Commits · bf428935670c67e7afd3340ee0c6f66240c13658 · Roger Ferrer / llvm-epi-0.8 · GitLab

This project is archived. Its data is read-only.

Aug 06, 2008
- Reinstate this optimization, but without the miscompile. Thanks to Bill for · bf428935
  Nick Lewycky authored Aug 06, 2008
  bf428935
- Just grep for through the LL code instead of the ASM code · 0e966d3e
  Bill Wendling authored Aug 06, 2008
  0e966d3e
- Add default architecture. · bc6786e7
  Bill Wendling authored Aug 05, 2008
  bc6786e7
- Testcase for PR2629. · 3dfa168d
  Bill Wendling authored Aug 05, 2008
  3dfa168d
Aug 05, 2008
- Revert r53282. This was causing a miscompile on Linux. Also, the transformation · ee12a7ae
  Bill Wendling authored Aug 05, 2008
  ee12a7ae
Jul 29, 2008
- Add -unroll-allow-partial command line option that enabled the loop unroller to · 98b5c16e
  Matthijs Kooijman authored Jul 29, 2008
  98b5c16e
- Restructure ArgumentPromotion a bit. Instead of just having a single boolean · fd307045
  Matthijs Kooijman authored Jul 29, 2008
  fd307045
Jul 28, 2008
- Add support for eliminating stores that store the same value that was just loaded. · 3f338974
  Owen Anderson authored Jul 28, 2008
  3f338974
Jul 25, 2008
- Put the LICM of constant GlobalVariables, introduced in r53945, under a · 5f36a32e
  Dan Gohman authored Jul 24, 2008
  5f36a32e
Jul 23, 2008
- "Allow LICM to sink or lift loads from constant memory. Also add a test · 8a8fb908
  Chris Lattner authored Jul 23, 2008
  8a8fb908
- Enable first-class aggregates support. · fa1211f6
  Dan Gohman authored Jul 23, 2008
  fa1211f6
Jul 21, 2008
- Add the PR number to the test. · 60bae3fa
  Dan Gohman authored Jul 21, 2008
  60bae3fa
- Fix a bug in LSR's dead-PHI cleanup. If a PHI has a def-use chain that · 7ad3cd8c
  Dan Gohman authored Jul 21, 2008
  7ad3cd8c
Jul 17, 2008
- Make GlobalOpt preserve address spaces when scalar replacing aggregate globals. · 8b69d77a
  Matthijs Kooijman authored Jul 17, 2008
  8b69d77a
- Fix PR2553 · c600c53d
  Chris Lattner authored Jul 17, 2008
  c600c53d
Jul 16, 2008
- Add a few cases to instcombine's extractvalue testcase. · f22f34f0
  Matthijs Kooijman authored Jul 16, 2008
  f22f34f0
- Un-XFAIL multdeadretval, since instcombine now properly handles the mess... · c9d4e06f
  Matthijs Kooijman authored Jul 16, 2008
  c9d4e06f
- Fix PR2296. Do not transform x86_sse2_storel_dq into a full-width store. · c9709455
  Evan Cheng authored Jul 16, 2008
  c9709455
Jul 15, 2008
- XFAIL the multdeadretval test for now, I will be fixing instcombine to make it work again tomorrow. · af3594cc
  Matthijs Kooijman authored Jul 15, 2008
  af3594cc
- Remove a few tests which no longer hold for deadargelim (since it is now · 121a2062
  Matthijs Kooijman authored Jul 15, 2008
  121a2062
- Add a testcase for the canonicalizations now performed by deadargelim. · d881fb0a
  Matthijs Kooijman authored Jul 15, 2008
  d881fb0a
- Make deadargelim a bit less smart, so it doesn't choke on nested structs as · c1da8744
  Matthijs Kooijman authored Jul 15, 2008
  c1da8744
- Fix typo. · d337446b
  Matthijs Kooijman authored Jul 15, 2008
  d337446b
Jul 14, 2008
- Fix PR2506 by being a bit more careful about reverse fact propagation when · 16395e51
  Chris Lattner authored Jul 14, 2008
  16395e51
- Fix mishandling of the infinite loop case when merging two blocks. This · 80b03a1b
  Chris Lattner authored Jul 13, 2008
  80b03a1b
Jul 12, 2008
- Enhance analysis of srem. · f76aa23b
  Nick Lewycky authored Jul 12, 2008
  f76aa23b
Jul 11, 2008
- Add another optimization from PR2330. Also catch some missing cases that are · f95b64ac
  Nick Lewycky authored Jul 11, 2008
  f95b64ac
- Fix folding of icmp's of i1 where the comparison is signed. The code · 6af608b8
  Chris Lattner authored Jul 11, 2008
  6af608b8
- Fix a bogus optimization: folding (slt (zext i1 A to i32), 1) -> (slt i1 A, true) · 4fa8bb34
  Chris Lattner authored Jul 11, 2008
  4fa8bb34
- make this condition more precise. · 1be09d9e
  Chris Lattner authored Jul 11, 2008
  1be09d9e
Jul 10, 2008
- Restructure dead argument elimination, try #3 :-) · e0f3ab82
  Matthijs Kooijman authored Jul 10, 2008
  e0f3ab82
- Fix overzealous optimization. Thanks to Duncan Sands for pointing out my error! · 6193a564
  Nick Lewycky authored Jul 10, 2008
  6193a564
- Fix a case where vector comparison constant folding would cause an · 67136cff
  Chris Lattner authored Jul 10, 2008
  67136cff
- elementwise comparison of vector constants was completely wrong. Fix · b69689e1
  Chris Lattner authored Jul 10, 2008
  b69689e1
Jul 09, 2008
- Fold (a < 8) && (b < 8) into (a|b) < 8 for unsigned less or greater than. · f9c27c34
  Nick Lewycky authored Jul 09, 2008
  f9c27c34
- Fold ((1 << a) & 1) to (a == 0). · 364661c4
  Nick Lewycky authored Jul 09, 2008
  364661c4
- Fix a broken test. Neither load is eliminable without changing the CFG. · 7212e201
  Chris Lattner authored Jul 09, 2008
  7212e201
- Reduce x - y to -y when we know the 'x' part will get masked off anyways. · 0d3645e6
  Nick Lewycky authored Jul 09, 2008
  0d3645e6
- If loop induction variable's start value is less then its exit value then do not split the loop. · 51cbf928
  Devang Patel authored Jul 09, 2008
  51cbf928
Jul 08, 2008
- 'Optimize' test · 4f1a9ffc
  Chris Lattner authored Jul 08, 2008
  4f1a9ffc