Skip to content

Latest commit

 

History

History
55 lines (42 loc) · 2.15 KB

README.md

File metadata and controls

55 lines (42 loc) · 2.15 KB

What is the performance difference in R vs Python?

Setup

  1. Both scripts used the defaults as much as posible.
  2. Both scripts use as close to identical code as posible and are read in the normal fassion.
  3. The R script was devoloped in R Studio and the Python script in VS Code.
  4. The version of R was 3.6.3 64-bit while the version of Python was 3.7.7 64-bit with TensorFlow 2.1.
  5. Both scripts were run on an Intel Core i7-6700K with 64GB of Corsair Vengeance LPX DDR4 Ram and 2 NVIDIA GeForce GTX 1070 video cards (see note).
  6. Only the model training time was considered in results.
  7. Both scripts were run from the command line. 
    d:
cd d:\repos\MindMimicLabs\tensorflow-performance\r-vs-python
rscript simple_rnn.r
-- or --
python simple_rnn.py
  8. A fresh shell was opened between each run.
  9. Open Hardware Monitor was used to report the Max GPU and CPU

Results

Below can be found the results for the 2 different scripts.

NOTE: Both scripts were run WITH CPU ONLY because R requires Conda to enable GPU support despite the logging output saying "Adding visible gpu devices: 0, 1"

R

settings = batch_size/epochs/sample_length/units

settings Loss Time (min) Max CPU % Max GPU 1 % Max GPU 2 %
1024/10/11/40 5.770 47.13 34.7 0 10
0256/10/11/20 5.343 37.17 31.1 6 21
0128/10/11/20 4.750 36.52 86.7 0 11
0064/10/11/20 4.422 38.41 35.5 6 07
0064/10/11/10 4.886 38.25 33.1 0 12

Python

settings = batch_size/epochs/sample_length/units

settings Loss Time (min) Max CPU % Max GPU 1 % Max GPU 2 %
1024/10/11/40 6.203 00:05:44 78.5 0 05
0256/10/11/20 5.950 00:05:54 79.3 0 15
0128/10/11/20 5.397 00:06:19 85.7 0 03
0064/10/11/20 4.367 00:07:53 82.0 0 02
0064/10/11/10 4.800 00:07:13 83.2 0 17

Conclusion

There is some amount of loss in performance when using R. Waiting on another Lab mate to confirm results.