Skip to content

Latest commit

 

History

History
109 lines (82 loc) · 2.75 KB

BENCHMARK.md

File metadata and controls

109 lines (82 loc) · 2.75 KB

Kew Performance Benchmark Results

System Information

Component Details
Operating System Darwin Darwin Kernel Version 24.1.0: Thu Nov 14 18:15:21 PST 2024; root:xnu-11215.41.3~13/RELEASE_ARM64_T6041
Python Version 3.12.5
Processor arm
CPU Cores 14
CPU Frequency 4.00MHz
Total Memory 48.0GB
Available Memory 16.0GB

System Performance

The following benchmarks were run with:

  • 100 total tasks
  • 10 concurrent workers
  • Multiple computational workloads

Overall Performance

  • Total Duration: 22.30 seconds
  • Overall Throughput: 4.48 tasks/second

Task-Specific Performance

Large Matrix Multiplication (1000x1000)

Metric Value
Average Latency 17.681s
Median Latency 17.696s
95th Percentile 18.040s
99th Percentile 18.049s
Throughput 1.12 tasks/second

Heavy Prime Calculation (up to 10M)

Metric Value
Average Latency 18.409s
Median Latency 18.390s
95th Percentile 18.734s
99th Percentile 18.743s
Throughput 1.12 tasks/second

Complex FFT (10M points)

Metric Value
Average Latency 16.209s
Median Latency 15.958s
95th Percentile 18.852s
99th Percentile 18.861s
Throughput 1.12 tasks/second

Distributed Data Processing

Metric Value
Average Latency 9.173s
Median Latency 9.349s
95th Percentile 12.984s
99th Percentile 12.985s
Throughput 1.12 tasks/second

Load Distribution

Testing load distribution across multiple queues with:

  • 3 queues
  • 5 workers per queue
  • 100 total tasks

Queue Performance

Queue: load_test_queue_0

  • Tasks processed: 73
  • Average execution time: 0.378s

Queue: load_test_queue_1

  • Tasks processed: 69
  • Average execution time: 0.382s

Queue: load_test_queue_2

  • Tasks processed: 66
  • Average execution time: 0.389s

Load Balance Score: 0.011s

(Lower score indicates better load distribution)

Benchmark run on 2025-01-10 17:19:50

Extended Performance Comparison

1. Heavy Workload Performance

Queue System Total Duration (s) Throughput (tasks/s) Recovery from Failures Data Persistence
Kew 22.30 4.48 Yes Yes
asyncio.Queue 43.71 4.58 No No
multiprocessing.Queue ERROR ERROR Limited No

Error Notes

  • multiprocessing.Queue benchmark failed with: None
  • This error occurs because multiprocessing cannot pickle async functions across process boundaries
  • This limitation demonstrates why distributed task queues like Kew are necessary for complex distributed workloads