Commits · 93bcc7a6d2d5011547d933b8f0b8da61f4f6cca1 · Roger Ferrer / llvm-epi-0.8 · GitLab

This is an archived project. Repository and other project resources are read-only.

Mar 05, 2008
- Fix a coalescer bug wrt how dead copy interval is shortened. · 29b502e0
  Evan Cheng authored Mar 05, 2008
  29b502e0
- Clarify that CALLSEQ_START..END may not be nested, · 8ee39c61
  Dale Johannesen authored Mar 05, 2008
  8ee39c61
- Generalize FP constant shrinking optimization to apply to any vt · 78e9cab2
  Chris Lattner authored Mar 05, 2008
  78e9cab2
- Improve comment, pass in the original VT so that we can shrink a long double constant · 3dc38990
  Chris Lattner authored Mar 05, 2008
  3dc38990
- Codegen support for i128 UINT_TO_FP. This just fixes a · da7897c4
  Dan Gohman authored Mar 05, 2008
  da7897c4
- Add a target lowering hook to control whether it's worthwhile to compress fp constant. · 0a62cb44
  Evan Cheng authored Mar 05, 2008
  0a62cb44
- 64bit CAS on 32bit x86. · 357061a7
  Andrew Lenharth authored Mar 05, 2008
  357061a7
- Codegen support for i128 SINT_TO_FP. · d9d874b0
  Dan Gohman authored Mar 05, 2008
  d9d874b0
- Refactor code. Remove duplicated functions that basically do the same thing as · 63254466
  Evan Cheng authored Mar 05, 2008
  63254466
Mar 04, 2008
- Some improvements related to the computation of heights, depths of SUnits. · c62c2bb4
  Roman Levenstein authored Mar 04, 2008
  c62c2bb4
- Refactor ExpandConstantFP so it can optimize load from constpool of types... · 38caf774
  Evan Cheng authored Mar 04, 2008
  38caf774
- Did I say 'e = getNumOperands()'? I meant --e, of course. · 2ae70788
  Bill Wendling authored Mar 04, 2008
  2ae70788
- Rename isOperand() to isOperandOf() (and other similar methods). It always confuses me. · 567d2e5b
  Evan Cheng authored Mar 04, 2008
  567d2e5b
- Miscellaneous clean-ups based on Evan's feedback: · 0e541ea7
  Bill Wendling authored Mar 03, 2008
  0e541ea7
- Misc. APInt-ification in the DAGCombiner. · e1c4f995
  Dan Gohman authored Mar 03, 2008
  e1c4f995
- More APInt-ification. · 10f34077
  Dan Gohman authored Mar 03, 2008
  10f34077
Mar 03, 2008
- Yet more APInt-ification. · 0e238dc8
  Dan Gohman authored Mar 03, 2008
  0e238dc8
- More APInt-ification. · 2fa65b79
  Dan Gohman authored Mar 03, 2008
  2fa65b79
- More APInt-ification. · f2bbfa3b
  Dan Gohman authored Mar 03, 2008
  f2bbfa3b
- Go through the machine instruction's operands to make sure that we're not · 7921ad0d
  Bill Wendling authored Mar 03, 2008
  7921ad0d
- Make the register scavenger update the bookkeeping values for sub/super · 528083bc
  Bill Wendling authored Mar 03, 2008
  528083bc
- Multiple instructions can be inserted when eliminating frame indexes. We need · 4836d58f
  Bill Wendling authored Mar 03, 2008
  4836d58f
Mar 01, 2008
- all but CAS working on x86 · d032c333
  Andrew Lenharth authored Mar 01, 2008
  d032c333
- Add MVT::is128BitVector and is64BitVector. Shrink · 208cc8f1
  Dale Johannesen authored Mar 01, 2008
  208cc8f1
- Refactor / clean up code; remove td list scheduler special tie breaker (no real benefit). · 73bdf043
  Evan Cheng authored Mar 01, 2008
  73bdf043
Feb 29, 2008
- Don't fill eh frames even though these are text sections. · 26edb59d
  Evan Cheng authored Feb 29, 2008
  26edb59d
- If we reload a virtual register that's already been assigned, we want to mark · 811153a5
  Bill Wendling authored Feb 29, 2008
  811153a5
- Fix PR2112: don't run loop aligner if target doesn't have a TargetLowering object. · 2e26dc80
  Evan Cheng authored Feb 29, 2008
  2e26dc80
- No need for coalescer to update kills. Only copies are coalesced and those... · ca7c61e7
  Evan Cheng authored Feb 29, 2008
  ca7c61e7
- Remove redundant #include. · 88f83994
  Evan Cheng authored Feb 29, 2008
  88f83994
- More APInt-ification. · bd2fa566
  Dan Gohman authored Feb 29, 2008
  bd2fa566
- Use the new convertFromAPInt instead of convertFromZeroExtendedInteger, · 837a6dcc
  Dan Gohman authored Feb 29, 2008
  837a6dcc
- Use the new APInt-enabled form of getConstant instead of converting · ec6be4a7
  Dan Gohman authored Feb 29, 2008
  ec6be4a7
- Added option -align-loops=<true/false> to disable loop aligner pass. · 95a7be47
  Evan Cheng authored Feb 28, 2008
  95a7be47
Feb 28, 2008
- Interface of getByValTypeAlignment differed between · cbde4c22
  Dale Johannesen authored Feb 28, 2008
  cbde4c22
- Fix an assertion message. · c4c3de2b
  Dale Johannesen authored Feb 28, 2008
  c4c3de2b
- Keep track how many commutes are performed by the scheduler. · a465bfb8
  Evan Cheng authored Feb 28, 2008
  a465bfb8
- implement expand for ISD::DECLARE by just deleting it. · 9824ffef
  Chris Lattner authored Feb 28, 2008
  9824ffef
- Add a quick and dirty "loop aligner pass". x86 uses it to align its loops to 16-byte boundaries. · c799065c
  Evan Cheng authored Feb 28, 2008
  c799065c
Feb 27, 2008
- Handle load/store of misaligned vectors that are the · bf76a08e
  Dale Johannesen authored Feb 27, 2008
  bf76a08e