Bug ID: JDK-8355708 Two Float16 IR tests fail after JDK-8345125

Versions (Unresolved/Resolved/Fixed)

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 25
25Unresolved

The following two tests fail IR verification with "-ea -esa -XX:CompileThreshold=100 -XX:+UnlockExperimentalVMOptions -server -XX:-TieredCompilation" after JDK-8345125

compiler/c2/irTests/TestFloat16ScalarOperations.java

Failed IR Rules (1) of Methods (1)
----------------------------------
1) Method "public void TestFloat16ScalarOperations.testRounding2()" - [Failed IR rules: 1]:
   * @IR rule 2: "@compiler.lib.ir_framework.IR(phase={DEFAULT}, applyIfPlatformAnd={}, applyIfCPUFeatureOr={}, counts={"_#ADD_HF#_", " >0 ", "_#SUB_HF#_", " >0 ", "_#MUL_HF#_", " >0 ", "_#DIV_HF#_", " >0 ", "_#SQRT_HF#_", " >0 ", "_#FMA_HF#_", " >0 "}, failOn={}, applyIfPlatform={}, applyIfPlatformOr={}, applyIfOr={}, applyIfCPUFeatureAnd={"fphp", "true", "asimdhp", "true"}, applyIf={}, applyIfCPUFeature={}, applyIfAnd={}, applyIfNot={})"
     > Phase "PrintIdeal":
       - counts: Graph contains wrong number of nodes:
         * Constraint 1: "(\\d+(\\s){2}(AddHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!
         * Constraint 2: "(\\d+(\\s){2}(SubHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!
         * Constraint 3: "(\\d+(\\s){2}(MulHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!
         * Constraint 4: "(\\d+(\\s){2}(DivHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!
         * Constraint 5: "(\\d+(\\s){2}(SqrtHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!
         * Constraint 6: "(\\d+(\\s){2}(FmaHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 > 0 [given]
           - No nodes matched!

compiler/c2/irTests/MulHFNodeIdealizationTests.java

Failed IR Rules (1) of Methods (1)
----------------------------------
1) Method "public void compiler.c2.irTests.MulHFNodeIdealizationTests.test1()" - [Failed IR rules: 1]:
   * @IR rule 2: "@compiler.lib.ir_framework.IR(phase={DEFAULT}, applyIfPlatformAnd={}, applyIfCPUFeatureOr={}, counts={"_#ADD_HF#_", "1"}, failOn={"_#MUL_HF#_"}, applyIfPlatform={}, applyIfPlatformOr={}, applyIfOr={}, applyIfCPUFeatureAnd={"fphp", "true", "asimdhp", "true"}, applyIf={}, applyIfCPUFeature={}, applyIfAnd={}, applyIfNot={})"
     > Phase "PrintIdeal":
       - counts: Graph contains wrong number of nodes:
         * Constraint 1: "(\\d+(\\s){2}(AddHF.*)+(\\s){2}===.*)"
           - Failed comparison: [found] 0 = 1 [given]
           - No nodes matched!

[~jbhateja]Thanks for the suggestion. I have uploaded a patch for the same.
09-05-2025
A pull request was submitted for review. Branch: master URL: https://git.openjdk.org/jdk/pull/25141 Date: 2025-05-09 13:17:42 +0000
09-05-2025
[~bkilambi], Can you try by increasing the warmup for routines that are violating?
08-05-2025
Marking this as test bug for now. ILW = Tests fail IR verification (test bug), two tests at tier2 but problem listed, no workaround = MLH = P4
28-04-2025
Okay, I'll take care of it.
28-04-2025
I can do it but might take a little of time as it needs to go through some internal review. If it'd be quicker at your end, would you be able to do it please?
28-04-2025
Thanks, could you please problem list right-away? Otherwise, I can take care of the problem listing.
28-04-2025
It possibly failed on a machine without fp16 support. I'll look into this.
28-04-2025
Thanks, this is with an VM.Standard.A1.Flex (Altra processor from Ampere) on OCI. Could you please quickly problem list the affected (sub-) tests? We hit this a lot in our testing.
28-04-2025
Hi, I'll take a look. Can I ask what machine has this been tested on? These tests pass on N1/V1/V2 architectures.
28-04-2025
Looks like a regression from JDK-8345125.
28-04-2025