Notes on Disruptor performance test in java

Source: Internet
Author: User

Hardware description:

Model name: Intel (R) Xeon (R) CPU E5-2640 0 @ 2.50 GHz

Cpu core: 4

Cache size: 15360 KB

OS: Linux 64bit

JDK: 1.7.0 _ 21-b11

Disruptor 3.2.1:

RingBufferSize: 1024

ProducerType: ProducerType. MULTI

WaitStrategy: BlockingWaitStrategy

Test code (dtest.zip common-disruptor ):

The code is as follows: Copy code

Package org. chinasb. test;
 
Import java. util. concurrent. CountDownLatch;
 
Import org. chinasb. common. disruptor. Event;
Import org. chinasb. common. disruptor. dispatch. Dispatcher;
Import org. chinasb. common. disruptor. dispatch. RingbufferDispatcher;
 
Public class Test {
Static long LOGIC_LOOP = 0;
Static long COST = 0;
Static int NUM = 5000000;
 
Public static void main (String [] args ){
Boolean multithread = false;
Int buffer size = 1024;
If (args. length> 0 ){
If (args. length> 0 ){
NUM = Integer. parseInt (args [0]);
            }
If (args. length> 1 ){
Multithread = Boolean. parseBoolean (args [1]);
            }
If (args. length> 2 ){
BufferSize = Integer. parseInt (args [2]);
            }
If (args. length> 3 ){
LOGIC_LOOP = Long. parseLong (args [3]);
            }
        }
Final Dispatcher dispatcher = new RingbufferDispatcher (multithread, "task executor", bufferSize );
Long start = System. nanoTime ();
Final CountDownLatch countDownLatch = new CountDownLatch (NUM );
Final int coreSize = Runtime. getRuntime (). availableProcessors ();
For (int I = 0; I <coreSize; I ++ ){
New Thread (new Runnable (){
@ Override
Public void run (){
For (int I = 0; I <NUM/coreSize; I ++ ){
Dispatcher. dispatch (Event. wrap (new Runnable (){
@ Override
Public void run (){
// Logic
Long start = System. nanoTime ();
For (long I = 0; I <LOGIC_LOOP; I ++ ){}
Double elapsed = (System. nanoTime ()-start)/1000.0;
COST + = elapsed;
CountDownLatch. countDown ();
                            }
}));
                    }
                }
}). Start ();
        }
Try {
CountDownLatch. await ();
Double elapsed = (System. nanoTime ()-start)/1000000.0;
Long throughput = (long) (NUM/(elapsed/1000.0 ));
System. out. println ("throughput:" + throughput + "/sec in" + (long) elapsed + "ms ");
System. out. println ("average logic:" + COST/NUM + "us ");
} Catch (InterruptedException e ){
E. printStackTrace ();
        }
Dispatcher. shutdown ();
    }
}

 

Test 1 description: the consumption capability of an individual consumer event in multi-producer mode (no time-consuming business logic, only test the event distribution throughput capability). Each type of operation is tested three times.

Test result 1:
Test 2 description: the consumption capability of a single consumer event in the multi-producer mode (with time-consuming business logic, the average time consumption of each event is 1.8 ms, and the throughput capability of event distribution and processing is tested ), each type of operation is tested three times.
Test result 2:
 
Conclusion: in the test scenario 1, when the number of events exceeds 1000, the event distribution throughput is stable at around. In the test scenario 2, the distribution of two different event quantity types is involved, its event distribution and processing throughput is stable at around 540.
Based on the test results, if a Disruptor is introduced in the game to replace the traditional BlockQueue for message event processing, the number of messages to be carried should be increased by about three times.
Related Article

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: info-contact@alibabacloud.com and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.