- 4771288 While I'm at it, don't break codegen of mul by 3,5,9. by Chris Lattner · 21 years ago
- 596b97f Generate better code for multiplies by negative constants like -4, -1, -9, etc. by Chris Lattner · 21 years ago
- 8863f18 bug 122: by Reid Spencer · 21 years ago
- 76e2df2 Patches towards fixing PR341 by Chris Lattner · 21 years ago
- d2995df Improve codegen for the LLVM offsetof/sizeof "operator". Before we compiled by Chris Lattner · 21 years ago
- 8b486a1 Fix a regression from r1.224. In particular, codegen a cast from double -> by Chris Lattner · 21 years ago
- 3048373 Move the IntrinsicLowering header into the CodeGen directory, as per PR346 by Chris Lattner · 21 years ago
- 667ea02 Codegen sub C, X a little bit better for register pressure. Instead of by Chris Lattner · 21 years ago
- a6f9fe6 Fold setcc instructions into select and branches that are not in the same BB as by Chris Lattner · 21 years ago
- ccd9796 Do not fold loads into instructions if it is used more than once. In particular by Chris Lattner · 21 years ago
- f70c22b Rename Type::PrimitiveID to TypeId and ::getPrimitiveID() to ::getTypeID() by Chris Lattner · 21 years ago
- 4adf066 Remove support for llvm.isnan. Alkis wins :) by Chris Lattner · 21 years ago
- dc57244 Add basic support for the isunordered intrinsic. The isnan stuff still needs to go by Chris Lattner · 21 years ago
- 01cdb1b By far, one of the most common uses of isnan is to make 'isunordered' by Chris Lattner · 21 years ago
- 0ca2c8e Now that compare instructions aren't lumped in with the other twoargfp instructions, by Chris Lattner · 21 years ago
- b4fe76c Add direct support for the isnan intrinsic, implementing test/Regression/CodeGen/X86/isnan.llx by Chris Lattner · 21 years ago
- 6b5bd58 Fix for PR#366. We use getClassB() so that we can handle cast instructions by John Criswell · 21 years ago
- d029cd2 Convert to the new TargetMachine interface. by Chris Lattner · 21 years ago
- df04097 Add some notes to myself, no functional changes by Chris Lattner · 21 years ago
- 9f088e4 Generate branch machine instructions with MachineBasicBlock operands instead of by Brian Gaeke · 21 years ago
- b7cb9ff Two more improvements for null pointer handling: storing a null pointer by Chris Lattner · 21 years ago
- 9f1b531 Second half of my fixed-sized-alloca patch. This folds the LEA to compute by Chris Lattner · 21 years ago
- cb2fd55 Substantially improve code generation for address exposed locals (aka fixed by Chris Lattner · 21 years ago
- 2b10b08 Pass boolean constants into function calls more efficiently, generating: by Chris Lattner · 21 years ago
- c81e6ba Fix a fairly serious pessimizaion that was preventing us from efficiently by Chris Lattner · 21 years ago
- 9984fd0 Fix some comments, avoid sign extending booleans when zero extend works fine by Chris Lattner · 21 years ago
- 96e3b42 Generate more efficient code for casting booleans to integers (no sign extension required) by Chris Lattner · 21 years ago
- e7a31c9 Codegen floating point stores of constants into integer instructions. This by Chris Lattner · 21 years ago
- 260195d Make comparisons against the null pointer as efficient as integer comparisons by Chris Lattner · 21 years ago
- bbc130d Remove unneeded check by Chris Lattner · 21 years ago
- c8af02c Improve signed division by power of 2 *dramatically* from this: by Chris Lattner · 21 years ago
- 9eb9b8e Improve code generated for integer multiplications by 2,3,5,9 by Chris Lattner · 21 years ago
- 7799363 Remove unused #include by Chris Lattner · 21 years ago
- 1afe773 Make RequiresFPRegKill() take a MachineBasicBlock arg. by Brian Gaeke · 21 years ago
- 235aa5e In InsertFPRegKills(), use the machine-CFG itself rather than the by Brian Gaeke · 21 years ago
- ea9ca67 Update the machine-CFG edges whenever we see a branch. by Brian Gaeke · 21 years ago
- 53b54be Remove code to adjust the iterator for llvm.readio and llvm.writeio. by John Criswell · 21 years ago
- e5a4c15 Added support for the llvm.readio and llvm.writeio intrinsics. by John Criswell · 21 years ago
- 82c5a99 Implement a small optimization, which papers over the problem in by Chris Lattner · 21 years ago
- 87e18de Emit the immediate form of in/out when possible. by Chris Lattner · 21 years ago
- 133dbb1 Fix issues that the local allocator has dealing with instructions that implicitly use ST(0) by Chris Lattner · 21 years ago
- 8d2822e Use the fucomi[p] instructions to perform floating point comparisons instead by Chris Lattner · 21 years ago
- 9938286 Fix a bug in my load/cast folding patch. by Chris Lattner · 21 years ago
- 13c07fe Adjust some comments, fix a bug in my previous patch by Chris Lattner · 21 years ago
- feac3e1 On X86, casting an integer to floating point requires going through memory. by Chris Lattner · 21 years ago
- 95157f7 Implement folding of loads into floating point operations. This implements: by Chris Lattner · 21 years ago
- 6621ed9 Unify all of the code for floating point +,-,*,/ into one function by Chris Lattner · 21 years ago
- 8ebf1c3 This implements folding of constant operands into floating point operations by Chris Lattner · 21 years ago
- 462fa82 Restructure the mul/div/rem handling code to follow the pattern the other by Chris Lattner · 21 years ago
- 48b0c97 Codegen FP adds and subtracts with a constant more efficiently, generating: by Chris Lattner · 21 years ago
- 427aeb4 Two changes: by Chris Lattner · 21 years ago
- 85aa709 Silence a spurious warning by Chris Lattner · 21 years ago
- 6d804f4 Reversed the order of the llvm.writeport() operands so that the value by John Criswell · 21 years ago
- aee0cf3 Changed assertions to error messages. by John Criswell · 21 years ago
- ca6ea0f Changes recommended by Chris: by John Criswell · 21 years ago
- 4ffff9e Added the llvm.readport and llvm.writeport intrinsics for x86. These do by John Criswell · 21 years ago
- 7b92de1e Fix PR313: [x86] JIT miscompiles unsigned short to floating point by Chris Lattner · 21 years ago
- 48c937e Fix a minor bug in previous checking by Chris Lattner · 21 years ago
- e80e637 Improve codegen of long == and != comparisons against constants. Before, by Chris Lattner · 21 years ago
- 6ab06d5 Handle various other important cases of multiplying a long constant immediate. For by Chris Lattner · 21 years ago
- 028adc4 Efficiently handle a long multiplication by a constant. For this testcase: by Chris Lattner · 21 years ago
- 722070e Improve code generation of long shifts by 32. On this testcase: by Chris Lattner · 21 years ago
- 0652167 Bugfixes: inc/dec don't set the carry flag! by Chris Lattner · 21 years ago
- 92900a6 Improve code for passing constant longs as arguments to function calls. by Chris Lattner · 21 years ago
- 33f7fa3 Emit more efficient 64-bit operations when the RHS is a constant, and one by Chris Lattner · 21 years ago
- 7ba9230 Fix typeo by Chris Lattner · 21 years ago
- ab1d0e0 Add support for simple immediate handling to long instruction selection. by Chris Lattner · 21 years ago
- edd5e49 Implement negation of longs efficiently. For this testcase: by Chris Lattner · 21 years ago
- 502e36c Minor tweak to avoid an extra reg-reg copy that the register allocator has to eliminate by Chris Lattner · 21 years ago
- 29bf062 Two changes: by Chris Lattner · 21 years ago
- 28977af Support getelementptr instructions which use uint's to index into structure by Chris Lattner · 21 years ago
- bee8a09 Clean up code a bit. by Alkis Evlogimenos · 21 years ago
- 13ce339 Fix type in instruction builder instantiation by Alkis Evlogimenos · 21 years ago
- 68626c2 Generate slightly smaller code, "test R, R" instead of "cmp R, 0" by Chris Lattner · 21 years ago
- 352eb48 Codegen FP select instructions into X86 conditional moves. Annoyingly enough by Chris Lattner · 21 years ago
- 307ecba Fold comparisons into select instructions, making much better code and by Chris Lattner · 21 years ago
- 12d96a0 Add direct support for integer select instructions, though we still don't support by Chris Lattner · 21 years ago
- 6f2ab04 Fix a fairly major performance problem. If a PHI node had a constant as by Chris Lattner · 21 years ago
- ab18020 Malloc doesn't kill a load. This patch need not go into 1.2 though. by Chris Lattner · 21 years ago
- 85c84e7 Fix a really nasty bug that was breaking ijpeg in LLC mode. We were incorrectly by Chris Lattner · 21 years ago
- 5634b9f It helps if I save the file. :) by Chris Lattner · 21 years ago
- 317201d Rename the intrinsic enum values for llvm.va_* from Intrinsic::va_* to by Chris Lattner · 21 years ago
- 7dee5da Implement folding explicit load instructions into binary operations. For a by Chris Lattner · 21 years ago
- 721d2d4 Rearrange and refactor some code. No functionality changes. by Chris Lattner · 21 years ago
- 538607f Doxygenify some comments. by Misha Brukman · 21 years ago
- 2158522 Handle passing constant integers to functions much more efficiently. Instead by Chris Lattner · 21 years ago
- ce6096f Fix a minor code-quality issue. When passing 8 and 16-bit integer constants by Chris Lattner · 21 years ago
- 8295f20 A big X86 instruction rename. The instructions are renamed to make by Alkis Evlogimenos · 21 years ago
- ee35285 Eliminate the X86-specific BMI functions, using BuildMI instead. by Chris Lattner · 21 years ago
- 168aa90 Fix a miscompilation of 197.parser that occurs when you have single basic by Chris Lattner · 21 years ago
- 1ddf475 These two virtual methods are never called. by Chris Lattner · 21 years ago
- da474ad SHLD and SHRD take 32-bit operands but an 8-bit immediate. Rename them by Alkis Evlogimenos · 21 years ago
- 8e475b8 Floating point loads/stores act on memory operands. Rename them to by Alkis Evlogimenos · 21 years ago
- e35ba65 Rename SHL, SHR, SAR, SHLD and SHLR instructions to make them by Alkis Evlogimenos · 21 years ago
- 71e353e Uncomment assertions that register# != 0 on calls to by Alkis Evlogimenos · 21 years ago
- 8dd8d26 Fix some warnings, some of which were spurious, and some of which were real by Chris Lattner · 21 years ago
- 5f2c7b1 Teach the instruction selector how to transform 'array' GEP computations into X86 by Chris Lattner · 21 years ago
- b6bac51 * Make the previous patch more efficient by not allocating a temporary MachineInstr by Chris Lattner · 21 years ago
- 985fe3d add an inefficient way of folding structure and constant array indexes together by Chris Lattner · 21 years ago
- 5a83096 Implement special case for storing an immediate into memory so that we don't need by Chris Lattner · 21 years ago