llvm-svn: 19945

llvm-svn: 19944

LLVM make the very reasonable assumption that constant expressions will
have at least one operand! :)

llvm-svn: 19943

llvm-svn: 19942

llvm-svn: 19941

llvm-svn: 19940

* Change the FunctionCalls and AuxFunctionCalls vectors into std::lists.
  This makes many operations on these lists much more natural, and avoids
  *exteremely* expensive copying of DSCallSites (e.g. moving nodes around
  between lists, erasing a node from not the end of the vector, etc).

With a profile build of analyze, this speeds up BU DS from 25.14s to
12.59s on 176.gcc.  I expect that it would help TD even more, but I don't
have data for it.

This effectively eliminates removeIdenticalCalls and children from the
profile, going from 6.53 to 0.27s.

llvm-svn: 19939

added fp extend and removed a forgotten assert in more than 6 arg support (should break somewhere else now :) ) and fix an incorrect asm sequence for indirect calls

llvm-svn: 19938

llvm-svn: 19937

llvm-svn: 19934

llvm-svn: 19933

llvm-svn: 19932

llvm-svn: 19931

llvm-svn: 19930

the same.

llvm-svn: 19929

llvm-svn: 19928

llvm-svn: 19926

Based on the ilist changes avoid allocating an entire Use object for the
end of the Use chain.  This saves 8 bytes of memory for each Value allocated
in the program.  For 176.gcc, this reduces us from 69.5M -> 66.0M, a 5.3%
memory savings.

llvm-svn: 19925

llvm-svn: 19924

llvm-svn: 19923

llvm-svn: 19922

Add a new method, destroySentinal, that is used to delete it (instead of
requiring use of delete.

llvm-svn: 19921

llvm-svn: 19920

llvm-svn: 19919

cases it represented them as 'unsigned's, which are not enough for 64-bit
hosts.  In other cases, it represented them as uint64_t's, which are
inefficient for 32-bit hosts.

This patch unifies all of the sizes to use size_t instead.

llvm-svn: 19918

This file was schizophrenic when it came to representing sizes.  In some
cases it represented them as 'unsigneds', which are not enough for 64-bit
hosts.  In other cases, it represented them as uint64_t's, which are
inefficient for 32-bit hosts.

This patch unifies all of the sizes to use size_t instead.

llvm-svn: 19917

really need.  This reduces 4M of memory consumption reading 176.gcc.

llvm-svn: 19916

llvm-svn: 19915

first step towards a correct and complete stack.  also add some forms for things that were getting stuck in the nightly tester.

llvm-svn: 19914

used here.

llvm-svn: 19913

llvm-svn: 19912

BB as the load are not included in the Cand* sets at all.

llvm-svn: 19911

not invalidated on entry and on exit of the block.  This fixes some N^2
behavior in common cases, and speeds up gcc another 5% to 22.35s.

llvm-svn: 19910

llvm-svn: 19909

llvm-svn: 19908

This speeds up 176.gcc from 25.73s to 23.48s, which is 9.5%

llvm-svn: 19907

will cause us to miss cases where the input pointer to a load could be value
numbered to another load.  Something like this:


  %X = load int* %P1
  %Y = load int* %P2

Those are obviously the same if P1/P2 are the same.  The code this patch
removes attempts to handle that.  However, since GCSE iterates, this doesn't
actually buy us anything: GCSE will first replace P1 or P2 with the other
one, then the load can be value numbered as equal.

Removing this code speeds up gcse a lot.  On 176.gcc in debug mode, this
speeds up gcse from 29.08s -> 25.73s, a 13% savings.

llvm-svn: 19906

  %A = alloca int
  %V = load int* %A

value number %V to undef, not 0.

llvm-svn: 19905

memuse at the end, thus it is signed.

llvm-svn: 19904

llvm-svn: 19903