JDK-8269260 : Add AVX512 and other SSE + AVX combinations testing for tests which generate vector instructions
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 11,17,18
  • Priority: P3
  • Status: Resolved
  • Resolution: Fixed
  • CPU: x86
  • Submitted: 2021-06-23
  • Updated: 2025-04-22
  • Resolved: 2021-06-25
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 11 JDK 17 JDK 18
11.0.28-oracleFixed 17 b29Fixed 18Fixed
Related Reports
Relates :  
Relates :  
Description
JDK-8269179 example shows that we don't test enough wide vectors on x64.
I suggest to create new HotSpot test group for such tests and together with jdk_vector group run them with different SSE + AVX HotSpot flags combinations. For now it is:
-XX:UseAVX=3
-XX:UseAVX=2
-XX:UseAVX=1
-XX:UseAVX=0
-XX:UseAVX=0 -XX:UseSSE=3
-XX:UseAVX=0 -XX:UseSSE=2 (this is minimal setting for 64 bit)

Comments
Changeset: 824a5169 Author: Vladimir Kozlov <kvn@openjdk.org> Date: 2021-06-25 22:48:56 +0000 URL: https://git.openjdk.java.net/jdk17/commit/824a51693e10afba834823efb38195ee0d692e5e
25-06-2021

ILW = missing testing, no workaround = MMH = P3
24-06-2021

jdk_vector (slow, will run in hs-tier3) jdk/incubator/vector Have to split some tests which are too slow to run in tier1. hotspot_vector_1 = \ compiler/c2/cr6340864 \ compiler/codegen \ compiler/loopopts/superword \ compiler/vectorapi \ compiler/vectorization \ -compiler/codegen/aes \ -compiler/codegen/Test6875866.java \ -compiler/codegen/Test6935535.java \ -compiler/codegen/TestGCMStorePlacement.java \ -compiler/codegen/TestTrichotomyExpressions.java \ -compiler/loopopts/superword/Vec_MulAddS2I.java \ -compiler/vectorapi/VectorRebracket128Test.java hotspot_vector_2 = \ compiler/intrinsics \ compiler/codegen/aes \ compiler/codegen/Test6875866.java \ compiler/codegen/Test6935535.java \ compiler/loopopts/superword/Vec_MulAddS2I.java \ compiler/vectorapi/VectorRebracket128Test.java \ - compiler/intrinsics/string/TestStringLatin1IndexOfChar.java
24-06-2021

I got internal comment that we can not run with flags in tier1. So here what I am suggesting to run internally: hs-tier2: hotspot_vector_1 - run with all flags combinations listed in Description hotspot_vector_2 - run with `-XX:UseAVX=3` only hs-tier3: jdk_vector - run with all flags combinations listed in Description hotspot_vector_2 - run with all combinations except `-XX:UseAVX=3`
24-06-2021