JDK-8230078 : compiler/loopopts/superword/Vec_MulAddS2I.java is unexpectedly slow in windows
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 14,15
  • Priority: P4
  • Status: Closed
  • Resolution: Duplicate
  • Submitted: 2019-08-23
  • Updated: 2020-06-09
  • Resolved: 2020-06-09
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 15
15Resolved
Related Reports
Duplicate :  
JDK-8242310 - use reproducible random in hotspot compiler tests
Relates :  
JDK-8214751 - X86: Support for VNNI Instructions
Description
Test compiler/loopopts/superword/Vec_MulAddS2I.java time stat (on similar hosts):
tier1	macosx-x64-debug	41s	
tier1	linux-x64-debug	56s	
tier1	windows-x64-debug	5m 35s
Comments
oh, I somehow missed the outer loop which calls Vec_MulAddS2IImplement 5120k times.
09-06-2020
Fix changed test. See: https://hg.openjdk.java.net/jdk/jdk/rev/193e4179def8#l90.1 At least random is re-used and not created a lot of times. The only possible candidates are: 2020-04-08 22:58 iignatyev: 193e41 - OpenJDK 8242310 use reproducible random in hotspot compiler tests 2020-04-08 22:54 iignatyev: 5ab4f8 - OpenJDK 8241707 introduce randomness k/w to hotspot test suite 2020-04-08 22:58 It is possible to revert fix in test and compare duration if anyone is interested.
09-06-2020
JDK-8242310 didn't change the test (nor the product), j.u.Random is still used, and given that I don't think that the comparison w/ other platforms was fair.
09-06-2020
It seems using Random caused so significant performance degradation. Fixed by JDK-8242310.
09-06-2020
The test takes ~1m on the windows-debug for JDK 15 b22.
09-06-2020
[~jbhateja], any updates on this?
28-05-2020
Vivek, could you please have a look?
25-09-2019