From: Instance segmentation on distributed deep learning big data cluster
Experiment (1) Machine Configuration: Ubuntu 20.04 VM with 7 GB RAM, Number processors = 2, Number cores each processor = 3, Python Version=3.7 Anaconda, Java Version = 1.8.0–201, PySpark Version = 2.4.6, spark.driver.memory = 4 GB, The number of cores = 2 | Local mode spark stand-up time | 17 s |
Total execution time for 18 images as one patch task job | The total time was 43 s, with 2.3 s allocated to process a single image (one frame). | |
Experiment (2) Machine Configuration: Ubuntu 20.04 VM with 7 GB RAM, Number processors = 2, Number cores each processor = 3, Python Version=3.7 Anaconda, Java Version=1.8.0-201, PySpark Version=2.4.6, spark.driver.memory = 4 GB, The number of cores = 4 | Local mode spark stand-up time | 14 s |
Total execution time for 26 images as one patch task job | The total time was 48 s, with 1.8 s dedicated to processing a single image (one frame). |