LLVM/project 942fda2 — clang/lib/Headers avx10_2convertintrin.h, llvm/test/Analysis/CostModel/AArch64 arith-widening.ll
rebase Created using spr 1.3.4
rebase Created using spr 1.3.4
run 'git merge main'
Delta | File | |
---|---|---|
+65,595 | -0 | llvm/test/CodeGen/SPIRV/extensions/SPV_INTEL_long_composites/long-spec-const-composite.ll |
+7,782 | -7,782 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+102,453 | -36,858 | 8,560 files not shown |
+809,659 | -558,257 | 8,566 files |
[RISCV] Consolidate some DecoderNamespaces for standard extensions. (#128954) First thing to know is that the subtarget feature checks used to block accessing a decoder table are only a performance optimization and not required for functionality. The tables have their own predicate checks. I've removed them from all the standard extension tables. -RV32 Zacas decoder namespace has been renamed to RV32GPRPair, I think Zilsd(rv32 load/store pair) can go in here too. -The RV32 Zdinx table has been renamed to also use RV32GPRPair. -The Zfinx table has been renamed to remove superflous "RV" prefix. -Zcmp and Zcmt tables have been combined into a ZcOverlap table. I think Zclsd(rv32 compressed load/store pair) can go in here too. -All the extra standard extension tables are checked after the main standard extension table. This makes the common case of the main table matching occur earlier. -Zicfiss is the exception to this as it needs to be checked before the main table since it overrides some encodings from Zcmop. This can't be handled by a predicate based priority as Zicfiss only overrides a subset of Zcmop encodings.
fix order Created using spr 1.3.4
Delta | File | |
---|---|---|
+7,782 | -7,782 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v4i64.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+43,825 | -43,825 | 7,407 files not shown |
+681,199 | -531,648 | 7,413 files |
[𝘀𝗽𝗿] changes introduced through rebase Created using spr 1.3.4 [skip ci]
Delta | File | |
---|---|---|
+7,782 | -7,782 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v4i64.ll |
+43,825 | -43,825 | 7,401 files not shown |
+681,165 | -531,621 | 7,407 files |
Rebase, add option Created using spr 1.3.5
Delta | File | |
---|---|---|
+4,730 | -0 | llvm/test/tools/llvm-mca/RISCV/SiFiveP600/vlseg-vsseg.s |
+1,896 | -940 | llvm/test/Analysis/CostModel/AArch64/arith-widening.ll |
+230 | -1,254 | clang/lib/Lex/ModuleMap.cpp |
+1,298 | -0 | llvm/test/Transforms/LoopVectorize/AArch64/optsize_minsize.ll |
+1,240 | -0 | clang/lib/Lex/ModuleMapFile.cpp |
+530 | -554 | llvm/test/CodeGen/X86/vector-interleaved-store-i8-stride-7.ll |
+9,924 | -2,748 | 432 files not shown |
+27,460 | -8,770 | 438 files |
[RISCV] Add Xqccmp 0.1 Assembly Support (#128731) Xqccmp is a new spec by Qualcomm that makes a vendor-specific effort to solve the push/pop + frame pointers issue. Broadly, it takes the Zcmp instructions and reverse the order they push/pop registers in, which ends up matching the frame pointer convention. This extension adds a new instruction not present in Zcmp, `qc.cm.pushfp`, which will set `fp` to the incoming `sp` value after it has pushed the registers. This change duplicates the Zcmp implementation, with minor changes to mnemonics (for the `qc.` prefix), predicates, and the addition of `qc.cm.pushfp`. There is also new logic to prevent combining Xqccmp and Zcmp. Xqccmp is kept separate to Xqci for decoding/encoding etc, as the specs are separate today. Specification: https://github.com/quic/riscv-unified-db/releases/tag/Xqccmp_extension-0.1.0
Merge branch 'main' into users/mtrofin/02-25-_ctxprof_don_t_inline_weak_symbols_after_instrumentation
Fix formatting Created using spr 1.3.5
Delta | File | |
---|---|---|
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm90.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm60.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm70.ll |
+26 | -2,694 | lldb/tools/lldb-dap/lldb-dap.cpp |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log10.ll |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log.ll |
+18,594 | -4,360 | 1,391 files not shown |
+65,876 | -28,157 | 1,397 files |
[RISCV][MC] Add assembler support for XRivosVisni (#128773) This implements assembler support for the XRivosVisni custom/vendor extension from Rivos Inc. which is defined in: https://github.com/rivosinc/rivos-custom-extensions (See src/xrivosvisni.adoc) Codegen support will follow in separate changes.
[RISCV] Merge some of the Sifive decoder tables. (#128794) This makes a single table for vector and another table for system. I left sf.cease out of system because its not in custom encoding space. The other system instructions are in the custom part of OPC_SYSTEM.
Delta | File | |
---|---|---|
+13 | -20 | llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp |
+7 | -7 | llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td |
+20 | -27 | 2 files |
rebase Created using spr 1.3.4
Delta | File | |
---|---|---|
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm90.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm60.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm70.ll |
+40 | -4,869 | lldb/tools/lldb-dap/lldb-dap.cpp |
+4,241 | -0 | llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log.ll |
+22,085 | -5,702 | 927 files not shown |
+63,396 | -23,486 | 933 files |
rebase Created using spr 1.3.4
Delta | File | |
---|---|---|
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log10.ll |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log.ll |
+666 | -669 | lldb/unittests/UnwindAssembly/x86/Testx86AssemblyInspectionEngine.cpp |
+1,029 | -0 | llvm/test/CodeGen/AMDGPU/llvm.amdgcn.readfirstlane.ll |
+255 | -324 | llvm/test/CodeGen/AMDGPU/llvm.log2.ll |
+392 | -163 | llvm/test/CodeGen/AMDGPU/GlobalISel/usubsat.ll |
+3,870 | -2,822 | 391 files not shown |
+15,954 | -9,056 | 397 files |
[𝘀𝗽𝗿] changes introduced through rebase Created using spr 1.3.4 [skip ci]
Delta | File | |
---|---|---|
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log.ll |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log10.ll |
+666 | -669 | lldb/unittests/UnwindAssembly/x86/Testx86AssemblyInspectionEngine.cpp |
+1,029 | -0 | llvm/test/CodeGen/AMDGPU/llvm.amdgcn.readfirstlane.ll |
+255 | -324 | llvm/test/CodeGen/AMDGPU/llvm.log2.ll |
+392 | -163 | llvm/test/CodeGen/AMDGPU/GlobalISel/uaddsat.ll |
+3,870 | -2,822 | 391 files not shown |
+15,954 | -9,056 | 397 files |
[RISCV][NFC] Merge Xqci Decoder Tables (#128140) RISC-V has multiple decoder tables because there is no guarantee that non-standard extensions do not overlap with each other. Qualcomm's Xqci family of extensions are intended to be implemented together, and therefore we want a single decode table for this group of extensions. This should be more efficient overall, and allows us to use tablegen's existing mechanism that finds overlapping encodings within the group. To implement this, the key addition is `TRY_TO_DECODE_FEATURE_ANY`, which will use the provided decoder table if any of the features from the FeatureBitset (first argument) are enabled, rather than if all are enabled.
Delta | File | |
---|---|---|
+21 | -28 | llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp |
+26 | -22 | llvm/lib/Target/RISCV/RISCVInstrInfoXqci.td |
+47 | -50 | 2 files |
[RISCV] Correctly Decode Unsigned Immediates with Ranges (#128584) We currently have two operands upstream that are an unsigned immediate with a range constraint - `uimm8ge32` (for `cm.jalt`) and `uimm5gt3` (for `qc.shladd`). Both of these were using `decodeUImmOperand<N>` for decoding. For `Zcmt` this worked, because the generated decoder automatically checked for `cm.jt` first because the 8 undefined bits in `cm.jalt` are `000?????` in `cm.jt` (this is to do with the range lower-bound being a power-of-two). For Zcmt, this patch is NFC. We have less luck with `Xqciac` - `qc.shladd` is being decoded where the `uimm5` field is 3 or lower. This patch fixes this by introducing a `decodeUImmOperandGE<Width, LowerBound>` helper, which will corretly return `MCDisassembler::Fail` when the immediate is below the lower bound. I have added a test to show the encoding where `uimm5` is equal to 3 is no longer disassembled as `qc.shladd`.
[𝘀𝗽𝗿] changes introduced through rebase Created using spr 1.3.4 [skip ci]
Delta | File | |
---|---|---|
+7,194 | -3,880 | llvm/test/CodeGen/AMDGPU/bf16.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm60.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm70.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm90.ll |
+272 | -5,007 | lldb/tools/lldb-dap/lldb-dap.cpp |
+4,269 | -28 | llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll |
+28,775 | -8,915 | 1,280 files not shown |
+84,457 | -34,573 | 1,286 files |
comments Created using spr 1.3.5-bogner
Delta | File | |
---|---|---|
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm90.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm60.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm70.ll |
+40 | -4,869 | lldb/tools/lldb-dap/lldb-dap.cpp |
+4,241 | -0 | llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll |
+764 | -833 | llvm/test/CodeGen/AMDGPU/llvm.log.ll |
+22,085 | -5,702 | 813 files not shown |
+59,305 | -21,472 | 819 files |
Merge branch 'main' into users/junlarsen/codegen_replace_pointertype_getunqual_type_with_opaque_pointer_version_nfc_
Delta | File | |
---|---|---|
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm90.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm60.ll |
+5,680 | -0 | llvm/test/CodeGen/NVPTX/cmpxchg-sm70.ll |
+26 | -2,694 | lldb/tools/lldb-dap/lldb-dap.cpp |
+1,483 | -14 | llvm/test/CodeGen/NVPTX/cmpxchg.ll |
+666 | -669 | lldb/unittests/UnwindAssembly/x86/Testx86AssemblyInspectionEngine.cpp |
+19,215 | -3,377 | 565 files not shown |
+42,845 | -13,130 | 571 files |
rebase Created using spr 1.3.4
Delta | File | |
---|---|---|
+7,163 | -3,829 | llvm/test/CodeGen/AMDGPU/bf16.ll |
+4,269 | -28 | llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll |
+249 | -2,316 | lldb/tools/lldb-dap/lldb-dap.cpp |
+2,163 | -0 | llvm/test/CodeGen/X86/vector-lrint-f16.ll |
+190 | -1,086 | flang/test/Lower/allocatable-assignment.f90 |
+0 | -1,265 | flang/test/Lower/array-expression.f90 |
+14,034 | -8,524 | 1,008 files not shown |
+53,160 | -26,409 | 1,014 files |
rebase Created using spr 1.3.4
[RISCV] Add Qualcomm uC Xqcilia (Large Immediate Arithmetic) extension (#124706) This extension adds eight 48 bit large arithmetic instructions. The current spec can be found at: https://github.com/quic/riscv-unified-db/releases/latest This patch adds assembler only support.
Rebase Created using spr 1.3.5
Delta | File | |
---|---|---|
+7,782 | -7,782 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v4i64.ll |
+43,825 | -43,825 | 4,283 files not shown |
+513,606 | -420,700 | 4,289 files |
rebase Created using spr 1.3.4
Delta | File | |
---|---|---|
+7,782 | -7,782 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v4i64.ll |
+43,825 | -43,825 | 4,198 files not shown |
+510,585 | -434,236 | 4,204 files |
refactor Created using spr 1.3.4
Delta | File | |
---|---|---|
+30,641 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+27,139 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f32.v8f32.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i32.v8i32.ll |
+164,956 | -0 | 15,561 files not shown |
+1,914,489 | -576,768 | 15,567 files |
[RISCV] Assembler support for XRivosVizip (#127694) This implements assembler support for the XRivosVizip custom/vendor extension from Rivos Inc. which is defined in: https://github.com/rivosinc/rivos-custom-extensions (See src/xrivosvizip.adoc) Codegen support will follow in a separate change.
[RISCV] Simplify the debug messages in the disassembler. (#128102) Move the printing of "table" to the macro instantiation. Don't use string concatenation in the macro. Print DESC as it own string. This allows the "Trying " and " table:" to only appear in the binary once instead of being part of every string. Remove "custom opcode" from the messages, I don't think it provides much value after mentioning the vendor. I'm hoping to replace the macros with a table of features, decoder table pointer, and string that we can iterate over.
Delta | File | |
---|---|---|
+63 | -78 | llvm/lib/Target/RISCV/Disassembler/RISCVDisassembler.cpp |
+63 | -78 | 1 files |
Address @DavidSpickett's comment Created using spr 1.3.5-bogner
Delta | File | |
---|---|---|
+7,513 | -7,513 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+7,427 | -7,427 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+7,255 | -7,255 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4p0.v4p0.ll |
+6,967 | -6,967 | llvm/test/CodeGen/AMDGPU/shufflevector.v4i64.v4i64.ll |
+43,556 | -43,556 | 6,429 files not shown |
+608,030 | -497,429 | 6,435 files |
Merge branch 'main' into users/chapuni/yaml/newgen Conflicts: llvm/test/tools/llvm-cov/Inputs/branch-logical-mixed.cpp llvm/test/tools/llvm-cov/Inputs/branch-macros.cpp
Delta | File | |
---|---|---|
+30,641 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+27,139 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2p3.v8p3.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f32.v8f32.ll |
+164,956 | -0 | 30,003 files not shown |
+3,294,806 | -1,187,822 | 30,009 files |
Merge branch 'main' into users/ylzsx/r-call36
Delta | File | |
---|---|---|
+30,641 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i64.v8i64.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f16.v8f16.ll |
+27,655 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2bf16.v8bf16.ll |
+27,139 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i16.v8i16.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2f32.v8f32.ll |
+25,933 | -0 | llvm/test/CodeGen/AMDGPU/shufflevector.v2i32.v8i32.ll |
+164,956 | -0 | 10,990 files not shown |
+1,581,431 | -416,524 | 10,996 files |