Using JMH to find the fastest way to encode/decode UTF8

By

Overview

I had played with JMH before however this is the first time I have used JMH to solve a production problem.  I had some idea how to optimise the code involved but trying different combinations with JMH lead to a significant improvement.

In this test I am encoding a String as UTF8 to direct memory so it can be written to a TCP socket.  Also I need to take data written to direct memory via an NIO SocketChannel.read and encode it into a StringBuilder (which can be reused)

The tests

These tests involved a combination of using reflection to obtain the underlying data structure of String and StringBuilder but also using Unsafe to access the native memory.

To my surprise, access String via reflection appeared to be no faster, possibly slower. However @jponge pointed out that I needed to be returning a result from each benchamrk to avoid dead code elimination.  After I did this, the access via reflection was faster.

See encode_unsafeLoopCharAt  and encode_unsafeLoopCharAt

Another surprise was that accessing StringBuilder via reflection did make a difference.

See decode_usingSimpleLoop and decode_usingCharArray Accessing the underlying char[] was over 4x faster.  Note: this is clear an optimisation issue and future versions of Java might not have this problem.

All the results

The performance score is in operations per micro-second


 
 
 
 
 


  Benchmark
  i7-3790X 
  i7-4770K
  E5-2650v2
 



  DecodeMain.decode_fromUTF8
        7.3 
       6.8 
    6.3 
 



  DecodeMain.decode_usingCharArray
        8.4 
       8.1 
    8.8 
 



  DecodeMain.decode_usingCharArrayAndAddress
      12.7 
     11.8 
  11.3 
 



  DecodeMain.decode_usingSimpleLoop
        1.9
       1.9 
    1.8 
 



  EncodeMain.encode_simpleToUTF8
        6.5 
       6.6 
    5.9 
 



  EncodeMain.encode_unsafeLoopCharArray
      19.8 
     17.0 
  17.5 
 



  EncodeMain.encode_unsafeLoopCharAt
      15.3 
     12.0 
  13.7 
 



  EncodeMain.encode_unsafeLoopCharAtUnrolled
      15.3
     11.8
  13.0 
 



  EncodeMain.encode_usingSimpleLoop
        8.4 
       7.9 
    7.4 
 



  EncodeMain.encode_usingSimpleLoopUnrolled
        6.3 
       5.6 
    6.3 
 



 

Conclusion

In future this functionality might be built in to the JVM, however there is likely to be functionality which is not built in to the JVM which is causing a performance issues and having an alternative can make a difference.

Thank you

@jponge and @shiplev for your feedback.


Comments

Post a Comment

Popular posts from this blog

Java is Very Fast, If You Don’t Create Many Objects

By
Image
  You still have to watch how many objects you create. This article looks at a benchmark passing events over TCP/IP at 4 billion events per minute using the net.openhft.chronicle.wire.channel package in Chronicle Wire and why we still avoid object allocations..  One of the key optimisations is creating almost no garbage. Allocation is a very cheap operation and collection of very short-lived objects is also very cheap. Does this really make a difference? What difference does one small object per event (44 bytes) make to the performance in a throughput test where GC pauses are amortised? While allocation is as efficient as possible, it doesn’t avoid the memory pressure on the L1/L2 caches of your CPUs and when many cores are busy, they are contending for memory in the shared L3 cache.  Results Benchmark on a Ryzen 5950X with Ubuntu 22.10. JVM Vendor, Version No objects Throughput, Average Latency* One object per event Throughput, Average Latency* Azul Zulu 1.8.0_322 60.6 M event/s, 528
Read more

System wide unique nanosecond timestamps

By
A Unique Identifier can be very useful for tracing. Those ids are even more useful when they contain a high-resolution timestamp.  Not only do they record the time of an event, but if unique can help trace events as they pass through the system. Such unique timestamps however can be expensive depending on how they are implemented.   This post explores a lightweight means of producing a unique, monotonically increasing system-wide nano-second resolution timestamp available in our open-source library. Uses for Unique Identifiers Unique identifiers can be useful to associate with a piece of information so that information can be referred to later unambiguously. This could be an event, a request, an order id, or a customer id. They can naturally be used as a primary key in a database or key/value store to retrieve that information later. One of the challenges of generating these identifiers is avoiding creating duplicates while not having an increasing cost.  You could record every identi
Read more

Comparing Approaches to Durability in Low Latency Messaging Queues

By
Image
  A significant feature of Chronicle Queue Enterprise is support for TCP replication across multiple servers to ensure the high availability of application infrastructure. I generally believe that replicating data to a secondary system is faster than syncing to disk, assuming the round trip network delay wasn’t high due to quality networks and co-located redundant servers. This is the first time I have benchmarked it with a realistic example. Little’s Law and Why Latency Matters In many cases, the assumption is that the latency won't be a problem as long as throughput is high enough. However, latency is often a key factor in why the throughput isn’t high enough. Little’s law states, “ the long-term average number  L  of customers in a  stationary  system is equal to the long-term average effective arrival rate  λ  multiplied by the average time  W  that a customer spends in the system”. In computer terminology, the level of concurrency or parallelism a system has to support must be
Read more