simple ray python example cannot get it to execute in parallel

Question

Fairly new to ray, so please be patient. I created the following example to test the time difference between the ray and non-ray implementation of simple image arithmetics. Why is the following code not getting executed in parallel? I can see on the dashboard everything is loaded to one worker only

import ray
import cv2
import numpy as np
import time

ray.init(ignore_reinit_error=True)


@ray.remote
class image_import(object):
    def __init__(self):
        pass

    def color(self):
        return cv2.imread("frame30.png", flags=0)

    def black(self):
        return cv2.imread("frame30.png", flags=0)

    def green(self):
        return cv2.imread("frame30.png", flags=0)

    def yellow(self):
        return cv2.imread("frame30.png", flags=0)

    def blue(self):
        return cv2.imread("frame30.png", flags=0)
    
    def orange(self):
        return cv2.imread("frame30.png", flags=0)
    
    def multiply_matrices_1(self,clr,b_1):
        return np.dot(clr,b_1)
    
    def multiply_matrices_2(self,b_2,b_3):
        return np.dot(b_2,b_3)
    
    def multiply_matrices_3(self,b_4,b_5):
        return np.dot(b_4,b_5)
    
    def multi_1(self,val_1,val_2):
        return np.dot(val_1,val_2)
    
    def multi_2(self,val_2):
        return np.dot(val_2,val_2)
        
    def multi_3(self,val_4,val_5):
        return np.dot(val_4,val_5)

start = time.time()
counter_actor = image_import.remote()

clr= counter_actor.color.remote()
b_1= counter_actor.black.remote()
b_2= counter_actor.green.remote()
b_3= counter_actor.yellow.remote()
b_4= counter_actor.blue.remote()
b_5= counter_actor.orange.remote()


multi_1= counter_actor.multiply_matrices_1.remote(clr,b_1)
multi_2= counter_actor.multiply_matrices_2.remote(b_2,b_3)
multi_3= counter_actor.multiply_matrices_3.remote(b_3,b_4)

multi_4= counter_actor.multi_1.remote(multi_1,multi_2)
multi_5= counter_actor.multi_2.remote(multi_3)

addition= counter_actor.multi_3.remote(multi_4,multi_5)

results= ray.get(addition)
duration = time.time() - start
print(duration)

Kai · Accepted Answer · 2021-02-22 20:54:46Z

4

You are creating an actor (a remote object of your class), and actor tasks are executed serially:

https://docs.ray.io/en/master/walkthrough.html#calling-the-actor

For parallel execution just get rid of the class:

@ray.remote
def color(self):
    return cv2.imread("frame30.png", flags=0)

@ray.remote
def black(self):
    return cv2.imread("frame30.png", flags=0)

futures = [color.remote(), black.remote()]  # Will execute in parallel
ray.get(futures)

answered Feb 22, 2021 at 20:54

Kai

2611 silver badge3 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

PolarBear10 Over a year ago

thanks Kai! We noobs need that first push to get going :)

Collectives™ on Stack Overflow

simple ray python example cannot get it to execute in parallel

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related