Calculate 5 summaries while running benchmarks

Now you can specify which summary you want when calling benchmark_result_df:

julia> benchmark_result_df(rb)
18×4 DataFrame
 Row │ Size   Library   Seconds      GFLOPS
     │ Int64  String    Float64      Float64
─────┼──────────────────────────────────────────
   1 │    20  BLIS      4.29e-5        0.37296
   2 │    45  BLIS      4.5419e-5      4.01264
   3 │   100  BLIS      9.7717e-5     20.4673
   4 │    20  Gaius     2.62847e-7    60.8719
   5 │    45  Gaius     0.000264588    0.688807
   6 │   100  Gaius     5.9732e-5     33.4829
   7 │    20  MKL       4.40768e-7    36.3003
   8 │    45  MKL       2.47089e-6    73.7589
   9 │   100  MKL       5.46717e-6   365.82
  10 │    20  Octavian  2.39237e-7    66.8793
  11 │    45  Octavian  2.17778e-6    83.6862
  12 │   100  Octavian  4.73643e-6   422.259
  13 │    20  OpenBLAS  7.06423e-7    22.6493
  14 │    45  OpenBLAS  4.98383e-6    36.5682
  15 │   100  OpenBLAS  1.443e-5     138.6
  16 │    20  Tullio    2.6291e-7     60.8574
  17 │    45  Tullio    2.16711e-6    84.0981
  18 │   100  Tullio    0.000125153   15.9804

julia> benchmark_result_df(rb, :median)
18×4 DataFrame
 Row │ Size   Library   Seconds      GFLOPS
     │ Int64  String    Float64      Float64
─────┼──────────────────────────────────────────
   1 │    20  BLIS      4.71605e-5     0.339267
   2 │    45  BLIS      4.9872e-5      3.65436
   3 │   100  BLIS      0.000125569   15.9274
   4 │    20  Gaius     2.64962e-7    60.3861
   5 │    45  Gaius     0.000304132    0.599246
   6 │   100  Gaius     0.000279572    7.15381
   7 │    20  MKL       4.44111e-7    36.027
   8 │    45  MKL       2.50067e-6    72.8806
   9 │   100  MKL       5.87317e-6   340.532
  10 │    20  Octavian  2.40778e-7    66.4513
  11 │    45  Octavian  2.20433e-6    82.6781
  12 │   100  Octavian  4.83843e-6   413.357
  13 │    20  OpenBLAS  7.12453e-7    22.4576
  14 │    45  OpenBLAS  5.137e-6      35.4779
  15 │   100  OpenBLAS  1.5909e-5    125.715
  16 │    20  Tullio    2.63827e-7    60.6457
  17 │    45  Tullio    2.21511e-6    82.2758
  18 │   100  Tullio    0.000191986   10.4174

or plot

julia> plot(rb, measure=:median)

As an aside, seems like plot should probably display the plot instead of just returning the files where it saves them. I added that as an option.

JuliaLinearAlgebra / BLASBenchmarksCPU.jl

Calculate 5 summaries while running benchmarks #42