Update diff October 4th 2024, 2:29:14 pm

[github-actions]

from: llvm/llvm-project@a3cc4b6
to: llvm/llvm-project@6b3220a
commit: b51ed96

Change Logs

from a3cc4b61a0414a844058f064bd19d1d4085e0f6c to 6b3220afa62bb1e77acf18dc7b622df9223e1ea8

6b3220afa62bb1e77acf18dc7b622df9223e1ea8 [InstCombine] Avoid crash on aggregate types in SimplifyDemandedUseFPClass (#111128)
d2051919bb42087ebde2a8bdcd5b2676ad16d8eb [MSVC] work-around for compile time issue 102513 (#110986)
4da4fac322165420f0522cb554e5f6f25540d564 [ArgPromotion] Consider InvokeInst in Caller alias analysis (#110335)
47c8b95daeec8e6cb012344ed037024528a73295 [CFIFixup] Factor CFI remember/restore insertion into a helper (NFC) (#111066)
d68083feba2ff869dae96e938b1fa64296f84274 [FuncSpec] Update tests to use --include-generated-funcs (NFC)
2f245875b2f71272e6d7a78b4aed5be81109e9b9 [flang][OpenMP] Handle unstructured CF in compound loop constructs (#111111)
208f42fd9bb59c007b0d6abe7eaf8c4e829740ea [mlir][NVVM] Add constant memory space identifier (#111141)
8e33ff7d5669ec62ad26a0beb81cd257cb14258e [mlir][GPU][NFC] Move dump-ptx.mlir test case (#111142)
d8f22514eb0b83f302872af4689d27dce67501fc [Clang] Automatically enable -fconvergent-functions on GPU targets (#111076)
8d661fd9fd50fe4be9a9d5a8cc8f52f23925e7f6 [analyzer] use invalidateRegions() in VisitGCCAsmStmt (#109838)

Compilation time result (by files):
Top 5 improvements:
  ncnn/rnn.cpp.ll 1317034904 -> 1177141704 -10.62%
  opencv/samplewriteconfigfile.cpp.ll 112656940 -> 106740405 -5.25%
  ncnn/lstm_x86.cpp.ll 3050026510 -> 2976293192 -2.42%
  ncnn/lstm_x86_avx.cpp.ll 3282384822 -> 3212032419 -2.14%
  ncnn/lstm_x86_fma.cpp.ll 3270197928 -> 3202468949 -2.07%
Top 5 regressions:
  velox/CoverageUtil.cpp.ll 1391652543 -> 1426691821 +2.52%
  verilator/V3SchedPartition.cpp.ll 2070880355 -> 2095595124 +1.19%
  openusd/tessellation.cpp.ll 2020150337 -> 2042026485 +1.08%
  protobuf/arena.cc.ll 972329306 -> 978832893 +0.67%
  opencv/text_detector_swt.cpp.ll 5975540273 -> 6012702510 +0.62%

Overall: -0.01307853%

Compilation time result (by projects):
Top 5 improvements:
Top 5 regressions:

Overall: -0.01548739%

Improvements:
argpromotion.NumArgumentsPromoted 529061 -> 531838 +0.52%
licm.NumMovedCalls 18798 -> 18837 +0.21%
loop-rotate.NumNotRotatedDueToHeaderSize 12496 -> 12514 +0.14%
adce.NumBranchesRemoved 21228 -> 21240 +0.06%
function-attrs.NumReadOnlyArg 1509994 -> 1510656 +0.04%
constraint-elimination.NumCondsRemoved 1219026 -> 1219435 +0.03%
memcpyopt.NumMoveToCpy 6320 -> 6322 +0.03%
correlated-value-propagation.NumCmps 236045 -> 236115 +0.03%
function-attrs.NumNoCapture 3059813 -> 3060654 +0.03%
loop-idiom.NumMemCpy 7503 -> 7505 +0.03%
Regressions:
vector-combine.NumVecLoad 3992 -> 3976 -0.40%
memory-builtins.ObjectVisitorLoad 1644448 -> 1641196 -0.20%
sccp.NumDeadBlocks 664386 -> 663722 -0.10%
sccp.NumInstRemoved 1805222 -> 1803465 -0.10%
capture-tracking.NumNotCapturedBefore 15646031 -> 15630821 -0.10%
loop-simplifycfg.NumLoopBlocksDeleted 8486 -> 8480 -0.07%
licm.NumLoadStorePromoted 54389 -> 54353 -0.07%
dse.NumFastOther 121254 -> 121191 -0.05%
loop-simplifycfg.NumTerminatorsFolded 8005 -> 8001 -0.05%
memcpyopt.NumMemSetInfer 342207 -> 342059 -0.04%