2009-05-28 |
Evan Cheng | Fix MOVMSKPDrr encoding. |
blob | commitdiff | raw |
2009-05-28 |
Evan Cheng | Fix PSIGND encoding bug. Patch by Sean Callanan. |
blob | commitdiff | raw | diff to current |
2009-05-28 |
Bill Wendling | "The instructions MMX_PSADBWrm and MMX_PSADBWrr have... |
blob | commitdiff | raw | diff to current |
2009-05-27 |
Evan Cheng | Fix sfence jit encoding. Patch by Sean Callanan. |
blob | commitdiff | raw | diff to current |
2009-05-12 |
Evan Cheng | 80 col violations. |
blob | commitdiff | raw | diff to current |
2009-04-29 |
Nate Begeman | Fix infinite recursion in the C++ code which handles... |
blob | commitdiff | raw | diff to current |
2009-04-27 |
Nate Begeman | 2nd attempt, fixing SSE4.1 issues and implementing... |
blob | commitdiff | raw | diff to current |
2009-04-24 |
Rafael Espindola | Revert 69952. Causes testsuite failures on linux x86-64. |
blob | commitdiff | raw | diff to current |
2009-04-24 |
Nate Begeman | PR2957 |
blob | commitdiff | raw | diff to current |
2009-04-08 |
Rafael Espindola | Re-apply 68552. |
blob | commitdiff | raw | diff to current |
2009-04-07 |
Bill Wendling | Temporarily revert r68552. This was causing a failure... |
blob | commitdiff | raw | diff to current |
2009-04-07 |
Rafael Espindola | Reduce code duplication on the TLS implementation. |
blob | commitdiff | raw | diff to current |
2009-02-26 |
Evan Cheng | ADDS{D|S}rr_Int and MULS{D|S}rr_Int are not commutable... |
blob | commitdiff | raw | diff to current |
2009-02-23 |
Nate Begeman | Generate better code for v8i16 shuffles on SSE2 |
blob | commitdiff | raw | diff to current |
2009-02-10 |
Evan Cheng | Handle llvm.x86.sse2.maskmov.dqu in 64-bit. |
blob | commitdiff | raw | diff to current |
2009-02-05 |
Evan Cheng | A few more isAsCheapAsAMove. |
blob | commitdiff | raw | diff to current |
2009-01-28 |
Evan Cheng | The memory alignment requirement on some of the mov... |
blob | commitdiff | raw | diff to current |
2009-01-09 |
Dan Gohman | Whitespace and other minor adjustments to make SSE... |
blob | commitdiff | raw | diff to current |
2008-12-18 |
Mon P Wang | Fixed x86 code generation of multiple for v2i64. It... |
blob | commitdiff | raw | diff to current |
2008-12-03 |
Dan Gohman | Rename isSimpleLoad to canFoldAsLoad, to better reflect... |
blob | commitdiff | raw | diff to current |
2008-12-03 |
Dan Gohman | Mark x86's V_SET0 and V_SETALLONES with isSimpleLoad... |
blob | commitdiff | raw | diff to current |
2008-10-17 |
Evan Cheng | Fix lfence and mfence encoding. These look like MRM5r... |
blob | commitdiff | raw | diff to current |
2008-10-16 |
Dan Gohman | Fix the predicate for memop64 to be a regular load... |
blob | commitdiff | raw | diff to current |
2008-10-15 |
Dan Gohman | Now that predicates can be composed, simplify several of |
blob | commitdiff | raw | diff to current |
2008-10-10 |
Dale Johannesen | Fix SSE4.1 roundss, roundsd. While the instructions... |
blob | commitdiff | raw | diff to current |
2008-10-07 |
Anders Carlsson | Certain patterns involving the "movss" instruction... |
blob | commitdiff | raw | diff to current |
2008-10-02 |
Bill Wendling | "The original bug was a complaint that _mm_srli_si128... |
blob | commitdiff | raw | diff to current |
2008-09-26 |
Evan Cheng | Implement "punpckldq %xmm0, $xmm0" as "pshufd $0x50... |
blob | commitdiff | raw | diff to current |
2008-09-26 |
Evan Cheng | unpckhps requires sse1, punpckhdq requires sse2. |
blob | commitdiff | raw | diff to current |
2008-09-25 |
Evan Cheng | With sse3 and when the source is a load or has multiple... |
blob | commitdiff | raw | diff to current |
2008-09-25 |
Evan Cheng | pmovsxbq etc. requires sse4.1. |
blob | commitdiff | raw | diff to current |
2008-09-24 |
Evan Cheng | Fix patterns for SSE4.1 move and sign extend instructio... |
blob | commitdiff | raw | diff to current |
2008-09-12 |
Dan Gohman | Rename ConstantSDNode::getValue to getZExtValue, for... |
blob | commitdiff | raw | diff to current |
2008-09-05 |
Eli Friedman | Fix for PR2687: Add patterns to match sint_to_fp and... |
blob | commitdiff | raw | diff to current |
2008-08-28 |
Evan Cheng | FsFLD0S{S|D} and V_SETALLONES are as cheap as moves. |
blob | commitdiff | raw | diff to current |
2008-08-20 |
Dan Gohman | Tablegen generated code already tests the opcode value... |
blob | commitdiff | raw | diff to current |
2008-08-08 |
Dan Gohman | Add an EXTRACTPSmr pattern to match the pattern that |
blob | commitdiff | raw | diff to current |
2008-08-05 |
Evan Cheng | Fix PR2620: Fix X86cmppd selection code so it expects... |
blob | commitdiff | raw | diff to current |
2008-07-17 |
Nate Begeman | Fix a typo in last commit |
blob | commitdiff | raw | diff to current |
2008-07-17 |
Nate Begeman | SSE codegen for vsetcc nodes |
blob | commitdiff | raw | diff to current |
2008-07-10 |
Evan Cheng | Fix for PR2472. Use movss to set lower 32-bits of a... |
blob | commitdiff | raw | diff to current |
2008-06-16 |
Evan Cheng | Horizontal-add instructions are not commutative. |
blob | commitdiff | raw | diff to current |
2008-06-16 |
Evan Cheng | mpsadbw is commutable. |
blob | commitdiff | raw | diff to current |
2008-06-13 |
Duncan Sands | Disable some DAG combiner optimizations that may be |
blob | commitdiff | raw | diff to current |
2008-05-29 |
Evan Cheng | Implement vector shift up / down and insert zero with... |
blob | commitdiff | raw | diff to current |
2008-05-28 |
Dan Gohman | Fix the encoding for two more "rm" instructions that... |
blob | commitdiff | raw | diff to current |
2008-05-28 |
Mon P Wang | Fixed X86 encoding error CVTPS2PD and CVTPD2PS when... |
blob | commitdiff | raw | diff to current |
2008-05-24 |
Evan Cheng | Eliminate x86.sse2.punpckh.qdq and x86.sse2.punpckl... |
blob | commitdiff | raw | diff to current |
2008-05-24 |
Evan Cheng | Eliminate x86.sse2.movs.d, x86.sse2.shuf.pd, x86.sse2... |
blob | commitdiff | raw | diff to current |
2008-05-24 |
Evan Cheng | Remove x86.sse2.loadh.pd and x86.sse2.loadl.pd. These... |
blob | commitdiff | raw | diff to current |
2008-05-23 |
Evan Cheng | Use movlps / movhps to modify low / high half of 16... |
blob | commitdiff | raw | diff to current |
2008-05-23 |
Evan Cheng | Fix a duplicated pattern. |
blob | commitdiff | raw | diff to current |
2008-05-23 |
Dan Gohman | Use PMULDQ for v2i64 multiplies when SSE4.1 is availabl... |
blob | commitdiff | raw | diff to current |
2008-05-23 |
Evan Cheng | Bug: rcpps can only folds a load if the address is... |
blob | commitdiff | raw | diff to current |
2008-05-22 |
Evan Cheng | Add missing patterns. |
blob | commitdiff | raw | diff to current |
2008-05-20 |
Evan Cheng | movsd and movq do not require 16-byte alignment. This... |
blob | commitdiff | raw | diff to current |
2008-05-13 |
Nate Begeman | Fix one more encoding bug. |
blob | commitdiff | raw | diff to current |
2008-05-13 |
Nate Begeman | Fix and encoding error in the psrad xmm, imm8 instruction. |
blob | commitdiff | raw | diff to current |
2008-05-12 |
Nate Begeman | Teach Legalize how to scalarize VSETCC |
blob | commitdiff | raw | diff to current |
2008-05-12 |
Nate Begeman | Initial X86 codegen support for VSETCC. |
blob | commitdiff | raw | diff to current |
2008-05-10 |
Evan Cheng | Some clean up. |
blob | commitdiff | raw | diff to current |
2008-05-09 |
Evan Cheng | Add a pattern to do move the low element of a v4f32... |
blob | commitdiff | raw | diff to current |
2008-05-09 |
Evan Cheng | Handle a few more cases of folding load i64 into xmm... |
blob | commitdiff | raw | diff to current |
2008-05-08 |
Evan Cheng | Use movq to move low half of XMM register and zero... |
blob | commitdiff | raw | diff to current |
2008-05-08 |
Evan Cheng | Handle vector move / load which zero the destination... |
blob | commitdiff | raw | diff to current |
2008-05-03 |
Evan Cheng | Add separate intrinsics for MMX / SSE shifts with i32... |
blob | commitdiff | raw | diff to current |
2008-05-02 |
Evan Cheng | 80 column violation. |
blob | commitdiff | raw | diff to current |
2008-04-20 |
Chris Lattner | A better fix for my previous patch, MOVZQI2PQIrr just... |
blob | commitdiff | raw | diff to current |
2008-04-16 |
Dan Gohman | Add support for the form of the SSE41 extractps instruc... |
blob | commitdiff | raw | diff to current |
2008-04-10 |
Chris Lattner | Fix the x86-64 side of PR2108 by adding a v2f64 version of |
blob | commitdiff | raw | diff to current |
2008-04-05 |
Evan Cheng | Favors pshufd over shufps when shuffling elements from... |
blob | commitdiff | raw | diff to current |
2008-03-26 |
Evan Cheng | Fix some SSE4.1 instruction encoding bugs. |
blob | commitdiff | raw | diff to current |
2008-03-24 |
Evan Cheng | - SSE4.1 extractfps extracts a f32 into a gr32 register... |
blob | commitdiff | raw | diff to current |
2008-03-16 |
Nate Begeman | Add a couple missing SSE4 instructions |
blob | commitdiff | raw | diff to current |
2008-03-15 |
Evan Cheng | Replace all target specific implicit def instructions... |
blob | commitdiff | raw | diff to current |
2008-03-14 |
Evan Cheng | Fix some 80 col violations. |
blob | commitdiff | raw | diff to current |
2008-03-14 |
Evan Cheng | Fix a number of encoding bugs. SSE 4.1 instructions... |
blob | commitdiff | raw | diff to current |
2008-03-12 |
Evan Cheng | Clean up my own mess. |
blob | commitdiff | raw | diff to current |
2008-03-08 |
Evan Cheng | Implement x86 support for @llvm.prefetch. It correspond... |
blob | commitdiff | raw | diff to current |
2008-03-05 |
Evan Cheng | isTwoAddress = 1 -> Constraints. |
blob | commitdiff | raw | diff to current |
2008-03-05 |
Evan Cheng | PSLLWri etc. are two-address instructions. |
blob | commitdiff | raw | diff to current |
2008-02-18 |
Evan Cheng | - When DAG combiner is folding a bit convert into a... |
blob | commitdiff | raw | diff to current |
2008-02-16 |
Andrew Lenharth | llvm.memory.barrier, and impl for x86 and alpha |
blob | commitdiff | raw | diff to current |
2008-02-12 |
Nate Begeman | SSE4.1 64b integer insert/extract pattern support |
blob | commitdiff | raw | diff to current |
2008-02-11 |
Nate Begeman | Enable SSE4 codegen and pattern matching. |
blob | commitdiff | raw | diff to current |
2008-02-10 |
Nate Begeman | xmm0 variable blends |
blob | commitdiff | raw | diff to current |
2008-02-09 |
Nate Begeman | memopv16i8 had wrong alignment requirement, would have... |
blob | commitdiff | raw | diff to current |
2008-02-09 |
Nate Begeman | Skeleton of insert and extract matching, more to come |
blob | commitdiff | raw | diff to current |
2008-02-04 |
Nate Begeman | The rest of the SSE4.1 intrinsic patterns that are... |
blob | commitdiff | raw | diff to current |
2008-02-04 |
Nate Begeman | Some more SSE 4.1 intrinsic patterns. |
blob | commitdiff | raw | diff to current |
2008-02-03 |
Nate Begeman | SSE 4.1 Intrinsics and detection |
blob | commitdiff | raw | diff to current |
2008-01-24 |
Chris Lattner | Significantly simplify and improve handling of FP funct... |
blob | commitdiff | raw | diff to current |
2008-01-11 |
Chris Lattner | add some missing flags. |
blob | commitdiff | raw | diff to current |
2008-01-10 |
Chris Lattner | Start inferring side effect information more aggressive... |
blob | commitdiff | raw | diff to current |
2008-01-10 |
Chris Lattner | remove explicit sets of 'neverHasSideEffects' that... |
blob | commitdiff | raw | diff to current |
2008-01-06 |
Chris Lattner | rename isLoad -> isSimpleLoad due to evan's desire... |
blob | commitdiff | raw | diff to current |
2007-12-29 |
Chris Lattner | Remove attribution from file headers, per discussion... |
blob | commitdiff | raw | diff to current |
2007-12-20 |
Evan Cheng | Fix JIT encoding for CMPSD as well. |
blob | commitdiff | raw | diff to current |
2007-12-17 |
Bill Wendling | Add "mayHaveSideEffects" and "neverHasSideEffects"... |
blob | commitdiff | raw | diff to current |
2007-12-16 |
Chris Lattner | Fix the JIT encoding of cmp*ss, which aborts with this... |
blob | commitdiff | raw | diff to current |
next |