8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long #9068

merykitty · 2022-06-07T17:14:18Z

Hi,

This patch implements intrinsics for Integer/Long::compareUnsigned using the same approach as the JVM does for long and floating-point comparisons. This allows efficient and reliable usage of unsigned comparison in Java, which is a basic operation and is important for range checks such as discussed in #8620 .

Thank you very much.

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-8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long

Reviewers

Vladimir Kozlov (@vnkozlov - Reviewer) ⚠️ Review applies to b5627135
Jatin Bhateja (@jatin-bhateja - Committer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/9068/head:pull/9068
$ git checkout pull/9068

Update a local copy of the PR:
$ git checkout pull/9068
$ git pull https://git.openjdk.org/jdk pull/9068/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 9068

View PR using the GUI difftool:
$ git pr show -t 9068

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/9068.diff

bridgekeeper · 2022-06-07T17:14:39Z

👋 Welcome back merykitty! 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 · 2022-06-07T17:17:11Z

@merykitty The following labels will be automatically applied to this pull request:

core-libs
hotspot

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

merykitty · 2022-06-07T17:19:27Z

/label hotspot-compiler

mlbridge · 2022-06-07T17:19:43Z

Webrevs

openjdk · 2022-06-07T17:22:02Z

@merykitty
The hotspot-compiler label was successfully added.

vnkozlov

Please add microbenchmark and show its results.

src/hotspot/share/opto/subnode.hpp

merykitty · 2022-06-08T09:39:04Z

I have added a benchmark for the intrinsic. The result is as follows, thanks a lot:

                                                Before          After
Benchmark                 (size)  Mode  Cnt  Score   Error  Score   Error  Units
Integers.compareUnsigned     500  avgt   15  0.527 ± 0.002  0.498 ± 0.011  us/op
Longs.compareUnsigned        500  avgt   15  0.677 ± 0.014  0.561 ± 0.006  us/op

vnkozlov

Good. I submitted testing.
You need second review.

openjdk · 2022-06-08T17:54:40Z

@merykitty 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:

8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long

Reviewed-by: kvn, jbhateja

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 321 new commits pushed to the master branch:

b96ba19: 8289182: NMT: MemTracker::baseline should return void
779b4e1: 8287001: Add warning message when fail to load hsdis libraries
910053b: 8280235: Deprecated flag FlightRecorder missing from VMDeprecatedOptions test
7b3bf97: 8289401: Add dump output to TestRawRSACipher.java
86dc760: Merge
1504804: 8289398: ProblemList jdk/jfr/api/consumer/recordingstream/TestOnEvent.java on linux-x64 again
9b7805e: 8289069: Very slow C1 arraycopy jcstress tests after JDK-8279886
c42b796: 8288058: Broken links on constant-values page
a814293: 8275784: Bogus warning generated for record with compact constructor
6f9717b: 8288836: (fs) Files.writeString spec for IOException has "specified charset" when no charset is provided
... and 311 more: https://git.openjdk.org/jdk/compare/645be42f76b8983a9096ed90caa70b5c59dd822c...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.

As you do not have Committer status in this project an existing Committer must agree to sponsor your change. Possible candidates are the reviewers of this PR (@vnkozlov, @jatin-bhateja) but any other Committer may sponsor as well.

➡️ To flag this PR as ready for integration with the above commit message, type /integrate in a new comment. (Afterwards, your sponsor types /sponsor in a new comment to perform the integration).

vnkozlov · 2022-06-08T23:54:14Z

Tier1-4 testing passed - no new failures. I suggest to push it into JDK 20 after fork and after you get second review.

sviswa7 · 2022-06-10T00:05:53Z

@merykitty Could you please also add the micro benchmark where compareUnsigned result is stored directly in an integer and show the performance of that?

merykitty · 2022-06-22T02:57:43Z

Thanks @sviswa7 for the suggestion, the results of getting the value of compareUnsigned directly is as follow:

                                                       Before          After
Benchmark                       (size)  Mode  Cnt   Score   Error  Score   Error  Units
Integers.compareUnsignedDirect     500  avgt   15   0.639 ± 0.022  0.626 ± 0.002  us/op
Longs.compareUnsignedDirect        500  avgt   15   0.672 ± 0.011  0.609 ± 0.004  us/op

jatin-bhateja · 2022-06-28T05:20:03Z

src/hotspot/cpu/x86/x86_64.ad

+    __ cmpl($src1$$Register, $src2$$Register);
+    __ movl($dst$$Register, -1);
+    __ jccb(Assembler::below, done);


By placing compare adjacent to conditional jump in-order frontend can trigger macro-fusion.
Kindly refer section 3.4.2.2 of Intel's optimization manual.

I realised that by swapping the mov and the cmp instruction, the rule needs to have dst different from src1 and src2, which increases register pressure.

I do not follow your comment, allocation decisions purely based on LRGs interferences and data flow attributes attached to operands and is agnostic to encoding block contents.

Your suggestion requires us having additional TEMP dst for the match rule. Thanks.

Yes, macro fusion is a fine microarchitectural optimization which can reduce load on entire execution pipeline and is deterministic for specific pair of cmp + jump instructions, you have aggregated destination's defs and its usages towards the tail which can save TEMP attribution on destination operand and may save a redundant spill only for high register pressure blocks. I am ok with existing handling.

Thanks for your explanations.

jatin-bhateja · 2022-06-28T05:20:10Z

src/hotspot/cpu/x86/x86_64.ad

+    __ cmpq($src1$$Register, $src2$$Register);
+    __ movl($dst$$Register, -1);
+    __ jccb(Assembler::below, done);


Same as above.

jatin-bhateja · 2022-06-28T05:51:42Z

src/hotspot/cpu/x86/x86_64.ad

@@ -13022,6 +13022,32 @@ instruct testL_reg_mem2(rFlagsReg cr, rRegP src, memory mem, immL0 zero)
  ins_pipe(ialu_cr_reg_mem);
 %}

+// Manifest a CmpU result in an integer register.  Very painful.
+// This is the test to avoid.
+instruct cmpU3_reg_reg(rRegI dst, rRegI src1, rRegI src2, rFlagsReg flags)


Do you plan to add 32 bit support?
Integer pattern can be moved to common file x86.ad and 64 pattern can handled in 32/64 bit AD files.

Yes I will add support for 32-bit after this patch, basic rules are often put in the bit-specific ad file so I think it would be more preferable to follow that convention here.

jatin-bhateja · 2022-06-28T05:57:09Z

src/hotspot/share/opto/subnode.hpp

+    // Since it is not consumed by Bools, it is not really a Cmp.
+    init_class_id(Class_Sub);
+  }
+  virtual int Opcode() const;


In-lining may connect the inputs to constant, hence a Value routine may be useful here.

CmpU3 inherits the Value method from its superclass CmpU

Its fine then.

jatin-bhateja · 2022-06-28T05:58:42Z

src/hotspot/share/opto/subnode.hpp

+    init_class_id(Class_Sub);
+  }
+  virtual int Opcode() const;
+  virtual uint ideal_reg() const { return Op_RegI; }


Value routine to handle constant folding.

merykitty · 2022-06-28T12:43:29Z

@jatin-bhateja Thanks a lot for your reviews and suggestions, I have answered your comments.

merykitty · 2022-06-29T07:16:50Z

Thank you very much for your reviews
/integrate

openjdk · 2022-06-29T07:17:45Z

@merykitty
Your change (at version 0ab881a) is now ready to be sponsored by a Committer.

jatin-bhateja · 2022-06-29T10:32:47Z

/sponsor

openjdk · 2022-06-29T10:34:21Z

Going to push as commit 108cd69.
Since your change was applied there have been 321 commits pushed to the master branch:

b96ba19: 8289182: NMT: MemTracker::baseline should return void
779b4e1: 8287001: Add warning message when fail to load hsdis libraries
910053b: 8280235: Deprecated flag FlightRecorder missing from VMDeprecatedOptions test
7b3bf97: 8289401: Add dump output to TestRawRSACipher.java
86dc760: Merge
1504804: 8289398: ProblemList jdk/jfr/api/consumer/recordingstream/TestOnEvent.java on linux-x64 again
9b7805e: 8289069: Very slow C1 arraycopy jcstress tests after JDK-8279886
c42b796: 8288058: Broken links on constant-values page
a814293: 8275784: Bogus warning generated for record with compact constructor
6f9717b: 8288836: (fs) Files.writeString spec for IOException has "specified charset" when no charset is provided
... and 311 more: https://git.openjdk.org/jdk/compare/645be42f76b8983a9096ed90caa70b5c59dd822c...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-06-29T10:34:42Z

@jatin-bhateja @merykitty Pushed as commit 108cd69.

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

merykitty added 2 commits June 7, 2022 22:57

compare unsigned

69b4fc1

format

01c0a07

openjdk bot added the rfr label Jun 7, 2022

openjdk bot added hotspot core-libs labels Jun 7, 2022

merykitty changed the title ~~8283726: x86_64 intrinsics for compare method in Integer and Long~~ 8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long Jun 7, 2022

openjdk bot added the hotspot-compiler label Jun 7, 2022

vnkozlov reviewed Jun 7, 2022

View reviewed changes

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

merykitty added 2 commits June 8, 2022 17:26

review comments

2bd0c78

remove comments

b562713

vnkozlov approved these changes Jun 8, 2022

View reviewed changes

openjdk bot added the ready label Jun 8, 2022

add comparison for direct value of compare

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature.

GPG key ID: B5690EEEBB952194

Verified
Learn about vigilant mode

0ab881a

jatin-bhateja reviewed Jun 28, 2022

View reviewed changes

jatin-bhateja approved these changes Jun 29, 2022

View reviewed changes

openjdk bot added the sponsor label Jun 29, 2022

openjdk bot added the integrated label Jun 29, 2022

openjdk bot closed this Jun 29, 2022

openjdk bot removed ready rfr sponsor labels Jun 29, 2022

merykitty deleted the compareUnsigned branch August 12, 2022 19:22

shqking mentioned this pull request Dec 1, 2022

8287925: AArch64: intrinsics for compareUnsigned method in Integer and Long #11383

Closed

3 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long #9068

8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long #9068

merykitty commented Jun 7, 2022 •

edited by openjdk bot

Loading

bridgekeeper bot commented Jun 7, 2022

openjdk bot commented Jun 7, 2022

merykitty commented Jun 7, 2022

mlbridge bot commented Jun 7, 2022 •

edited

Loading

openjdk bot commented Jun 7, 2022

vnkozlov left a comment

merykitty commented Jun 8, 2022

vnkozlov left a comment

openjdk bot commented Jun 8, 2022 •

edited

Loading

vnkozlov commented Jun 8, 2022

sviswa7 commented Jun 10, 2022

merykitty commented Jun 22, 2022

jatin-bhateja Jun 28, 2022

merykitty Jun 28, 2022

jatin-bhateja Jun 29, 2022

merykitty Jun 29, 2022

jatin-bhateja Jun 29, 2022 •

edited

Loading

jatin-bhateja Jun 28, 2022

jatin-bhateja Jun 28, 2022

merykitty Jun 28, 2022

jatin-bhateja Jun 28, 2022

merykitty Jun 28, 2022

jatin-bhateja Jun 28, 2022

jatin-bhateja Jun 28, 2022

merykitty commented Jun 28, 2022

merykitty commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

jatin-bhateja commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long #9068

8283726: x86_64 intrinsics for compareUnsigned method in Integer and Long #9068

Conversation

merykitty commented Jun 7, 2022 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Jun 7, 2022

openjdk bot commented Jun 7, 2022

merykitty commented Jun 7, 2022

mlbridge bot commented Jun 7, 2022 • edited Loading

Webrevs

openjdk bot commented Jun 7, 2022

vnkozlov left a comment

Choose a reason for hiding this comment

merykitty commented Jun 8, 2022

vnkozlov left a comment

Choose a reason for hiding this comment

openjdk bot commented Jun 8, 2022 • edited Loading

vnkozlov commented Jun 8, 2022

sviswa7 commented Jun 10, 2022

merykitty commented Jun 22, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jatin-bhateja Jun 29, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

merykitty commented Jun 28, 2022

merykitty commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

jatin-bhateja commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

openjdk bot commented Jun 29, 2022

merykitty commented Jun 7, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented Jun 7, 2022 •

edited

Loading

openjdk bot commented Jun 8, 2022 •

edited

Loading

jatin-bhateja Jun 29, 2022 •

edited

Loading