| 2006-05-05 |
Chris Lattner | Teach the code generator to use cvtss2sd as extload... |
blob | commitdiff | raw |
| 2006-05-03 |
Evan Cheng | Use movsd to shuffle in the lowest two elements of... |
blob | commitdiff | raw | diff to current |
| 2006-04-25 |
Evan Cheng | Fix a typo. |
blob | commitdiff | raw | diff to current |
| 2006-04-25 |
Evan Cheng | Explicitly specify result type for def : Pat<> patterns... |
blob | commitdiff | raw | diff to current |
| 2006-04-24 |
Evan Cheng | Added X86 SSE2 intrinsics which can be represented... |
blob | commitdiff | raw | diff to current |
| 2006-04-24 |
Evan Cheng | Some missing movlps, movhps, movlpd, and movhpd patterns. |
blob | commitdiff | raw | diff to current |
| 2006-04-21 |
Evan Cheng | Now generating perfect (I think) code for "vector set... |
blob | commitdiff | raw | diff to current |
| 2006-04-19 |
Evan Cheng | Prefer {p}unpack* and mov*dup over {p}shuf* as well. |
blob | commitdiff | raw | diff to current |
| 2006-04-19 |
Evan Cheng | - Renamed AddedCost to AddedComplexity. |
blob | commitdiff | raw | diff to current |
| 2006-04-19 |
Evan Cheng | More mov{h|l}p{d|s} patterns. |
blob | commitdiff | raw | diff to current |
| 2006-04-19 |
Evan Cheng | - More mov{h|l}ps patterns. |
blob | commitdiff | raw | diff to current |
| 2006-04-18 |
Evan Cheng | - PEXTRW cannot take a memory location as its first... |
blob | commitdiff | raw | diff to current |
| 2006-04-18 |
Evan Cheng | Name change for clarity sake |
blob | commitdiff | raw | diff to current |
| 2006-04-18 |
Evan Cheng | Name change for clarity sake |
blob | commitdiff | raw | diff to current |
| 2006-04-18 |
Evan Cheng | Left a pattern out |
blob | commitdiff | raw | diff to current |
| 2006-04-18 |
Evan Cheng | Fixed an encoding bug: movd from XMM to R32. |
blob | commitdiff | raw | diff to current |
| 2006-04-17 |
Evan Cheng | Use movss to insert_vector_elt(v, s, 0). |
blob | commitdiff | raw | diff to current |
| 2006-04-17 |
Evan Cheng | Encoding bug |
blob | commitdiff | raw | diff to current |
| 2006-04-17 |
Evan Cheng | Errors in patterns preventing load folding |
blob | commitdiff | raw | diff to current |
| 2006-04-16 |
Evan Cheng | movduprm, movshduprm bugs |
blob | commitdiff | raw | diff to current |
| 2006-04-16 |
Evan Cheng | Encoding bugs |
blob | commitdiff | raw | diff to current |
| 2006-04-15 |
Evan Cheng | More encoding bugs |
blob | commitdiff | raw | diff to current |
| 2006-04-15 |
Evan Cheng | pslldrm, psrawrm, etc. encoding bug |
blob | commitdiff | raw | diff to current |
| 2006-04-15 |
Evan Cheng | hsubp{s|d} encoding bug |
blob | commitdiff | raw | diff to current |
| 2006-04-15 |
Evan Cheng | Silly bug |
blob | commitdiff | raw | diff to current |
| 2006-04-14 |
Evan Cheng | Some clean up |
blob | commitdiff | raw | diff to current |
| 2006-04-14 |
Evan Cheng | Last few SSE3 intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-14 |
Evan Cheng | Misc. SSE2 intrinsics: clflush, lfench, mfence |
blob | commitdiff | raw | diff to current |
| 2006-04-14 |
Evan Cheng | pcmpeq* and pcmpgt* intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-14 |
Evan Cheng | psll*, psrl*, and psra* intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-13 |
Evan Cheng | Doh. PANDrm, etc. are not commutable. |
blob | commitdiff | raw | diff to current |
| 2006-04-13 |
Evan Cheng | psad, pmax, pmin intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-13 |
Evan Cheng | Various SSE2 packed integer intrinsics: pmulhuw, pavgw... |
blob | commitdiff | raw | diff to current |
| 2006-04-13 |
Evan Cheng | padds{b|w}, paddus{b|w}, psubs{b|w}, psubus{b|w} intrin... |
blob | commitdiff | raw | diff to current |
| 2006-04-13 |
Evan Cheng | Naming inconsistency. |
blob | commitdiff | raw | diff to current |
| 2006-04-12 |
Evan Cheng | SSE / SSE2 conversion intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-12 |
Evan Cheng | All "integer" logical ops (pand, por, pxor) are now... |
blob | commitdiff | raw | diff to current |
| 2006-04-12 |
Evan Cheng | Promote v4i32, v8i16, v16i8 load to v2i64 load. |
blob | commitdiff | raw | diff to current |
| 2006-04-12 |
Evan Cheng | Various SSE2 conversion intrinsics |
blob | commitdiff | raw | diff to current |
| 2006-04-11 |
Evan Cheng | Added __builtin_ia32_storelv4si, __builtin_ia32_movqv4si, |
blob | commitdiff | raw | diff to current |
| 2006-04-11 |
Evan Cheng | gcc lower SSE prefetch into generic prefetch intrinsic... |
blob | commitdiff | raw | diff to current |
| 2006-04-11 |
Evan Cheng | Misc. intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-11 |
Evan Cheng | movnt* and maskmovdqu intrinsics |
blob | commitdiff | raw | diff to current |
| 2006-04-11 |
Evan Cheng | Added support for _mm_move_ss and _mm_move_sd. |
blob | commitdiff | raw | diff to current |
| 2006-04-10 |
Evan Cheng | Remove some bogus patterns; clean up. |
blob | commitdiff | raw | diff to current |
| 2006-04-10 |
Evan Cheng | Added some missing shuffle patterns. |
blob | commitdiff | raw | diff to current |
| 2006-04-10 |
Evan Cheng | movups / movupd |
blob | commitdiff | raw | diff to current |
| 2006-04-10 |
Evan Cheng | Conditional move of vector types. |
blob | commitdiff | raw | diff to current |
| 2006-04-08 |
Evan Cheng | ldmxcsr and stmxcsr. |
blob | commitdiff | raw | diff to current |
| 2006-04-07 |
Evan Cheng | Added patterns for MOVHPSmr and MOVLPSmr. |
blob | commitdiff | raw | diff to current |
| 2006-04-06 |
Evan Cheng | A MOVPS2SSmr, i.e. _mm_store_ss, encoding bug. |
blob | commitdiff | raw | diff to current |
| 2006-04-06 |
Evan Cheng | - movlp{s|d} and movhp{s|d} support. |
blob | commitdiff | raw | diff to current |
| 2006-04-06 |
Evan Cheng | POR encoded as PAND, yikes. |
blob | commitdiff | raw | diff to current |
| 2006-04-05 |
Evan Cheng | Support for comi / ucomi intrinsics. |
blob | commitdiff | raw | diff to current |
| 2006-04-05 |
Evan Cheng | Handle canonical form of e.g. |
blob | commitdiff | raw | diff to current |
| 2006-04-04 |
Evan Cheng | Added pslldq and psrldq. |
blob | commitdiff | raw | diff to current |
| 2006-04-04 |
Evan Cheng | Minor fixes + naming changes. |
blob | commitdiff | raw | diff to current |
| 2006-04-04 |
Evan Cheng | PSHUF* encoding bugs. |
blob | commitdiff | raw | diff to current |
| 2006-04-04 |
Evan Cheng | cmpps / cmppd encoding bug |
blob | commitdiff | raw | diff to current |
| 2006-04-04 |
Evan Cheng | Compact some intrinsic definitions. |
blob | commitdiff | raw | diff to current |
| 2006-04-03 |
Evan Cheng | Some SSE1 intrinsics: min, max, sqrt, etc. |
blob | commitdiff | raw | diff to current |
| 2006-04-03 |
Evan Cheng | Use movlpd to: store lower f64 extracted from v2f64. |
blob | commitdiff | raw | diff to current |
| 2006-04-03 |
Evan Cheng | - More efficient extract_vector_elt with shuffle and... |
blob | commitdiff | raw | diff to current |
| 2006-03-31 |
Evan Cheng | Use a X86 target specific node X86ISD::PINSRW instead... |
blob | commitdiff | raw | diff to current |
| 2006-03-31 |
Evan Cheng | Added support for SSE3 horizontal ops: haddp{s|d} and... |
blob | commitdiff | raw | diff to current |
| 2006-03-31 |
Evan Cheng | Add support to use pextrw and pinsrw to extract and... |
blob | commitdiff | raw | diff to current |
| 2006-03-30 |
Evan Cheng | Make sure all possible shuffles are matched. |
blob | commitdiff | raw | diff to current |
| 2006-03-30 |
Evan Cheng | More logical ops patterns |
blob | commitdiff | raw | diff to current |
| 2006-03-30 |
Evan Cheng | Add support for _mm_cmp{cc}_ss and _mm_cmp{cc}_ps intri... |
blob | commitdiff | raw | diff to current |
| 2006-03-30 |
Evan Cheng | Add 128-bit pmovmskb intrinsic support. |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | Change SSE pack operation definitions to fit what the... |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | - Added some SSE2 128-bit packed integer ops. |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | Need to special case splat after all. Make the second... |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | Floating point logical operation patterns should match... |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | - More shuffle related bug fixes. |
blob | commitdiff | raw | diff to current |
| 2006-03-29 |
Evan Cheng | - Only use pshufd for v4i32 vector shuffles. |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | Added aliases to scalar SSE instructions, e.g. addss... |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | movlps and movlpd should be modeled as two address... |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | Typo |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | * Prefer using operation of matching types. e.g unpcklp... |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | - Clean up / consoladate various shuffle masks. |
blob | commitdiff | raw | diff to current |
| 2006-03-28 |
Evan Cheng | Model unpack lower and interleave as vector_shuffle... |
blob | commitdiff | raw | diff to current |
| 2006-03-27 |
Chris Lattner | unbreak the build |
blob | commitdiff | raw | diff to current |
| 2006-03-27 |
Evan Cheng | Use pcmpeq to generate vector of all ones. |
blob | commitdiff | raw | diff to current |
| 2006-03-26 |
Evan Cheng | Remove X86:isZeroVector, use ISD::isBuildVectorAllZeros... |
blob | commitdiff | raw | diff to current |
| 2006-03-25 |
Evan Cheng | Build arbitrary vector with more than 2 distinct scalar... |
blob | commitdiff | raw | diff to current |
| 2006-03-25 |
Evan Cheng | Added SSE cachebility ops |
blob | commitdiff | raw | diff to current |
| 2006-03-25 |
Evan Cheng | Added 128-bit packed integer subtraction. |
blob | commitdiff | raw | diff to current |
| 2006-03-25 |
Evan Cheng | Added CVTSS2SI. |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Evan Cheng | Support for scalar to vector with zero extension. |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Evan Cheng | Added LDMXCSR |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Chris Lattner | plug the intrinsics into the patterns for movmsk* |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Evan Cheng | Handle BUILD_VECTOR with all zero elements. |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Evan Cheng | More efficient v2f64 shuffle using movlhps, movhlps... |
blob | commitdiff | raw | diff to current |
| 2006-03-24 |
Evan Cheng | Handle more shuffle cases with SHUFP* instructions. |
blob | commitdiff | raw | diff to current |
| 2006-03-23 |
Evan Cheng | Following icc's lead: use movdqa to load / store 128... |
blob | commitdiff | raw | diff to current |
| 2006-03-23 |
Evan Cheng | Add v4i32 <-> v4f32 bitconvert patterns. |
blob | commitdiff | raw | diff to current |
| 2006-03-23 |
Evan Cheng | Add 128-bit integer vector load and add (for testing). |
blob | commitdiff | raw | diff to current |
| 2006-03-22 |
Evan Cheng | SHUFP* are two address code. |
blob | commitdiff | raw | diff to current |
| 2006-03-22 |
Evan Cheng | - Supposely movlhps is faster / better than unpcklpd. |
blob | commitdiff | raw | diff to current |
| next |
projects / oota-llvm.git / history