[PATCH] D68252: [Stats] Add ALWAYS_ENABLED_STATISTIC enabled regardless of LLVM_ENABLE_STATS.

Volodymyr Sapsai via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Sep 30 16:12:59 PDT 2019


vsapsai added a comment.

On release build with assertions disabled and LLVM_ENABLE_STATS being 0 the impact on the compile time is the following

  Metric: compile_time
  
  Program                                        output0jM4H6 output_KEgWj diff  
   test-suite...tTests/2003-05-07-VarArgs.test     0.03         0.02       -25.2%
   test-suite...sts/2003-05-31-CastToBool.test     0.02         0.02       22.1% 
   test-suite...ing-flt/Equivalencing-flt.test     0.61         0.73       20.7% 
   test-suite...s/2009-12-07-StructReturn.test     0.02         0.02       17.1% 
   test-suite...ts/2003-07-06-IntOverflow.test     0.02         0.02       17.0% 
   test-suite...ts/block-copied-in-cxxobj.test     0.03         0.03       16.2% 
   test-suite...e/UnitTests/ObjC/messages.test     0.03         0.04       15.9% 
   test-suite...r-algebra/kernels/2mm/2mm.test     0.08         0.06       -14.7%
   test-suite.../instance-method-metadata.test     0.37         0.42       14.0% 
   test-suite...ource/UnitTests/printargs.test     0.02         0.02       12.4% 
   test-suite...itTests/2003-05-26-Shorts.test     0.02         0.02       -12.2%
   test-suite...-06-20-StaticBitfieldInit.test     0.02         0.02       12.2% 
   test-suite...tTests/2002-05-03-NotTest.test     0.02         0.02       11.5% 
   test-suite...tTests/C++11/stdthreadbug.test     0.28         0.26       -10.3%
   test-suite...e/Benchmarks/Misc/flops-7.test     0.03         0.02       -10.2%
   Geomean difference                                                       nan% 
         output0jM4H6  output_KEgWj        diff
  count  542.000000    542.000000    424.000000
  mean   0.550511      0.547688     -0.001379  
  std    2.215555      2.205268      0.045026  
  min    0.000000      0.000000     -0.252280  
  25%    0.015875      0.016725     -0.021092  
  50%    0.040950      0.040000     -0.000875  
  75%    0.350750      0.355700      0.017304  
  max    30.343300     30.326700     0.220588

To me it looks that the change in compile time is within the noise.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D68252/new/

https://reviews.llvm.org/D68252





More information about the llvm-commits mailing list