Loading...

XML

Word

Printable

Type: Enhancement
Resolution: Fixed
Priority: P4
Fix Version/s: 22
Affects Version/s: 21, 22
Component/s: hotspot
Labels:

Subcomponent:
compiler
Resolved In Build:
b02
CPU:

x86
OS:

generic

Issue	Fix Version	Assignee	Priority	Status	Resolution	Resolved In Build
JDK-8352136	21.0.8-oracle	Tobias Hartmann	P4	Resolved	Fixed	b01

Adding an intrinsic for float/double modulus operations for x86. This addresses a performance degradation and removes dependence on x87 floating point for platforms that support AVX.

- - Sort By Name
  - Sort By Date
  - Ascending
  - Descending
  - Thumbnails
  - List
  - Download All

time.openjdk21-gcc1221fix
0.9 kB
2023-05-26 08:53
time.openjdk21-intrinsic
2 kB
2023-05-26 08:55

backported by

JDK-8352136 Add intrinsic for float/double modulo for x86 AVX2 and AVX512

Resolved

caused by

JDK-8302191 Performance degradation for float/double modulo on Linux

Closed

causes

JDK-8312233 Performance regression in SharedRuntime::frem/drem() on x86 with AVX2 after JDK-8308966

Closed

JDK-8314056 Remove runtime platform check from frem/drem

Resolved

relates to

JDK-8305689 Consider adding an intrinsic for StrictMath.IEEEremainder

Open

JDK-8309636 Optimize double drem for Blender.java

Open

JDK-8305948 Performance degradation for float/double modulo on Linux

Resolved

JDK-8302191 Performance degradation for float/double modulo on Linux

Closed

links to

Commit openjdk/jdk/5d5ae352

Review openjdk/jdk/14224

Review(master) openjdk/jdk21u-dev/1668

(3 relates to, 3 links to)

Assignee:: Scott Gibbons (Inactive)

Reporter:: Scott Gibbons (Inactive)

Votes:: 0 Vote for this issue

Watchers:: 6 Start watching this issue

Created:: 2023-05-26 07:55

Updated:: 2 days ago 02:46

Resolved:: 2023-06-12 08:09

Details

Backports

Description

Attachments

Attachments

Issue Links

Activity

People

Dates