From 2e4842efdd3a40d9e9f9ae07962513329ab3c184 Mon Sep 17 00:00:00 2001 From: Chris Lattner Date: Sun, 12 Oct 2008 19:47:48 +0000 Subject: [PATCH] add some more notes llvm-svn: 57423 --- docs/ReleaseNotes.html | 43 +++++++++++++++++++++++++++++++++++------- 1 file changed, 36 insertions(+), 7 deletions(-) diff --git a/docs/ReleaseNotes.html b/docs/ReleaseNotes.html index c9027e6bf3c..2355b382cc6 100644 --- a/docs/ReleaseNotes.html +++ b/docs/ReleaseNotes.html @@ -24,8 +24,6 @@

Written by the LLVM Team

- -

Introduction @@ -70,7 +68,14 @@ It includes a large number of features and refinements from LLVM 2.3.

Machine LICM Machine Sinking LegalizeDAGTypes + llc -enable-value-prop, propagation of value info (sign/zero ext info) from + one MBB to another --> + +
@@ -79,7 +84,11 @@ It includes a large number of features and refinements from LLVM 2.3.

-

....

+

The LLVM IR generated by llvm-gcc now doesn't name instructions. Use the + instnamer pass if you want them.

+ + +

LLVM API Changes:

@@ -94,6 +103,8 @@ converted to strings using llvm::GetConstantStringInfo defined via "llvm/Analysis/ValueTracking.h".
  • API change: BinaryOperator::create -> Create (CmpInst, CastInst too)
  • +
  • Various header files like "llvm/ADT/iterator" were given a .h suffix. + Change your code to #include "llvm/ADT/iterator.h" instead.
  • @@ -182,14 +193,14 @@ this section. is gone.

    -
  • fast isel, -O0 compile times

  • +
  • -O0 compile times overall much faster

  • Attrs changes?

  • Initial PIC16 port

  • -
  • builtin sync_compare_and_swap builtins + intrinsics

  • +
  • Support the rest of the atomic __sync builtins

  • ...

  • @@ -230,7 +241,7 @@ for the C, C++, Objective-C, Ada, and Fortran front-ends.

  • use diet patch landed: saved 15% IR memory footprint
  • LLVM IR now directly represents "common" linkage, instead of representing it as a form of weak linkage.
  • - +
  • DebugInfoBuilder
  • ...
  • @@ -248,6 +259,10 @@ LLVM 2.4 optimizers support a few major enhancements:

      +
    • GVN now does local PRE?
    • + +
    • Matthijs' Dead argument elimination rewrite
    • +
    • Old-ADCE used control dependence and deleted output-free infinite loops. Added a new Loop deletion pass (for deleting output free provably-finite loops) and rewrote ADCE to be simpler faster, and not need control dependence.
    • @@ -258,6 +273,10 @@ and rewrote ADCE to be simpler faster, and not need control dependence.
    • Tail duplication was is removed from the standard optimizer sequence.
    • +
    • Various helper functions (ComputeMaskedBits, ComputeNumSignBits, etc) were +pulled out of instcombine and put into a new ValueTracking.h file, where they +can be reused by other passes.
    • +
    @@ -274,7 +293,15 @@ which allows us to implement more aggressive algorithms and make it run faster:

    @@ -292,7 +319,8 @@ faster:

    @@ -326,6 +354,7 @@ faster: