FP_REG_KILL instructions at the end of blocks involved with critical edges.
Fix a bug where FP_REG_KILL instructions weren't inserted in fall through
unconditional branches. Perhaps this will fix some linscan problems?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@11019 91177308-0d34-0410-b5e6-96231b3b80d8
the globals directly. This doesn't save any substantial time, however, because the
globals graph only contains globals!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10997 91177308-0d34-0410-b5e6-96231b3b80d8
function to find the globals, iterate over all of the globals directly. This
speeds the function up from 14s to 6.3s on perlbmk, reducing DSA time from
53->46s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10996 91177308-0d34-0410-b5e6-96231b3b80d8
This reduces the number of nodes allocated, then immediately merged and DNE'd
from 2193852 to 1298049. unfortunately this only speeds DSA up by ~1.5s (of
53s), because it's spending most of its time waddling through the scalar map :(
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10992 91177308-0d34-0410-b5e6-96231b3b80d8
Also, use RC::merge when possible, reducing the number of nodes allocated, then immediately merged away from 2985444 to 2193852 on perlbmk.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10991 91177308-0d34-0410-b5e6-96231b3b80d8
it to be off. If it looks like it's completely unnecessary after testing, I
will remove it completely (which is the hope).
* Callers of the DSNode "copy ctor" can not choose to not copy links.
* Make node collapsing not create a garbage node in some cases, avoiding a
memory allocation, and a subsequent DNE.
* When merging types, allow two functions of different types to be merged
without collapsing.
* Use DSNodeHandle::isNull more often instead of DSNodeHandle::getNode() == 0,
as it is much more efficient.
*** Implement the new, more efficient reachability cloner class
In addition to only cloning nodes that are reachable from interesting
roots, this also fixes the huge inefficiency we had where we cloned lots
of nodes, only to merge them away immediately after they were cloned.
Now we only actually allocate a node if there isn't one to merge it into.
* Eliminate the now-obsolete cloneReachable* and clonePartiallyInto methods
* Rewrite updateFromGlobalsGraph to use the reachability cloner
* Rewrite mergeInGraph to use the reachability cloner
* Disable the scalar map scanning code in removeTriviallyDeadNodes. In large
SCC's, this is extremely expensive. We need a better data structure for the
scalar map, because we really want to scan the unique node handles, not ALL
of the scalars.
* Remove the incorrect SANER_CODE_FOR_CHECKING_IF_ALL_REFERRERS_ARE_FROM_SCALARMAP code.
* Move the code for eliminating integer nodes from the trivially dead
eliminator to the dead node eliminator.
* removeDeadNodes no longer uses removeTriviallyDeadNodes, as it contains a
superset of the node removal power.
* Only futz around with the globals graph in removeDeadNodes if it is modified
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10987 91177308-0d34-0410-b5e6-96231b3b80d8
efficient in the case where a function calls into the same graph multiple times
(ie, it either contains multiple calls to the same function, or multiple calls
to functions in the same SCC graph)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10986 91177308-0d34-0410-b5e6-96231b3b80d8
when joining we need to check if we overlap with the second interval
or any of its aliases.
Also make joining intervals the default.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10973 91177308-0d34-0410-b5e6-96231b3b80d8
out that the problem was actually the writer writing out a 'null' value
because it didn't normalize it. This fixes:
test/Regression/Assembler/2004-01-22-FloatNormalization.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10967 91177308-0d34-0410-b5e6-96231b3b80d8
is a move between two registers, at least one of the registers is
virtual and the two live intervals do not overlap.
This results in about 40% reduction in intervals, 30% decrease in the
register allocators running time and a 20% increase in peephole
optimizations (mainly move eliminations).
The option can be enabled by passing -join-liveintervals where
appropriate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10965 91177308-0d34-0410-b5e6-96231b3b80d8
virtReg lives on the stack. Now a virtual register has an entry in the
virtual->physical map or the virtual->stack slot map but never in
both.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10958 91177308-0d34-0410-b5e6-96231b3b80d8
map was only used to implement a marginal GlobalsGraph optimization, and it
actually slows the analysis down (due to the overhead of keeping it), so just
eliminate it entirely.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10955 91177308-0d34-0410-b5e6-96231b3b80d8
in terms of it.
Though clonePartiallyInto is not cloning partial graphs yet, this change
dramatically speeds up inlining of graphs with many scalars. For example,
this change speeds up the BU pass on 253.perlbmk from 69s to 36s, because
it avoids iteration over the scalar map, which can get pretty large.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10951 91177308-0d34-0410-b5e6-96231b3b80d8
fact "profitable" to do so. This makes compactification "free" for small
programs (ie, it is completely disabled) and even helps large programs by
not having to encode pointless compactification planes.
On 176.gcc, this saves 50K from the bytecode file, which is, alas only
a couple percent.
This concludes my head bashing against the bytecode format, at least for
now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10922 91177308-0d34-0410-b5e6-96231b3b80d8
This shrinks the bytecode file for 176.gcc by about 200K (10%), and 254.gap by
about 167K, a 25% reduction. There is still a lot of room for improvement in
the encoding of the compaction table.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10915 91177308-0d34-0410-b5e6-96231b3b80d8
This shrinks the bytecode file for 176.gcc by about 200K (10%), and 254.gap by
about 167K, a 25% reduction. There is still a lot of room for improvement in
the encoding of the compaction table.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10914 91177308-0d34-0410-b5e6-96231b3b80d8
the bytecode file for 176.gcc by about 200K (10%), and 254.gap by about 167K,
a 25% reduction. There is still a lot of room for improvement in the encoding
of the compaction table.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10913 91177308-0d34-0410-b5e6-96231b3b80d8
Fix some problem cases where I was building the slot calculator in bytecode
writer mode instead of asmwriter mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10911 91177308-0d34-0410-b5e6-96231b3b80d8
type planes. This saves about 5k on 176.gcc, and is needed for a subsequent
patch of mine I'm working on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10908 91177308-0d34-0410-b5e6-96231b3b80d8
bytecode files when compiling 176.gcc, but more importantly will make it
easier to eliminate CPR's in the future (no new .bc revision will be
required to support them)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10884 91177308-0d34-0410-b5e6-96231b3b80d8
of forcing them to go through ConstantPointerRef's. This allows bytecode
files to mirror .ll files, allows more efficient encoding, and makes it easier
to eventually eliminate CPR's.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10883 91177308-0d34-0410-b5e6-96231b3b80d8
returning error codes. Because they don't return an error code, they can
return the value read, which simplifies the code and makes the reader more
efficient (yaay!).
Also eliminate the special case code for little endian machines.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10871 91177308-0d34-0410-b5e6-96231b3b80d8
intended to save size (and does on small programs), but on big programs it
actually increases the size of the program slightly. The deal is that many
functions end up using the characters that the string contained, and the
characters are no longer in the global constant table, so they have to be
emitted in function specific constant pools.
This pessimization will be fixed in subsequent patches.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10864 91177308-0d34-0410-b5e6-96231b3b80d8
It's not clear why the code was looking for signed chars < 0, but it can't
matter to the assembler anyway, so the check goes away. This also fixes
compatibility with arrays of [us]byte that have constantexprs in them.
Also slightly restructure some code to be cleaner.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10854 91177308-0d34-0410-b5e6-96231b3b80d8
It's not clear why the code was looking for signed chars < 0, but it can't
matter to the assembler anyway, so the check goes away.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10853 91177308-0d34-0410-b5e6-96231b3b80d8
because that makes it abort. Also, fix a typo in a comment.
This checkin brought to you by the "It only takes about 30 seconds to run
ENABLE_LLI tests on Shootout on zion, even if they all dump core" fund.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10844 91177308-0d34-0410-b5e6-96231b3b80d8
dominates the normal destination, not the exceptional dest (ie, the result
of a call is undefined on an exception)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10841 91177308-0d34-0410-b5e6-96231b3b80d8
Since this really only makes sense for these two, change hte instance variable
to reflect whether we are writing a bytecode file or not. This makes it
reasonable to add bcwriter specific stuff to it as necessary.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10837 91177308-0d34-0410-b5e6-96231b3b80d8
LiveIntervals::Interval::liveAt. Both were considering the live ranges
closed in the end, when they are actually open.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10835 91177308-0d34-0410-b5e6-96231b3b80d8
Using the SlotCalculator is total overkill for this file, a simple map
will suffice. Why doesn't this use the NameMangler interface?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10823 91177308-0d34-0410-b5e6-96231b3b80d8
when an implicitely defined register is later used by an alias. For example:
call foo
%reg1024 = mov %AL
The call implicitely defines EAX but only AL is used. Before this fix
no information was available on AL. Now EAX and all its aliases except
AL get defined and die at the call instruction whereas AL lives to be
killed by the assignment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10813 91177308-0d34-0410-b5e6-96231b3b80d8
testcase test/Regression/Assembler/ConstantExprFold.llx
Note that these kinds of things only rarely show up in source code, but are
exceedingly common in the intermediate stages of algorithms like SCCP. By
folding things (especially relational operators) that use symbolic constants,
we are able to speculatively fold more conditional branches, which can
lead to some big simplifications.
It would be easy to add a lot more special cases here, so if you notice
SCCP missing anything "obvious", you know what to make smarter. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10812 91177308-0d34-0410-b5e6-96231b3b80d8
Move a bunch of (now) private stuff from ConstantFolding.h into
ConstantFolding.cpp.
This _finally_ gets us to a place where we have a sane constant folder. The
rules are:
1. LLVM clients now use ConstantExpr::get* methods to fold constants. If they
cannot be folded, a constantexpr is created, so these methods always return
valid Constant*'s.
2. The implementation of ConstantExpr::get* uses the functions exposed by
ConstantFolding.h to try to fold constants. If they cannot be folded,
they should return a null pointer.
3. The implementation of ConstantFolding can do whatever it wants, and only
has one client (Constants.cpp)
This cuts down on the wierd dependencies, and eliminates the two interfaces.
The old constanthandling interface was especially bad for clients to use
because almost none of them took the failure condition into consideration,
thus leading to obscure problems.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10807 91177308-0d34-0410-b5e6-96231b3b80d8
this whole refactoring: allow constant folding methods to return something
other than predefined classes, allow them to return generic Constant*'s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10806 91177308-0d34-0410-b5e6-96231b3b80d8
constants as being "true" when evaluating branches. This was introduced
because we now create constantexprs for the constants instead of failing the
fold.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10778 91177308-0d34-0410-b5e6-96231b3b80d8
* Implement SCCP of load instructions, implementing Transforms/SCCP/loadtest.ll
This allows us to fold expressions like "foo"[2], even if the pointer is only
a conditional constant.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10767 91177308-0d34-0410-b5e6-96231b3b80d8
LiveVariables::HandlePhysRegDef private they use information that is
not in memory when LiveVariables finishes the analysis.
Also update the TwoAddressInstructionPass to not use this interface.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10755 91177308-0d34-0410-b5e6-96231b3b80d8
The first change (which is disabled) compactifies all of the function constant
pools into the global constant pool, in an attempt to reduce the amount of
duplication and overhead. Unfortunately, as the comment indicates, this is
not yet a win, so it is disabled.
The second change sorts the typeid's so that those types that can be used
by instructions in the program appear earlier in the table than those that
cannot (such as structures and arrays). This causes the instructions to
be able to use the dense encoding more often, saving about 5K on 254.gap.
This is only a .65% savings though, unfortunately. :(
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10754 91177308-0d34-0410-b5e6-96231b3b80d8
Fix iterator invalidation problems which was causing -mstrip to miss some
entries, and read free'd memory. This shrinks the symbol table of 254.gap
from 333 to 284 bytes! :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10751 91177308-0d34-0410-b5e6-96231b3b80d8
occurs when the symbol table for a module has been stripped, making all of the
function local symbols go away.
This saves 6728 bytes in the stripped bytecode file of 254.gap (which obviously
has 841 functions), which isn't a ton, but helps and was easy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10750 91177308-0d34-0410-b5e6-96231b3b80d8
* Refactor reader stuff out of include/llvm/Bytecode/Primitives.h. This is
internal implementation details for the reader, not public interfaces!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10739 91177308-0d34-0410-b5e6-96231b3b80d8
This should get hunked over to the Sparc backend, along with
MachineCodeForInstruction and a bunch of files in include/llvm/Codegen,
but those battles will have to wait for a later time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10731 91177308-0d34-0410-b5e6-96231b3b80d8
of the register allocator as follows:
before after
mesa 2.3790 1.5994
vpr 2.6008 1.2078
gcc 1.9840 0.5273
mcf 0.2569 0.0470
eon 1.8468 1.4359
twolf 0.9475 0.2004
burg 1.6807 1.3300
lambda 1.2191 0.3764
Speedups range anyware from 30% to over 400% :-)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@10712 91177308-0d34-0410-b5e6-96231b3b80d8
