Batched Throughput

Problem

Compute batching throughput. The number of batches is ceil(n_requests / batch_size), total time is batches * latency_ms, and throughput is n_requests per second. Return [batches, total_time_ms, throughput]; non-positive arguments return [0, 0.0, 0.0].

Hint

Larger batches raise throughput but add queueing delay for the first requests.

Test cases

case 1

in: [100,32,50]

expected: [4,200,500]

case 2

in: [10,10,100]

expected: [1,100,100]

case 3

in: [7,3,30]

expected: [3,90,77.77777777777779]

case 4

in: [0,5,10]

expected: [0,0,0]

Starter codePython

import math

def batched_throughput(n_requests, batch_size, latency_ms):
    # Returns [batches, total_time_ms, throughput]
    # Your code here
    pass

4 test cases · run your code against them in the editor