Improving AArch64 intrinsics

JEP 315 works by improving intrinsics on AArch64 processors. The current string and array intrinsic are improved. Also a new intrinsic is implemented for the sine, cosine, and logarithmic functions in java.lang.Math.

In order to enhance application performance, intrinsics utilize an assembly code that is specific to CPU architecture. It does not execute generic Java code.

Note that you will see AArch64 processors having an implementation of most of the intrinsics. However, JEP 315 implemented an optimized intrinsic for the following methods in the java.lang.Math class that was not up to the mark:

  • sin()
  • cos()
  • log()

It is also worth noting that some of the intrinsics that are previously implemented in the AArch64 port might not be completely optimal. Such intrinsic can take advantage of features such as memory address alignment or software prefetching instructions. Some of those methods are listed as follows:

  • String::compareTo
  • String::indexOf
  • StringCoding::hasNegatives
  • Arrays::equals
  • StringUTF16::compress
  • StringLatin1::inflate
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
52.14.151.45