On Sat, 8 Jun 2024 23:30:38 GMT, Shaojin Wen <d...@openjdk.org> wrote:
>> After PR #16245, C2 optimizes stores into primitive arrays by combining
>> values into larger stores. In the UUID.toString method,
>> ByteArrayLittleEndian can be removed, making the code more elegant and
>> faster.
>
> Shaojin Wen has updated the pull request incrementally with one additional
> commit since the last revision:
>
> change method name, putHex -> putHex4, and fix comments
> I think we don't need to change them back everywhere, but only need to
> rewrite `ByteArrayLittleEndian` and `ByteArray` so that they no longer use
> `VarHandle`.
>
> Maybe I should rewrite #14636 without using `Unsafe`, so more people might
> agree with it.
You are right, ByteArray and ByteArrayLittleEndian have good performance after
removing Unsafe. This is similar to the previous version of java.io.Bits
class ByteArrayLittleEndian {
public static void setInt(byte[] array, int offset, int value) {
array[offset ] = (byte) value;
array[offset + 1] = (byte) (value >> 8);
array[offset + 2] = (byte) (value >> 16);
array[offset + 3] = (byte) (value >> 24);
}
}
class HexDigits {
public static void putHex4(byte[] array, int offset, int value) {
// Prepare an int value so C2 generates a 4-byte write instead of two
2-byte writes
ByteArrayLittleEndian.setInt(
array,
offset,
(DIGITS[value & 0xff] << 16) | DIGITS[(value >> 8) & 0xff]);
}
}
-------------
PR Comment: https://git.openjdk.org/jdk/pull/19610#issuecomment-2156255748
