8376544: [lworld] Use profiling around acmp to speculate that one operand is a value object

[marc-chevalier]

To do acmp, if the pointer comparison fails, we check nullity of operands, whether operands have the same class, and if this class is a value class. And if everything works, we call isSubstitutable.

Currently, when we have profiling information, we use it to speculate that operands are null or of an identity class (in which case, pointer comparison is all we should do). This patch proposes to take advantage of the profiling when it hints that operands are value objects. By speculating that at least one operand is of a given value class, the call to isSubstitutable can be intrinsified later, and thus, spare the Java call.

Speculating on one operand is enough since we will check that operands are of the same type.

Some relevant microbenchmark in valhalla.acmp.array.Value032 (in ns/op (lower is better)):

	branch_obj_equals000	branch_val_equals000	branch_obj_equals025	branch_val_equals025
Before	4.361 ± 0.153 ns/op	0.907 ± 0.015 ns/op	3.332 ± 0.040 ns/op	0.845 ± 0.032 ns/op
With this fix	0.911 ± 0.019 ns/op	0.899 ± 0.021 ns/op	0.838 ± 0.023 ns/op	0.806 ± 0.012 ns/op

Full results of micro:valhalla.acmp.array.Value032

In more details, before:

Benchmark                                       Mode  Cnt  Score   Error  Units
Value032.branch_obj_equals000                   avgt   15  4.087 ± 0.122  ns/op
Value032.branch_obj_equals025                   avgt   15  3.117 ± 0.067  ns/op
Value032.branch_obj_equals050                   avgt   15  2.235 ± 0.083  ns/op
Value032.branch_obj_equals075                   avgt   15  1.292 ± 0.077  ns/op
Value032.branch_obj_equals100                   avgt   15  0.316 ± 0.004  ns/op
Value032.branch_val_equals000                   avgt   15  0.891 ± 0.024  ns/op
Value032.branch_val_equals025                   avgt   15  0.839 ± 0.033  ns/op
Value032.branch_val_equals050                   avgt   15  0.869 ± 0.087  ns/op
Value032.branch_val_equals075                   avgt   15  0.816 ± 0.023  ns/op
Value032.branch_val_equals100                   avgt   15  0.906 ± 0.032  ns/op
Value032.result_obj_equals000                   avgt   15  4.115 ± 0.185  ns/op
Value032.result_obj_equals025                   avgt   15  3.326 ± 0.346  ns/op
Value032.result_obj_equals050                   avgt   15  2.353 ± 0.516  ns/op
Value032.result_obj_equals075                   avgt   15  1.226 ± 0.040  ns/op
Value032.result_obj_equals100                   avgt   15  0.326 ± 0.006  ns/op
Value032.result_val_equals000                   avgt   15  0.888 ± 0.025  ns/op
Value032.result_val_equals025                   avgt   15  0.824 ± 0.028  ns/op
Value032.result_val_equals050                   avgt   15  0.891 ± 0.039  ns/op
Value032.result_val_equals075                   avgt   15  0.892 ± 0.005  ns/op
Value032.result_val_equals100                   avgt   15  0.908 ± 0.021  ns/op
Value032NullFree.branch_val_equals000           avgt   15  0.186 ± 0.012  ns/op
Value032NullFree.branch_val_equals025           avgt   15  0.461 ± 0.012  ns/op
Value032NullFree.branch_val_equals050           avgt   15  0.468 ± 0.015  ns/op
Value032NullFree.branch_val_equals075           avgt   15  0.465 ± 0.014  ns/op
Value032NullFree.branch_val_equals100           avgt   15  0.189 ± 0.007  ns/op
Value032NullFree.result_val_equals000           avgt   15  0.184 ± 0.004  ns/op
Value032NullFree.result_val_equals025           avgt   15  0.302 ± 0.005  ns/op
Value032NullFree.result_val_equals050           avgt   15  0.448 ± 0.014  ns/op
Value032NullFree.result_val_equals075           avgt   15  0.577 ± 0.014  ns/op
Value032NullFree.result_val_equals100           avgt   15  0.183 ± 0.004  ns/op
Value032NullFreeNonAtomic.branch_val_equals000  avgt   15  0.182 ± 0.006  ns/op
Value032NullFreeNonAtomic.branch_val_equals025  avgt   15  0.470 ± 0.015  ns/op
Value032NullFreeNonAtomic.branch_val_equals050  avgt   15  0.474 ± 0.016  ns/op
Value032NullFreeNonAtomic.branch_val_equals075  avgt   15  0.468 ± 0.017  ns/op
Value032NullFreeNonAtomic.branch_val_equals100  avgt   15  0.188 ± 0.009  ns/op
Value032NullFreeNonAtomic.result_val_equals000  avgt   15  0.190 ± 0.007  ns/op
Value032NullFreeNonAtomic.result_val_equals025  avgt   15  0.322 ± 0.017  ns/op
Value032NullFreeNonAtomic.result_val_equals050  avgt   15  0.473 ± 0.020  ns/op
Value032NullFreeNonAtomic.result_val_equals075  avgt   15  0.592 ± 0.026  ns/op
Value032NullFreeNonAtomic.result_val_equals100  avgt   15  0.189 ± 0.017  ns/op

After:

Benchmark                                       Mode  Cnt  Score   Error  Units
Value032.branch_obj_equals000                   avgt   15  0.967 ± 0.039  ns/op
Value032.branch_obj_equals025                   avgt   15  0.890 ± 0.032  ns/op
Value032.branch_obj_equals050                   avgt   15  0.640 ± 0.019  ns/op
Value032.branch_obj_equals075                   avgt   15  0.466 ± 0.022  ns/op
Value032.branch_obj_equals100                   avgt   15  0.320 ± 0.006  ns/op
Value032.branch_val_equals000                   avgt   15  0.913 ± 0.025  ns/op
Value032.branch_val_equals025                   avgt   15  0.843 ± 0.029  ns/op
Value032.branch_val_equals050                   avgt   15  0.818 ± 0.019  ns/op
Value032.branch_val_equals075                   avgt   15  0.850 ± 0.018  ns/op
Value032.branch_val_equals100                   avgt   15  0.893 ± 0.018  ns/op
Value032.result_obj_equals000                   avgt   15  0.927 ± 0.034  ns/op
Value032.result_obj_equals025                   avgt   15  0.865 ± 0.031  ns/op
Value032.result_obj_equals050                   avgt   15  0.632 ± 0.024  ns/op
Value032.result_obj_equals075                   avgt   15  0.470 ± 0.015  ns/op
Value032.result_obj_equals100                   avgt   15  0.320 ± 0.007  ns/op
Value032.result_val_equals000                   avgt   15  0.889 ± 0.021  ns/op
Value032.result_val_equals025                   avgt   15  0.811 ± 0.023  ns/op
Value032.result_val_equals050                   avgt   15  0.890 ± 0.025  ns/op
Value032.result_val_equals075                   avgt   15  0.923 ± 0.036  ns/op
Value032.result_val_equals100                   avgt   15  0.896 ± 0.027  ns/op
Value032NullFree.branch_val_equals000           avgt   15  0.184 ± 0.006  ns/op
Value032NullFree.branch_val_equals025           avgt   15  0.462 ± 0.015  ns/op
Value032NullFree.branch_val_equals050           avgt   15  0.460 ± 0.027  ns/op
Value032NullFree.branch_val_equals075           avgt   15  0.446 ± 0.014  ns/op
Value032NullFree.branch_val_equals100           avgt   15  0.183 ± 0.006  ns/op
Value032NullFree.result_val_equals000           avgt   15  0.188 ± 0.006  ns/op
Value032NullFree.result_val_equals025           avgt   15  0.317 ± 0.009  ns/op
Value032NullFree.result_val_equals050           avgt   15  0.461 ± 0.012  ns/op
Value032NullFree.result_val_equals075           avgt   15  0.584 ± 0.017  ns/op
Value032NullFree.result_val_equals100           avgt   15  0.181 ± 0.005  ns/op
Value032NullFreeNonAtomic.branch_val_equals000  avgt   15  0.184 ± 0.006  ns/op
Value032NullFreeNonAtomic.branch_val_equals025  avgt   15  0.465 ± 0.019  ns/op
Value032NullFreeNonAtomic.branch_val_equals050  avgt   15  0.451 ± 0.018  ns/op
Value032NullFreeNonAtomic.branch_val_equals075  avgt   15  0.468 ± 0.017  ns/op
Value032NullFreeNonAtomic.branch_val_equals100  avgt   15  0.187 ± 0.009  ns/op
Value032NullFreeNonAtomic.result_val_equals000  avgt   15  0.184 ± 0.007  ns/op
Value032NullFreeNonAtomic.result_val_equals025  avgt   15  0.315 ± 0.011  ns/op
Value032NullFreeNonAtomic.result_val_equals050  avgt   15  0.464 ± 0.021  ns/op
Value032NullFreeNonAtomic.result_val_equals075  avgt   15  0.581 ± 0.018  ns/op
Value032NullFreeNonAtomic.result_val_equals100  avgt   15  0.179 ± 0.002  ns/op

It is same or better.

Tested successfully with tier1-4,stress,valhalla-stress

Thanks,
Marc

---

Progress

• Change must not contain extraneous whitespace
• Commit message must refer to an issue
• Change must be properly reviewed (1 review required, with at least 1 Committer)

Issue

• JDK-8376544: [lworld] Use profiling around acmp to speculate that one operand is a value object (Bug - P3)

Reviewers

• Tobias Hartmann (@TobiHartmann - Committer)

Contributors

• Tobias Hartmann <thartmann@openjdk.org>

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/valhalla.git pull/2237/head:pull/2237
$ git checkout pull/2237

Update a local copy of the PR:
$ git checkout pull/2237
$ git pull https://git.openjdk.org/valhalla.git pull/2237/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 2237

View PR using the GUI difftool:
$ git pr show -t 2237

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/valhalla/pull/2237.diff

Using Webrev

Link to Webrev Comment

[marc-chevalier]

/contributor add @TobiHartmann

For the first prototype, and significant refactoring.

[bridgekeeper]

👋 Welcome back mchevalier! A progress list of the required criteria for merging this PR into lworld will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

[openjdk]

@marc-chevalier This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8376544: [lworld] Use profiling around acmp to speculate that one operand is a value object

Co-authored-by: Tobias Hartmann <thartmann@openjdk.org>
Reviewed-by: thartmann

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 16 new commits pushed to the lworld branch:

• 2d66adb: 8380015: [lworld] runtime/valhalla/inlinetypes/field_layout/ValueRandomLayoutTest.java divides by zero during statistic printing
• 0ca8597: 8343835: [lworld] C2: assert fails with "no mismatched stores, except on raw memory: StoreD StoreI"
• 632fc4f: 8379333: [lworld] [REDO] JNI NewWeakGlobalRef should throw IdentityException for value types
• ... and 13 more: https://git.openjdk.org/valhalla/compare/186c02e81371521b479f17ea7065b52669b1b2bb...lworld

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the lworld branch, type /integrate in a new comment.

[openjdk]

@marc-chevalier
Contributor Tobias Hartmann <thartmann@openjdk.org> successfully added.

[marc-chevalier]

After 8368988: [lworld] Rename classes in valhalla/inlinetypes to be distinct, the class in TestNewAcmp.java was copy-pasted here. I'm just redoing the fix as it makes my IDE complain again.

[mlbridge]

Webrevs

• 01: Full - Incremental (3873dbfa)
• 00: Full (728d2220)

[TobiHartmann]

Looks good to me otherwise. Thanks for working on this!

[TobiHartmann]

Still good. Thanks!

[marc-chevalier]

/integrate

Thanks for review!

[openjdk]

Going to push as commit 067e6d8.
Since your change was applied there have been 17 commits pushed to the lworld branch:

• 19fe8ba: 8379863: [lworld] C2: assert(!value->obj_is_scalar_replaced() || realloc_failures) failed: object should be reallocated already
• 2d66adb: 8380015: [lworld] runtime/valhalla/inlinetypes/field_layout/ValueRandomLayoutTest.java divides by zero during statistic printing
• 0ca8597: 8343835: [lworld] C2: assert fails with "no mismatched stores, except on raw memory: StoreD StoreI"
• ... and 14 more: https://git.openjdk.org/valhalla/compare/186c02e81371521b479f17ea7065b52669b1b2bb...lworld

Your commit was automatically rebased without conflicts.

[openjdk]

@marc-chevalier Pushed as commit 067e6d8.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.