Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

pre-commit: PR115305 #1653

Closed
wants to merge 2 commits into from
Closed

pre-commit: PR115305 #1653

wants to merge 2 commits into from

Conversation

dtcxzyw
Copy link
Owner

@dtcxzyw dtcxzyw commented Nov 7, 2024

Link: llvm/llvm-project#115305
Requested by: @nikic

@github-actions github-actions bot mentioned this pull request Nov 7, 2024
Open
@github-actions GitHub Actions
Copy link
Contributor

github-actions bot commented Nov 7, 2024

runner: buildkitsandbox
baseline runner: cseadmin-PowerEdge-R830
baseline: llvm/llvm-project@0c0d7a6
patch: llvm/llvm-project#115305
sha256: 4c1f328c7ea0f1c16851ad9516c081ba7007216047c5694d56e8e55431c4e88f
commit: 328db26
78 files changed, 63949 insertions(+), 64047 deletions(-)

Compilation time result (by files):
Top 5 improvements:
  opencv/contours.cpp.ll 1251118544 -> 1226233259 -1.99%
  lightgbm/metadata.cpp.ll 4606235589 -> 4576405479 -0.65%
  faiss/ScalarQuantizer.cpp.ll 6139148278 -> 6099723476 -0.64%
  lvgl/lv_draw_sw_blend_to_argb8888.ll 1970786676 -> 1958953804 -0.60%
  gromacs/freeenergydispatch.cpp.ll 666828081 -> 662976333 -0.58%
Top 5 regressions:
  spike/s_subMagsF32.ll 62734103 -> 64435810 +2.71%
  grpc/utils.cc.ll 77022249 -> 77511627 +0.64%
  php/optimize_temp_vars_5.ll 272796103 -> 274402149 +0.59%
  openusd/warped_motion.c.ll 860050901 -> 864381223 +0.50%
  diesel-rs/ivb4nc5xs1gnmkx.ll 4478431502 -> 4490665120 +0.27%

Overall: -0.06983438%

Compilation time result (by projects):
Top 5 improvements:
  cxxopts 11686294587 -> 11665352417 -0.18%
  g2o 447038542310 -> 446348440329 -0.15%
  log-rs 361794312 -> 361294547 -0.14%
  rayon-rs 7196709957 -> 7187306419 -0.13%
  libzmq 20935591288 -> 20908682706 -0.13%
Top 5 regressions:

Overall: -0.05380193%

Improvements:
  correlated-value-propagation.NumAddNSW 188049 -> 189026 +0.52%
  correlated-value-propagation.NumAddNW 347343 -> 348084 +0.21%
  correlated-value-propagation.NumNSW 413511 -> 414251 +0.18%
  correlated-value-propagation.NumNW 766132 -> 766373 +0.03%
  correlated-value-propagation.NumPhis 940035 -> 940192 +0.02%
  indvars.NumElimCmp 46394 -> 46398 +0.01%
  instcombine.NegatorNumValuesVisited 18405612 -> 18407022 +0.01%
  instcombine.NumSunkInst 2698354 -> 2698502 +0.01%
  instcombine.NegatorTotalNegationsAttempted 17655711 -> 17656613 +0.01%
  gvn.IsValueFullyAvailableInBlockNumSpeculationsMax 462907 -> 462924 +0.00%
Regressions:
  correlated-value-propagation.NumSIToFP 1438 -> 1297 -9.81%
  correlated-value-propagation.NumSRems 1199 -> 1111 -7.34%
  correlated-value-propagation.NumSDivs 16901 -> 16456 -2.63%
  correlated-value-propagation.NumUDivURemsNarrowed 8418 -> 8324 -1.12%
  correlated-value-propagation.NumSICmps 44557 -> 44287 -0.61%
  correlated-value-propagation.NumSExt 36113 -> 35926 -0.52%
  correlated-value-propagation.NumSubNUW 25765 -> 25662 -0.40%
  correlated-value-propagation.NumMinMax 9103 -> 9081 -0.24%
  correlated-value-propagation.NumNNeg 73860 -> 73682 -0.24%
  correlated-value-propagation.NumSubNW 94598 -> 94447 -0.16%

mitsuba3/jitallocator.cpp.ll crash
mitsuba3/x86rapass.cpp.ll crash
darktable/introspection_dither.c.ll crash
darktable/introspection_basecurve.c.ll crash
darktable/introspection_retouch.c.ll crash
darktable/introspection_colorzones.c.ll crash
darktable/common.c.ll crash
darktable/introspection_cacorrectrgb.c.ll crash
darktable/introspection_lut3d.c.ll crash
darktable/introspection_toneequal.c.ll crash
darktable/introspection_tonecurve.c.ll crash
darktable/introspection_denoiseprofile.c.ll crash
darktable/introspection_colorin.c.ll crash
darktable/introspection_liquify.c.ll crash
darktable/introspection_sharpen.c.ll crash
darktable/introspection_colormapping.c.ll crash
darktable/introspection_hazeremoval.c.ll crash
darktable/introspection_cacorrect.c.ll crash
darktable/print_settings.c.ll crash
darktable/introspection_ashift.c.ll crash
darktable/amaze.cc.ll crash
darktable/introspection_filmicrgb.c.ll crash
darktable/introspection_sigmoid.c.ll crash
darktable/introspection_demosaic.c.ll crash
darktable/introspection_censorize.c.ll crash
darktable/RawImageDataFloat.cpp.ll crash
darktable/introspection_colorequal.c.ll crash
darktable/introspection_highlights.c.ll crash
darktable/introspection_blurs.c.ll crash
darktable/introspection_soften.c.ll crash
darktable/RawImageDataU16.cpp.ll crash
darktable/avif.c.ll crash
darktable/introspection_primaries.c.ll crash
darktable/introspection_grain.c.ll crash
darktable/introspection_channelmixerrgb.c.ll crash
328db26 pre-commit: Update
27 27 bench/abseil-cpp/optimized/time_zone_format.cc.ll
2 2 bench/arrow/optimized/concatenate.cc.ll
9 9 bench/cmake/optimized/zstd_compress.c.ll
4 4 bench/cpython/optimized/unicodeobject.ll
68 68 bench/fmt/optimized/chrono-test.cc.ll
136 136 bench/gromacs/optimized/gmx_bar.cpp.ll
169 169 bench/gromacs/optimized/surfacearea.cpp.ll
1 1 bench/harfbuzz/optimized/harfbuzz.cc.ll
5 5 bench/hermes/optimized/BigIntSupport.cpp.ll
12 12 bench/libzmq/optimized/pipe.cpp.ll
7 5 bench/llvm/optimized/AArch64ISelLowering.cpp.ll
7 6 bench/llvm/optimized/X86ISelLowering.cpp.ll
36 36 bench/luau/optimized/isocline.c.ll
41 43 bench/ncnn/optimized/convolution_x86.cpp.ll
102 104 bench/ncnn/optimized/gemm_x86.cpp.ll
106 108 bench/ncnn/optimized/gemm_x86_avx.cpp.ll
73 73 bench/oiio/optimized/imagebufalgo.cpp.ll
58 58 bench/oiio/optimized/imagecache.cpp.ll
2 1 bench/opencv/optimized/stereobm.cpp.ll
125 120 bench/openjdk/optimized/DrawGlyphList.ll
69 69 bench/openusd/optimized/json.cpp.ll
19 19 bench/openusd/optimized/warped_motion.c.ll
3 3 bench/postgres/optimized/selfuncs.ll
96 96 bench/postgres/optimized/varlena.ll
54 55 bench/qemu/optimized/util_readline.c.ll
24 24 bench/spdlog/optimized/spdlog.cpp.ll
30 31 bench/stockfish/optimized/tbprobe.ll
12 20 bench/z3/optimized/mpz.cpp.ll
78 78 bench/zstd/optimized/zstd_compress.c.ll

nikic
nikic reviewed Nov 7, 2024
View reviewed changes
Copy link

@nikic nikic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are various minor regressions

bench/harfbuzz/optimized/harfbuzz.cc.ll
@@ -105368,7 +105368,7 @@ if.end: ; preds = %for.body
%tobool.not = icmp eq i32 %add14.i.i62, 0
%spec.select = tail call i32 @llvm.umax.i32(i32 %add14.i.i62, i32 1)
%inc = zext i1 %tobool.not to i32
%spec.select23 = add nuw nsw i32 %add14.i.i36, %inc
%spec.select23 = add i32 %add14.i.i36, %inc
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

flag inference regression

bench/arrow/optimized/concatenate.cc.ll
@@ -2929,7 +2929,7 @@ if.then.i.i.i.i279: ; preds = %if.else.i.i
_ZNKSt6vectorIN5arrow12_GLOBAL__N_15RangeESaIS2_EE12_M_check_lenEmPKc.exit.i.i.i: ; preds = %if.else.i.i
%sub.ptr.div.i.i.i.i.i = ashr exact i64 %sub.ptr.sub.i.i.i.i.i, 4
%.sroa.speculated.i.i.i.i = call i64 @llvm.umax.i64(i64 %sub.ptr.div.i.i.i.i.i, i64 1)
%add.i.i.i.i = add i64 %.sroa.speculated.i.i.i.i, %sub.ptr.div.i.i.i.i.i
%add.i.i.i.i = add nsw i64 %.sroa.speculated.i.i.i.i, %sub.ptr.div.i.i.i.i.i
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

flag inference improvement

bench/fmt/optimized/chrono-test.cc.ll
@@ -71721,10 +71721,10 @@ _ZNSt20back_insert_iteratorIN3fmt3v1019basic_memory_bufferIcLm500ESaIcEEEEaSEOc.
store i64 %inc.pre-phi.i.i47, ptr %size_.i.i, align 8
%arrayidx.i.i49 = getelementptr inbounds i8, ptr %24, i64 %23
store i8 46, ptr %arrayidx.i.i49, align 1
%.sroa.speculated193 = tail call i32 @llvm.umin.i32(i32 %precision, i32 %.sroa.speculated)
%.sroa.speculated193 = tail call i32 @llvm.smin.i32(i32 %precision, i32 %.sroa.speculated)
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

umin -> smin regression

bench/libzmq/optimized/pipe.cpp.ll
%cmp7 = icmp slt i32 %outhwm_, 1
%cmp10 = icmp eq i32 %1, 0
%or.cond = select i1 %cmp7, i1 true, i1 %cmp10
%out.0 = select i1 %or.cond, i32 0, i32 %add4
%3 = lshr i32 %2, 1
%div.i9 = select i1 %or.cond10, i32 0, i32 %3
%3 = sdiv i32 %2, 2
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

regression

bench/llvm/optimized/AArch64ISelLowering.cpp.ll
%108 = icmp ult i32 %107, 8
%106 = add nsw i8 %105, -8
%107 = call i8 @llvm.fshl.i8(i8 %106, i8 %106, i8 5)
%108 = icmp ult i8 %107, 8
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

improvement

bench/llvm/optimized/X86ISelLowering.cpp.ll
i32 4, label %287
i32 8, label %288
i32 16, label %289
%trunc = trunc nuw nsw i32 %.sroa.speculated to i8
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

improvement (I think?)

bench/ncnn/optimized/convolution_x86.cpp.ll
%69 = add nuw nsw i32 %68, 3
%70 = and i32 %69, -4
%.sroa.speculated = tail call i32 @llvm.umin.i32(i32 %70, i32 %.sroa.speculated52)
%66 = sdiv i32 %65, %.sroa.speculated52
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

regression (sdiv -> udiv no longer canonicalized)

bench/ncnn/optimized/gemm_x86_avx.cpp.ll

36: ; preds = %26
%37 = fmul fast float %17, 1.250000e-01
%38 = sitofp i32 %.sroa.speculated99 to float
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

regression (sitofp no longer canonicalized)

bench/opencv/optimized/stereobm.cpp.ll
@@ -1721,9 +1721,10 @@ _ZNK2cv11_InputArray6getMatEi.exit: ; preds = %187, %190

312: ; preds = %304
call void @_ZN2cv16ParallelLoopBodyD2Ev(ptr noundef nonnull align 8 dereferenceable(56) %37) #16
%.sroa.6.8.insert.ext = zext nneg i32 %234 to i64
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

regression

@nikic
Copy link

nikic commented Nov 25, 2024

/add-label regression
/close

@github-actions github-actions bot closed this Nov 25, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@nikic nikic nikic left review comments

Assignees
No one assigned
Labels
regression
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@dtcxzyw @nikic