add abstraction for uint16_t type index.
Test: m test-art-host
Change-Id: I47708741c7c579cbbe59ab723c1e31c5fe71f83a
|
|
Rationale:
Rather than half-baked reconstructing cycles during loop optimizations,
this CL passes the SCC computed during induction variable analysis
to the loop optimizer (trading some memory for more optimizations).
This further improves CaffeineLogic from 6000us down to 4200us (dx)
and 2200us to 1690us (jack). Note that this is on top of prior
improvements in previous CLs. Also, some narrowing type concerns
are taken care of during transfer operations.
Test: test-art-host
Change-Id: Ice2764811a70073c5014b3a05fb51f39fd2f4c3c
|
|
Rationale:
Similar to the previous CL, this helps to eliminate more dead induction.
Now, CaffeineLogic, when compiled with dx (rather than jack) improves
by a 1.5 speedup (9000us -> 6000us).
Note:
We need to run the simplifier before induction analysis to trigger
the select simplification first. Although a bit of a compile-time hit,
it seems a good idea to run a simplifier here again anyway.
Test: test-art-host
Change-Id: I93b91ca40a4d64385c64393028e8d213f0c904a8
|
|
Rationale:
This is a commonly used construct (e.g. x = !x for booleans
and x ^= 1 for integers). This CL prepares some upcoming
optimizations that exploit such inductions.
Change-Id: I46edffb9de1075a836995daf5c2dfff7891f3034
Test: 530-checker-loops2 and induction_var_analysis_test
|
|
Rationale:
Few extra comments, added replacement of loop control functionality.
All in preparation of bigger loop optimizations CL.
Change-Id: Ie2243c3c8b57b2901654b05fb200e9d80a0cb641
Test: test-art
|
|
Extended unit tests.
Rationale:
Now that inductions are typed, comparisons should take
the type into account as well.
Change-Id: Ie533bb728d2006390646657573fa743e6fc37ff2
|
|
Rationale:
This changelist implements allowing narrowing conversions within
inductions and loop control. More induction and loops recognized,
more bounds eliminated. We all win. The basic idea is pretty simple
(record type with detected induction) but one has to get all the
details right, as illustrated by the many new unit tests.
BUG=27151098
Change-Id: I254020bfa5fa623799b31bbbb5ccc97d4d5a0100
|
|
Change-Id: I6f7785abece7ead31b4737fc6b54712c26582e2b
|
|
First step towards merging the two passes, which will later result in
HGraphBuilder directly producing SSA form. This CL mostly just updates
tests broken by not being able to inspect the pre-SSA form.
Using HLocals outside the HGraphBuilder is now deprecated.
Bug: 27150508
Change-Id: I00fb6050580f409dcc5aa5b5aa3a536d6e8d759e
|
|
Just like aget(-wide), the value operand of aput(-wide) bytecode
instructions can be both int/long and float/double. This patch builds
on the previous mechanism for resolving type of ArrayGets to type the
values of ArraySets based on the reference type of the array.
Bug: 22538329
Change-Id: Ic86abbb58de146692de04476b555010b6fcdd8b6
|
|
This reverts commit 68289a531484d26214e09f1eadd9833531a3bc3c.
Now uses Primitive::Is64BitType instead of Primitive::ComponentSize
because it was incorrectly optimized by GCC.
Bug: 26208284
Bug: 24252151
Bug: 24252100
Bug: 22538329
Bug: 25786318
Change-Id: Ib39f3da2b92bc5be5d76f4240a77567d82c6bebe
|
|
This reverts commit d9510dfc32349eeb4f2145c801f7ba1d5bccfb12.
Bug: 26208284
Bug: 24252151
Bug: 24252100
Bug: 22538329
Bug: 25786318
Change-Id: I5f491becdf076ff51d437d490405ec4e1586c010
|
|
This patch refactors the SsaBuilder to do the following:
1) All phis are constructed live and marked dead if not used or proved
to be conflicting.
2) Primitive type propagation, now not a separate pass, identifies
conflicting types and marks corresponding phis dead.
3) When compiling --debuggable, DeadPhiHandling used to revive phis
which had only environmental uses but did not attempt to resolve
conflicts. This pass was removed as obsolete and is now superseded
by primitive type propagation (identifying conflicting phis) and
SsaDeadPhiEliminiation (keeping phis live if debuggable + env use).
4) Resolving conflicts requires correct primitive type information
on all instructions. This was not the case for ArrayGet instructions
which can have ambiguous types in the bytecode. To this end,
SsaBuilder now runs reference type propagation and types ArrayGets
from the type of the input array.
5) With RTP being run inside the SsaBuilder, it is not necessary to
run it as a separate optimization pass. Optimizations can now assume
that all instructions of type kPrimNot have reference type info after
SsaBuilder (with the exception of NullConstant).
6) Graph now contains a reference type to be assigned to NullConstant.
All reference type instructions therefore have RTI, as now enforced
by the SsaChecker.
Bug: 24252151
Bug: 24252100
Bug: 22538329
Bug: 25786318
Change-Id: I7a3aee1ff66c82d64b4846611c547af17e91d260
|
|
A critical edge would not be split if the predecessor ends with
TryBoundary. This would eventually trip liveness analysis because
a back edge block would have smaller liveness position than a nested
loop.
Another implication of this change is that an edge between a loop's
pre-header ending with TryBoundary and the header will be split,
guaranteeing that a pre-header always has just one successor.
Bug: 25493695
Bug: 25454012
Change-Id: I5a13b8bb74509b48f5d628906f7158af007f99ae
|
|
Rationale: for loops that may not be taken, this taken-test
can be used by clients of the induction variable
analysis to ensure trip-count evaluation is valid.
Change-Id: Ia64749e2389b7224e69d6a49bb604b1964c11068
|
|
|
|
Rationale: used by dynamic BCE (done in another CL).
Change-Id: Ia6ce75da57b5298fba74622822ae0bae69c74188
|
|
This reverts commit 89c0d32437011bbe492fe14c766cd707046ce043.
Change-Id: I603a49794e155cc97410b8836c8ea425bfdc98eb
|
|
This reverts commit 2c1ffc3a06e9ed0411e29e7dc2558b5d657ede7a.
Change-Id: I3291070c373e661fa578f5a38becbb5a502baf94
|
|
This helps inlining and type check elimination.
e.g:
void foo(ArrayList a) {
int size = a.size(); // this can be inlined now.
}
Change-Id: I3ffeaa79d9df444aa19511c83c544cb5f9d9ab20
|
|
As shown in the induction analysis presentation, trip-counts need to
deal with potential taken/not-taken situations (so that trip-count
is either valid in the full loop or just in the loop-body proper)
and potential finite/infinite situations (the latter can still be
analyzed but may need to run-time test later to guard against the
infinite conditions). This CL provides that information.
Change-Id: I0445d8e836b80a3614af217ce3e39d766e77b986
|
|
Rationale:
(1) Analyze multi-way phis (requested by Nicolas, Igor, and Mingyao).
(2) Analyze trip count for restricted != loops
(3) Added unit test for public API of range analysis (static methods
were already well-tested).
Change-Id: I9285d22d3bb927f141204cc4697ea6fe5120994d
|
|
Rationale: by computing an upper bound on the trip count of each
loop after induction var analysis has completed, a
simple range analysis yields lower and upper bounds on
all induced expressions in a loop; this analysis
plugs directly into BCE (follow-up CL).
Change-Id: I46a3fe48721ca372547199b39a3498c47992597d
|
|
Rationale:
Saves some memory for nodes that are not really required, and
yields slightly more readable debugging strings.
Change-Id: I95b64b48869699137b5d49e26eb20091e264de7a
|
|
Various improvements:
(1) Introduced period sequences.
(2) Extended all transfer functions to deal with all cases;
also refactored these to read more compactly.
(3) Improved debugging output for constants for readability.
(4) Used direct pointer in mappings for clarify.
(5) Several induction info "constructors" for readability.
(6) Various other changes suggested in earlier code reviews.
Change-Id: I9d5381f1676b63d30cea6f5e304d4b7bda7acb96
|
|
Rationale:
Induction variable analysis forms the basis of a wide
variety of compiler optimizations. This implementation
finds induction variables using the elegant SSA-based
algorithm defined by [Gerlek et al.].
Change-Id: I79b8dce33ffb8b283c179699a8dff5bd196f75b2
|
