Update diff November 6th 2024, 12:49:26 pm

[github-actions]

from: llvm/llvm-project@c96a85a
to: llvm/llvm-project@e3a0775
commit: 0f4e21b

Change Logs

from c96a85abfde822f2eda9076eb40078389b21f23e to e3a0775651190a23d8234615b9fdadd81c1c24bc

e3a0775651190a23d8234615b9fdadd81c1c24bc [VectorCombine] foldExtractedCmps - (re-)enable fold on non-commutative binops
38fffa630ee80163dc65e759392ad29798905679 [LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548)
5a062191f7b9467aaddb6fe4b84c16e60fe85cc8 [SPIR-V] Ensure correct pointee types of some OpenCL Extended Instructions' pointer arguments (#114846)
d8139ae50f72046a2fce36055d8dc936b50d20ef [lldb][test] Use -gdwarf to fix same CU breakpoint test on Windows on Arm
cab606c30661a746b2513a8330e0c8eca771913e [LoongArch] Enable alias analysis by default (#114980)
d8354d63db66e5d67d74b24b1611b578700f1134 [VectorCombine] Extend test coverage for #114901 with commuted test case
3c4e6c17f066d9cf5a5b065a05bdff472f721bf6 [flang][Driver] When linking with the Fortran runtime, the addArchSpecificRPath() should be called too (#114837)
2bbb6c067020bef50435c2ba5addcd66d1b9a1dd [clang][NFC] Spell out DynTypedNode instead of auto (#114427)
eab7be5d42ad30c9992ff72c3be9298702001dc8 [libc++] Forward more algorithms to the classic algorithms (#114674)
56077e5ac09eb2d6b7ca818abce2bbbcf179f529 [LLD][COFF] Add support for locally imported EC symbols (#114985)
ea6b8fa4b9b48a11c2657bedf35ad5291b1e2b9c [SDAG] Merge multiple-result libcall expansion into DAG.expandMultipleResultFPLibCall() (#114792)
40556d08491f530e03746fb188b38e7f9cb272c7 [MLIR][Tensor] Fix out-of-bounds FoldEmptyTensorWithDimOp crash (#112196)
f87484d5910c1c708bfd93ef588d6ff8307e2477 Fix libnvptxcompiler_static.a absolute path (#115015)
5a16ed96c5362aa8e9610fa266d6f6202b19edc3 [RISCV] Add +unaligned-scalar-mem to spacemit-x60 (#115125)
c75353313ed73c6dc04beb322954bb905906f4a1 [X86] combineConcatVectorOps - add 256-bit concat(shuffle(),shuffle()) handling
9f8c3d3796ebf7ddd4a85134ff109cf03a0b9b5e [lldb][test] Correct typo in breakpoint test file name
2d56de9e7e4a3accde42b4d7d329acd007989df8 Revert "[ARM] Add extra tests for CVE-2024-7883 with undef/poison"
d77a36e01b8fed496b29c3b2c12526f8dc380766 [LoopVectorize] Use new getUniqueLatchExitBlock routine (#108231)
f363f9d61eaff7090a19d226ea8786b2987d4fcc [ORC][Runtime] Add dlupdate for elf (#110406)
ebfafa2511f92eed484895f3265ced40f4c1fc70 [SPIR-V] Fix OpFunctionParameter vs. OpTypeFunction types for pointer arguments when there are functions with aggregate arguments (#115044)
3d4d033ceaf9e72491a20e9210f396aa3ec52fa5 [AArch64][Arm] Add nested double reduction tests. NFC
2f48765b45ba87d780caf7d058d416b5dda32d7e [libc][bazel] Remove -mllvm --tail-merge-threshold=0 from Bazel. (#115061)
270bfb2f2abc48ec916fce7e677fe3cc6f2908d0 [X86] Add test coverage for #114959
e29d092af8b0ed2b15ce1dfd9fc4caef1976eef7 [X86] getFauxShuffleMask - add ISD::SHL/SRL handling
c6f3b7bcd0596d30f8dabecdfb9e44f9a07b6e4c [libc++] Refactor the configuration macros to being always defined (#112094)
5acc4a3dc0e2145d2bfef47f1543bb291c2b866a [libc++] Remove <stddef.h> includes from the granularized <cstddef> headers (#114788)
8c565de5ec6d49143ba9ae7c73b188314d31e563 [LoongArch] Support llvm.lround intrinsics with i32 return type. (#114733)
ed9dab67e2932baf11bfa514b07b159c3bffd518 [ARM] Add extra tests for CVE-2024-7883 with undef/poison
b4263ddbe7cbcc9e0b5b0ea07c252056355301d0 [libclc] Use __clc_max in CLC functions
7be30fd5335ca7fe050ee1789ea2648f014daf1b [libclc] Move abs/abs_diff to CLC library
37ce18951fded6be1de319b05b968918cb45c00b [RISCV] Add requirement of asserts
8431494094c8732d1426763d3e1aae322fa76830 [clang] Make source locations space usage diagnostics numbers easier to read (#114999)
c0a7b60fd1b244782032fefc261c4442c54c3935 Revert "[flang][OpenMP] Add alias analysis for omp private" (#115135)
69d0bab82689d470e3fd68f50ca8b8d28f3e2294 [RISCV] Add load/store clustering in post machine schedule (#111504)
08411c855f77bd7416725c280ad3dccdc00b7dd6 [RISCV] Correct fcopysign pattern for zdinx (#114954)
dc55d31f4cf5c97b56f6b7e1c24b70674cc15a01 [mlir][tensor] Fix a crash in ExtractOp::fold (#115001)

Compilation time result (by files):
Top 5 improvements:
  ncnn/innerproduct_x86_fma.cpp.ll 8320445299 -> 7849920590 -5.66%
  ncnn/innerproduct_x86_f16c.cpp.ll 3558730793 -> 3361651644 -5.54%
  ncnn/requantize_x86_avx512.cpp.ll 4313603264 -> 4074906838 -5.53%
  ncnn/mish_x86_fma.cpp.ll 178210264 -> 168432171 -5.49%
  ncnn/requantize_x86_fma.cpp.ll 4456669428 -> 4220107210 -5.31%
Top 5 regressions:
  faiss/IndexFlat.cpp.ll 570861757 -> 577314409 +1.13%
  faiss/IndexIVFSpectralHash.cpp.ll 1738984925 -> 1754919087 +0.92%
  lightgbm/metadata.cpp.ll 4707410159 -> 4748261869 +0.87%
  gromacs/domdec_constraints.cpp.ll 1890411281 -> 1900990661 +0.56%
  gromacs/pull.cpp.ll 3401100191 -> 3419290047 +0.53%

Overall: -0.03191132%

Compilation time result (by projects):
Top 5 improvements:
  ncnn 542305508282 -> 537967798321 -0.80%
  gromacs 980315792324 -> 977617868327 -0.28%
  simdjson 7564437433 -> 7547001847 -0.23%
  darktable 138021621140 -> 137731857043 -0.21%
  mitsuba3 206399956753 -> 206116652037 -0.14%
Top 5 regressions:

Overall: -0.02640127%