[8u] RFR for backport of 8216060 (CRC32 3/4): [PPC64] Vector CRC implementation should be used by interpreter and be faster for short arrays

Gustavo Romero gromero at linux.vnet.ibm.com
Fri Sep 27 03:32:14 UTC 2019


Hi,

Could the following backport be reviewed please?

Bug     : https://bugs.openjdk.java.net/browse/JDK-8216060
Original: http://hg.openjdk.java.net/jdk/jdk/rev/e17d7ffad9e3
Backport: http://cr.openjdk.java.net/~gromero/crc32_jdk8u/for-review/8216060/

It enhances performance for CRC32 computed over short arrays and also wires
up a faster CRC32 implementation to be used by the interpreter. It's also a
preparation to the last change in this patchset.

It's a PPC64-only change.

It was necessary to backport it to:

- Remove CRC32C code, basically removing is_crc32c in crc32(), and adapt callers
- Remove mentions to CRC32C in generate_CRC32_updateBytes() stub code mark
- Remove invertCRC when it's not necessary, like in
	a. kernel_crc32_1word()
	b. kernel_crc32_vpmsum()

Thank you.

Best regards,
Gustavo


More information about the jdk8u-dev mailing list