aboutsummaryrefslogtreecommitdiff
path: root/lib/Target
AgeCommit message (Collapse)Author
2006-04-27Support for passing 128-bit vector arguments via XMM registers.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27992 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-27OopsEvan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27989 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-27Bug fix: not updating NumIntRegs.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27988 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-27- Clean up formal argument lowering code. Prepare for vector pass by value work.Evan Cheng
- Fixed vararg support. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27985 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-26Fix fastcc failures.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27980 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-26Switching over FORMAL_ARGUMENTS mechanism to lower call arguments.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27975 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-25Keep the stack from on darwin 16-byte aligned. This fixes many JITNate Begeman
failres. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27973 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-25Separate LowerOperation() into multiple functions, one per opcode.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27972 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-25Fix a typo.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27968 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-25No functionality changes, but cleaner code with correct comments.Nate Begeman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27966 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-25Explicitly specify result type for def : Pat<> patterns (if it produces a vectorEvan Cheng
result). Otherwise tblgen will pick the default (v16i8 for 128-bit vector). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27965 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24Added X86 SSE2 intrinsics which can be represented as vector_shuffles. This isEvan Cheng
a temporary workaround for the 2-wide vector_shuffle problem (i.e. its mask would have type v2i32 which is not legal). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27964 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24Add a new entry.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27963 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24Special case handling two wide build_vector(0, x).Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27961 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24Some missing movlps, movhps, movlpd, and movhpd patterns.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27960 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24A little bit more build_vector enhancement for v8i16 cases.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27959 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24Remove a completed entry.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27958 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-24MakeMIInst() should handle jump table index operands.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27955 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-23Add a noteChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27954 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-23MOVL shuffle (i.e. movd or movss / movsd from memory) of undef, V2 == V2Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27953 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22Optimized stores to the constant pool, while cool, are unnecessary.Nate Begeman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27948 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22JumpTable support! What this represents is working asm and jit support forNate Begeman
x86 and ppc for 100% dense switch statements when relocations are non-PIC. This support will be extended and enhanced in the coming days to support PIC, and less dense forms of jump tables. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27947 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22Don't do all the lowering stuff for 2-wide build_vector's. Also, minor ↵Evan Cheng
optimization for shuffle of undef. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27946 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22Fix a performance regression. Use {p}shuf* when there are only two distinct ↵Evan Cheng
elements in a build_vector. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27945 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-22Teach the JIT how to relocate LI, this fixes the JIT on Prolangs-C/TimberWolfMCChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27943 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21Revamp build_vector lowering to take advantage of movss and movd instructions.Evan Cheng
movd always clear the top 96 bits and movss does so when it's loading the value from memory. The net result is codegen for 4-wide shuffles is much improved. It is near optimal if one or more elements is a zero. e.g. __m128i test(int a, int b) { return _mm_set_epi32(0, 0, b, a); } compiles to _test: movd 8(%esp), %xmm1 movd 4(%esp), %xmm0 punpckldq %xmm1, %xmm0 ret compare to gcc: _test: subl $12, %esp movd 20(%esp), %xmm0 movd 16(%esp), %xmm1 punpckldq %xmm0, %xmm1 movq %xmm1, %xmm0 movhps LC0, %xmm0 addl $12, %esp ret or icc: _test: movd 4(%esp), %xmm0 #5.10 movd 8(%esp), %xmm3 #5.10 xorl %eax, %eax #5.10 movd %eax, %xmm1 #5.10 punpckldq %xmm1, %xmm0 #5.10 movd %eax, %xmm2 #5.10 punpckldq %xmm2, %xmm3 #5.10 punpckldq %xmm3, %xmm0 #5.10 ret #5.10 There are still room for improvement, for example the FP variant of the above example: __m128 test(float a, float b) { return _mm_set_ps(0.0, 0.0, b, a); } _test: movss 8(%esp), %xmm1 movss 4(%esp), %xmm0 unpcklps %xmm1, %xmm0 xorps %xmm1, %xmm1 movlhps %xmm1, %xmm0 ret The xorps and movlhps are unnecessary. This will require post legalizer optimization to handle. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27939 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21Fix the commentNate Begeman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27938 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21Change the PPC JIT to use a Static relocation modelNate Begeman
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27937 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21fix thinkoChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27935 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21add some low-prio notesChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27934 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-21Now generating perfect (I think) code for "vector set" with a single non-zeroEvan Cheng
scalar value. e.g. _mm_set_epi32(0, a, 0, 0); ==> movd 4(%esp), %xmm0 pshufd $69, %xmm0, %xmm0 _mm_set_epi8(0, 0, 0, 0, 0, a, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0); ==> movzbw 4(%esp), %ax movzwl %ax, %eax pxor %xmm0, %xmm0 pinsrw $5, %eax, %xmm0 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27923 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20Fix the CodeGen/PowerPC/buildvec_canonicalize.ll regression last night.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27908 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20add a noteChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27907 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20remove some v9 specific codeChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27900 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20Remove this obsolete fileChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27895 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20This target is no longer built. The ,v files now live in the reoptimizer.Chris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27885 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20- Added support to turn "vector clear elements", e.g. pand V, <-1, -1, 0, -1>Evan Cheng
to a vector shuffle. - VECTOR_SHUFFLE lowering change in preparation for more efficient codegen of vector shuffle with zero (or any splat) vector. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27875 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20Make sure that the new instructions selected have the right type. This fixesChris Lattner
CodeGen/PowerPC/2006-04-19-vmaddfp-crash.ll git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27868 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-20Handle v2i64 BUILD_VECTOR custom lowering correctly. v2i64 is a legal type,Evan Cheng
but i64 is not. If possible, change a i64 op to a f64 (e.g. load, constant) and then cast it back. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27849 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19isSplatMask() bug: first element can be an undef.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27847 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19- Added support to do aribitrary 4 wide shuffle with no more than threeEvan Cheng
instructions. - Fixed a commute vector_shuff bug. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27845 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19Prefer {p}unpack* and mov*dup over {p}shuf* as well.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27844 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19Renamed AddedCost to AddedComplexity.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27843 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19- Renamed AddedCost to AddedComplexity.Evan Cheng
- Added more movhlps and movlhps patterns. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27842 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19Commute vector_shuffle to match more movlhps, movlp{s|d} cases.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27840 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19More mov{h|l}p{d|s} patterns.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27836 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19- More mov{h|l}ps patterns.Evan Cheng
- Increase cost (complexity) of patterns which match mov{h|l}ps ops. These are preferred over shufps in most cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27835 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19Allow "let AddedCost = n in" to increase pattern complexity.Evan Cheng
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27834 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19add a noteChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27832 91177308-0d34-0410-b5e6-96231b3b80d8
2006-04-19add a noteChris Lattner
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@27828 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.2.3-18-g5258 (git 2.32.0) at 2025-07-17 01:17:23 +0000