gistfile1.txt

~/local/pytorch  (rec_fn_bench_models)$ PYTHONPATH="$(pwd)/benchmarks/experimental_components" python benchmarks/record_function_benchmark/record_function_bench.py
Benchmarking RecordFunction overhead for lstm_jit
Running warmup... finished
Benchmarking with RecordFunction, 1 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58c18>
Record function overhead: with_rec_fn
lstm_jit
  Median: 620.98 ms
  IQR:    26.02 ms (608.89 to 634.91)
  194 measurements, 1 runs per measurement, 1 thread

Benchmarking without RecordFunction, 1 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58c50>
Record function overhead: without_rec_fn
lstm_jit
  Median: 604.08 ms
  IQR:    20.82 ms (592.95 to 613.77)
  199 measurements, 1 runs per measurement, 1 thread

Benchmarking with RecordFunction, 2 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58b00>
Record function overhead: with_rec_fn
lstm_jit
  Median: 403.32 ms
  IQR:    14.65 ms (394.82 to 409.47)
  298 measurements, 1 runs per measurement, 2 threads

Benchmarking without RecordFunction, 2 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58c88>
Record function overhead: without_rec_fn
lstm_jit
  Median: 397.79 ms
  IQR:    14.74 ms (390.69 to 405.44)
  302 measurements, 1 runs per measurement, 2 threads

Benchmarking with RecordFunction, 4 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58e10>
Record function overhead: with_rec_fn
lstm_jit
  Median: 270.11 ms
  IQR:    8.21 ms (265.75 to 273.96)
  444 measurements, 1 runs per measurement, 4 threads

Benchmarking without RecordFunction, 4 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58860>
Record function overhead: without_rec_fn
lstm_jit
  Median: 266.64 ms
  IQR:    5.20 ms (264.13 to 269.33)
  449 measurements, 1 runs per measurement, 4 threads

Benchmarking with RecordFunction, 8 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58828>
Record function overhead: with_rec_fn
lstm_jit
  Median: 228.70 ms
  IQR:    4.04 ms (226.92 to 230.97)
  526 measurements, 1 runs per measurement, 8 threads

Benchmarking without RecordFunction, 8 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58438>
Record function overhead: without_rec_fn
lstm_jit
  Median: 223.99 ms
  IQR:    8.61 ms (218.91 to 227.52)
  539 measurements, 1 runs per measurement, 8 threads

Benchmarking with RecordFunction, 16 threads ... finished
<utils.common.Measurement object at 0x7fb23fb584a8>
Record function overhead: with_rec_fn
lstm_jit
  Median: 243.43 ms
  IQR:    15.63 ms (233.13 to 248.75)
  497 measurements, 1 runs per measurement, 16 threads

Benchmarking without RecordFunction, 16 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58390>
Record function overhead: without_rec_fn
lstm_jit
  Median: 238.64 ms
  IQR:    12.47 ms (233.67 to 246.14)
  501 measurements, 1 runs per measurement, 16 threads

Benchmarking with RecordFunction, 32 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58b38>
Record function overhead: with_rec_fn
lstm_jit
  Median: 335.79 ms
  IQR:    22.56 ms (324.11 to 346.67)
  358 measurements, 1 runs per measurement, 32 threads

Benchmarking without RecordFunction, 32 threads ... finished
<utils.common.Measurement object at 0x7fb23fb58a90>
Record function overhead: without_rec_fn
lstm_jit
  Median: 330.86 ms
  IQR:    15.63 ms (323.63 to 339.26)
  361 measurements, 1 runs per measurement, 32 threads

[--- Record function overhead --]
                      |  lstm_jit
1 threads: ----------------------
      with_rec_fn     |    620
      without_rec_fn  |    600
2 threads: ----------------------
      with_rec_fn     |    400
      without_rec_fn  |    400
4 threads: ----------------------
      with_rec_fn     |    270
      without_rec_fn  |    267
8 threads: ----------------------
      with_rec_fn     |    229
      without_rec_fn  |    220
16 threads: ---------------------
      with_rec_fn     |    240
      without_rec_fn  |    240
32 threads: ---------------------
      with_rec_fn     |    340
      without_rec_fn  |    330

Times are in milliseconds (ms).