8286197: C2: Optimize MemorySegment shape in int loop by rwestrel · Pull Request #8555 · openjdk/jdk

rwestrel · 2022-05-05T14:57:11Z

This is another small enhancement for a code shape that showed up in a
MemorySegment micro benchmark. The shape to optimize is the one from test1:

for (int i = 0; i < size; i++) {
    long j = i * UNSAFE.ARRAY_INT_INDEX_SCALE;

    j = Objects.checkIndex(j, size * 4);

    if (((base + j) & 3) != 0) {
        throw new RuntimeException();
    }

    v += UNSAFE.getInt(base + j);
}

In that code shape, the loop iv is first scaled, result is then casted
to long, range checked and finally address of memory location is
computed.

The alignment check is transformed so the loop body has no check In
order to eliminate the range check, that loop is transformed into:

for (int i1 = ..) {
    for (int i2 = ..) {
        long j = (i1 + i2) * UNSAFE.ARRAY_INT_INDEX_SCALE;

        j = Objects.checkIndex(j, size * 4);

        v += UNSAFE.getInt(base + j);
    }
}

The address shape is (AddP base (CastLL (ConvI2L (LShiftI (AddI ...

In this case, the type of the ConvI2L is [min_jint, max_jint] and type
of CastLL is [0, max_jint] (the CastLL has a narrower type).

I propose transforming (CastLL (ConvI2L into (ConvI2L (CastII in that
case. The convI2L and CastII types can be set to [0, max_jint]. The
new address shape is then:

(AddP base (ConvI2L (CastII (LShiftI (AddI ...

which optimize well.

(LShiftI (AddI ...
is transformed into
(AddI (LShiftI ...
because one of the AddI input is loop invariant (i2) and we have:

(AddP base (ConvI2L (CastII (AddI (LShiftI ...

Then because the ConvI2L and CastII types are [0, max_jint], the AddI
is pushed through the ConvI2L and CastII:

(AddP base (AddL (ConvI2L (CastII (LShiftI ...

base and one of the inputs of the AddL are loop invariant so this
transformed into:

(AddP (AddP ...) (ConvI2L (CastII (LShiftI ...

The (AddP ...) is loop invariant so computed before entry. The
(ConvI2L ...) only depends on the loop iv.

The resulting address is a shift + an add. The address before
transformation requires 2 adds + a shift. Also after unrolling, the
adress of the second access in the loop is cheaper to compute as it
can be derived from the address of the first access.

For all of this to work:

I added a CastLL::Ideal transformation:
(CastLL (ConvI2L into (ConvI2l (CastII
I also had to prevent split if to transform (LShiftI (Phi for the
iv Phi of a counted loop.

test2 and test3 test 1) and 2) separately.

Progress

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

Issue

JDK-8286197: C2: Optimize MemorySegment shape in int loop

Reviewers

Vladimir Kozlov (@vnkozlov - Reviewer) ⚠️ Review applies to a122f0cf
Tobias Hartmann (@TobiHartmann - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/8555/head:pull/8555
$ git checkout pull/8555

Update a local copy of the PR:
$ git checkout pull/8555
$ git pull https://git.openjdk.org/jdk pull/8555/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 8555

View PR using the GUI difftool:
$ git pr show -t 8555

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/8555.diff

bridgekeeper · 2022-05-05T14:57:26Z

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

openjdk · 2022-05-05T15:00:03Z

@rwestrel The following label will be automatically applied to this pull request:

hotspot-compiler

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

mlbridge · 2022-05-05T15:05:47Z

Webrevs

vnkozlov

Good suggestion. I have comments.

vnkozlov · 2022-05-05T15:48:57Z

src/hotspot/share/opto/loopopts.cpp

-  // Check for having no control input; not pinned.  Allow
+// Check for having no control input; not pinned.  Allow


Wrong removed space.

fixed in new commit

vnkozlov · 2022-05-05T15:57:35Z

src/hotspot/share/opto/castnode.cpp

+  if (in1 != NULL && in1->Opcode() == Op_ConvI2L) {
+    const Type* t = Value(phase);
+    const Type* t_in = phase->type(in1);
+    if (t != Type::TOP && t_in != Type::TOP && t != t_in) {


t != t_in does not mean that type is narrower in general case. I think we need to check ranges (types meet?).

Thanks for looking at this. t is the result of Value() which takes the type of its input into account so, AFAICT, there's no way t can be wider than t_in. Am I missing something? If not I could add an assert. What do you think?

There is no specialized CastLLNode::Value() and ConstraintCastNode only calls filter_speculative() which do call join(). May be it is indeed enough. Yes, would be nice to have an assert to make sure we got it right.

The new commit adds an assert

rwestrel · 2022-05-05T16:04:58Z

Good suggestion. I have comments.

Thanks for reviewing this.

vnkozlov

Good. I will start testing.

openjdk · 2022-05-06T16:14:31Z

@rwestrel 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:

8286197: C2: Optimize MemorySegment shape in int loop

Reviewed-by: kvn, 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 588 new commits pushed to the master branch:

94b473e: 8280454: G1: ClassLoaderData verification keeps CLDs live that causes problems with VerifyDuringGC during Remark
900d967: 8287924: Avoid redundant HashMap.containsKey call in EnvHelp.mapToHashtable
d482d7f: 8286160: (fs) Files.exists returns unexpected results with C:\pagefile.sys because it's not readable
edff51e: 8284858: Start of release updates for JDK 20
2671443: 8286171: HttpClient/2 : Expect:100-Continue blocks indefinitely when response is not 100
59b0de6: 8288048: Build failure with GCC 6 after JDK-8286562
db4405d: 8288078: linux-aarch64-optimized build fails in Tier5 after JDK-8287567
7e948f7: 8287903: Reduce runtime of java.math microbenchmarks
3fa9984: 8287854: Dangling reference in ClassVerifier::verify_class
4285e3d: 8288023: AArch64: disable PAC-RET when preview is enabled
... and 578 more: https://git.openjdk.org/jdk/compare/29c2e54cf6fe472bd75a75fedf4ecf66e204647a...master

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 master branch, type /integrate in a new comment.

vnkozlov · 2022-05-07T00:18:23Z

Testing passed. It needs second review.

rwestrel · 2022-05-30T14:35:33Z

Any one else for this one?

TobiHartmann

Looks good to me but I'm wondering if we should delay this to JDK 20 as we are late for the JDK 19 release and had many issues with Cast/ConvNodes before.

src/hotspot/share/opto/castnode.cpp

src/hotspot/share/opto/castnode.hpp

test/hotspot/jtreg/compiler/c2/irTests/TestConvI2LCastLongLoop.java

Co-authored-by: Tobias Hartmann <tobias.hartmann@oracle.com>

….java Co-authored-by: Tobias Hartmann <tobias.hartmann@oracle.com>

rwestrel · 2022-06-07T14:38:55Z

Looks good to me but I'm wondering if we should delay this to JDK 20 as we are late for the JDK 19 release and had many issues with Cast/ConvNodes before.

Thanks for the review. I have no objection to delaying until JDK 20.

rwestrel · 2022-06-10T08:16:24Z

/integrate

openjdk · 2022-06-10T08:17:40Z

Going to push as commit dae4c49.
Since your change was applied there have been 588 commits pushed to the master branch:

94b473e: 8280454: G1: ClassLoaderData verification keeps CLDs live that causes problems with VerifyDuringGC during Remark
900d967: 8287924: Avoid redundant HashMap.containsKey call in EnvHelp.mapToHashtable
d482d7f: 8286160: (fs) Files.exists returns unexpected results with C:\pagefile.sys because it's not readable
edff51e: 8284858: Start of release updates for JDK 20
2671443: 8286171: HttpClient/2 : Expect:100-Continue blocks indefinitely when response is not 100
59b0de6: 8288048: Build failure with GCC 6 after JDK-8286562
db4405d: 8288078: linux-aarch64-optimized build fails in Tier5 after JDK-8287567
7e948f7: 8287903: Reduce runtime of java.math microbenchmarks
3fa9984: 8287854: Dangling reference in ClassVerifier::verify_class
4285e3d: 8288023: AArch64: disable PAC-RET when preview is enabled
... and 578 more: https://git.openjdk.org/jdk/compare/29c2e54cf6fe472bd75a75fedf4ecf66e204647a...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-06-10T08:17:56Z

@rwestrel Pushed as commit dae4c49.

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

test & fix

75bdea6

extra test comment

4b90672

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label May 5, 2022

whitespaces

6493429

openjdk bot added the rfr Pull request is ready for review label May 5, 2022

vnkozlov reviewed May 5, 2022

View reviewed changes

review

a122f0c

vnkozlov approved these changes May 6, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label May 6, 2022

TobiHartmann approved these changes Jun 7, 2022

View reviewed changes

src/hotspot/share/opto/castnode.cpp Outdated Show resolved Hide resolved

src/hotspot/share/opto/castnode.hpp Outdated Show resolved Hide resolved

test/hotspot/jtreg/compiler/c2/irTests/TestConvI2LCastLongLoop.java Show resolved Hide resolved

rwestrel and others added 3 commits June 7, 2022 16:37

Update src/hotspot/share/opto/castnode.cpp

2d78533

Co-authored-by: Tobias Hartmann <tobias.hartmann@oracle.com>

Update src/hotspot/share/opto/castnode.hpp

6f7e00c

Co-authored-by: Tobias Hartmann <tobias.hartmann@oracle.com>

Update test/hotspot/jtreg/compiler/c2/irTests/TestConvI2LCastLongLoop…

8261097

….java Co-authored-by: Tobias Hartmann <tobias.hartmann@oracle.com>

TobiHartmann approved these changes Jun 10, 2022

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Jun 10, 2022

openjdk bot closed this Jun 10, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Jun 10, 2022

		// Check for having no control input; not pinned. Allow
		// Check for having no control input; not pinned. Allow

Conversation

rwestrel commented May 5, 2022 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Uh oh!

bridgekeeper bot commented May 5, 2022

Uh oh!

openjdk bot commented May 5, 2022

Uh oh!

mlbridge bot commented May 5, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

vnkozlov May 5, 2022

Choose a reason for hiding this comment

Uh oh!

rwestrel May 6, 2022

Choose a reason for hiding this comment

Uh oh!

vnkozlov May 5, 2022

Choose a reason for hiding this comment

Uh oh!

rwestrel May 5, 2022

Choose a reason for hiding this comment

Uh oh!

vnkozlov May 5, 2022

Choose a reason for hiding this comment

Uh oh!

rwestrel May 6, 2022

Choose a reason for hiding this comment

Uh oh!

rwestrel commented May 5, 2022

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented May 6, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

vnkozlov commented May 7, 2022

Uh oh!

rwestrel commented May 30, 2022

Uh oh!

TobiHartmann left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rwestrel commented Jun 7, 2022

Uh oh!

rwestrel commented Jun 10, 2022

Uh oh!

openjdk bot commented Jun 10, 2022

Uh oh!

openjdk bot commented Jun 10, 2022

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

3 participants

rwestrel commented May 5, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented May 5, 2022 •

edited

Loading

openjdk bot commented May 6, 2022 •

edited

Loading