8351994: Enable Extended EVEX to REX2/REX demotion when src and dst are the same #24431

vamsi-parasa · 2025-04-04T01:15:36Z

Intel APX NDD instructions are encoded using EVEX encoding. The goal of this PR is to enable optimized instruction encoding for Intel APX NDD instructions when the non-destructive destination is same as the first source.

For example:

eaddl r18, r18, r25 can be encoded as addl r18, r25 using APX REX2 encoding
eaddl r2, r2, r7 can be encoded as addl r2, r7 using non-APX legacy encoding

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-8351994: Enable Extended EVEX to REX2/REX demotion when src and dst are the same (Sub-task - P4)

Reviewers

Sandhya Viswanathan (@sviswa7 - Reviewer)
Jatin Bhateja (@jatin-bhateja - Reviewer)
Emanuel Peter (@eme64 - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/24431/head:pull/24431
$ git checkout pull/24431

Update a local copy of the PR:
$ git checkout pull/24431
$ git pull https://git.openjdk.org/jdk.git pull/24431/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 24431

View PR using the GUI difftool:
$ git pr show -t 24431

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/24431.diff

Using Webrev

Link to Webrev Comment

…re the same

bridgekeeper · 2025-04-04T01:16:29Z

👋 Welcome back sparasa! 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 · 2025-04-04T01:17:19Z

@vamsi-parasa 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:

8351994: Enable Extended EVEX to REX2/REX demotion when src and dst are the same

Reviewed-by: sviswanathan, jbhateja, epeter

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 669 new commits pushed to the master branch:

a3f9e22: 8358218: Problemlist jdk/incubator/vector/PreferredSpeciesTest.java#id0
84002d1: 8228773: URLClassLoader constructors should include API note warning that the parent should not be null
3a3ea7e: 8357598: Toolkit.removeAWTEventListener should handle null listener in AWTEventListenerProxy
... and 666 more: https://git.openjdk.org/jdk/compare/2ec61f0f34d2fa313ae3ed5d6aa0881ec091b347...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 (@sviswa7, @jatin-bhateja, @eme64) 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).

openjdk · 2025-04-04T01:17:57Z

@vamsi-parasa 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 · 2025-04-04T01:21:34Z

Webrevs

src/hotspot/cpu/x86/assembler_x86.cpp

sviswa7 · 2025-04-09T23:15:12Z

src/hotspot/cpu/x86/assembler_x86.cpp

  }
 }

+bool Assembler::is_demotable(bool no_flags, int dst_enc, int nds_enc, int src_enc) {


src_enc is not being used in this method so could be removed.

Please see the src_enc removed in the updated code.

sviswa7 · 2025-04-09T23:16:30Z

src/hotspot/cpu/x86/assembler_x86.cpp

  InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
  // Encoding Format : eevex_prefix (4 bytes) | opcode_cc | modrm
-  int encode =  evex_prefix_and_encode_ndd(0, 0, dst->encoding(), VEX_SIMD_F2, /* MAP4 */VEX_OPCODE_0F_3C, &attributes);
+  int encode =  evex_prefix_and_encode_ndd(0, 0, dst->encoding(), VEX_SIMD_F2, /* MAP4 */VEX_OPCODE_0F_3C, &attributes); //TODO: check this


This should not be demoted.

Please see the demotion disabled for esetzucc instruction in the updated code.

sviswa7 · 2025-04-09T23:18:09Z

src/hotspot/cpu/x86/assembler_x86.cpp

+                           InstructionAttr *attributes, bool no_flags, bool use_prefixq) {
+  // Demote RegRegReg instructions
+  if (!no_flags && dst_enc == nds_enc) {
+    return use_prefixq? prefixq_and_encode(dst_enc, src_enc) : prefix_and_encode(dst_enc, src_enc);


Nit pick, need space before ? as below:
use_prefixq ? prefixq_and_encode

Please see the space fixed in the updated code.

sviswa7 · 2025-04-09T23:19:51Z

src/hotspot/cpu/x86/assembler_x86.cpp

-                           InstructionAttr *attributes, bool no_flags) {
+                           InstructionAttr *attributes, bool no_flags, bool use_prefixq) {
+  // Demote RegRegReg instructions
+  if (!no_flags && dst_enc == nds_enc) {


This could be replaced by call to is_demotable().

Please see the updated code refactored to use is_demotable().

sviswa7 · 2025-04-09T23:20:59Z

src/hotspot/cpu/x86/assembler_x86.cpp

+int Assembler::evex_prefix_and_encode_ndd(int dst_enc, int nds_enc, VexSimdPrefix pre, VexOpcode opc,
+                           InstructionAttr *attributes, bool no_flags, bool use_prefixq) {
+  // Demote RegReg and RegRegImm instructions
+  if (!no_flags && dst_enc == nds_enc) {


This could be replaced by call to is_demotable().

src/hotspot/cpu/x86/assembler_x86.cpp

openjdk · 2025-04-16T22:51:47Z

@vamsi-parasa this pull request can not be integrated into master due to one or more merge conflicts. To resolve these merge conflicts and update this pull request you can run the following commands in the local repository for your personal fork:

git checkout demotion-jdk
git fetch https://git.openjdk.org/jdk.git master
git merge FETCH_HEAD
# resolve conflicts and follow the instructions given by git merge
git commit -m "Merge master"
git push

jatin-bhateja · 2025-04-17T11:55:40Z

Hi Vamsi,
We should extend following peephole optimization rules for LEA detection to cover new NDD patterns.

https://github.com/openjdk/jdk/blob/master/src/hotspot/cpu/x86/x86_64.ad#L14590
https://github.com/openjdk/jdk/blob/master/src/hotspot/cpu/x86/x86_64.ad#L14599
https://github.com/openjdk/jdk/blob/master/src/hotspot/cpu/x86/x86_64.ad#L14608

I have created a follow up JBS for this
https://bugs.openjdk.org/browse/JDK-8354939

vamsi-parasa · 2025-04-28T15:31:18Z

Hi Sandhya (@sviswa7) and Jatin (@jatin-bhateja),

Could you please review the refactored changes?

Thanks,
Vamsi

sviswa7 · 2025-04-29T18:23:30Z

src/hotspot/cpu/x86/assembler_x86.cpp

  InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ true, /* uses_vl */ false);
-  int encode = evex_prefix_and_encode_ndd(src1->encoding(), dst->encoding(), src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_3C, &attributes);
-  emit_int16((0x40 | cc), (0xC0 | encode));
+  evex_opcode_and_int16_ndd(dst->encoding(), src1->encoding(), src2->encoding(), VEX_SIMD_NONE, VEX_OPCODE_0F_3C, &attributes, 0x40 | cc, 0xC0, false);


This could be instead call to:
evex_opcode_int16_ndd(dst->encoding(), src2->encoding(), src1->encoding(), ..)
We can remove the separate method evex_opcode_and_int16_ndd() method altogether.

Actually, I tried implementing using only one function but due to the argument ordering changes in the inner functions, it was not possible to do it using one high level function like evex_opcode_int16_ndd.

Co-authored-by: Jatin Bhateja <jatin.bhateja@intel.com>

vamsi-parasa · 2025-05-22T20:44:59Z

Hi Jatin (@jatin-bhateja),

Incorporated the changes suggested for cpu_family and is_P6_or_later() and other minor changes. Please let me know if everything looks good.

Thanks,
Vamsi

sviswa7

Updates look good.

vamsi-parasa · 2025-05-23T15:33:48Z

@vamsi-parasa Testing launched, ping me again in 24h :)

Thanks Emanuel (@eme64)! Please let me know if there're are any issues with the tests.

eme64 · 2025-05-26T05:52:07Z

@vamsi-parasa Testing looked good, though now you pushed some more changes. I'd like to run tests one more time before integration. Please let me know when you are ready :)

vamsi-parasa · 2025-05-26T22:57:11Z

@vamsi-parasa Testing looked good, though now you pushed some more changes. I'd like to run tests one more time before integration. Please let me know when you are ready :)

Hi Emanuel (@eme64),

Thanks for the update! The new changes got approved and are ready for testing.
Could you please launch the tests?

Thanks,
Vamsi

eme64 · 2025-05-27T06:43:33Z

@vamsi-parasa Launched!

vamsi-parasa · 2025-05-28T20:54:04Z

@vamsi-parasa Launched!

Hi Emanuel (@eme64),

Could you pls let me know when the testing is completed?
Will integrate it if everything looks good.

Thanks,
Vamsi

sviswa7 · 2025-05-30T18:06:30Z

We plan to integrate this PR on Monday June 2nd so as to not get very close to the upcoming fork next Thursday.

eme64 · 2025-05-31T16:20:10Z

@vamsi-parasa @sviswa7 Testing passed!
We are celebrating ascension weekend, hence we may be a little less responsive than usual ;)

eme64

Looks reasonable to me, thanks for the work!

sviswa7 · 2025-05-31T16:28:45Z

@vamsi-parasa @sviswa7 Testing passed! We are celebrating ascension weekend, hence we may be a little less responsive than usual ;)

Thanks a lot @eme64. Really appreciate all your help.

vamsi-parasa · 2025-05-31T21:39:20Z

@vamsi-parasa @sviswa7 Testing passed! We are celebrating ascension weekend, hence we may be a little less responsive than usual ;)

Thank you, Emanuel! :)

vamsi-parasa · 2025-05-31T21:39:30Z

/integrate

openjdk · 2025-05-31T21:40:06Z

@vamsi-parasa
Your change (at version 3378eae) is now ready to be sponsored by a Committer.

sviswa7 · 2025-05-31T23:07:32Z

/sponsor

openjdk · 2025-05-31T23:07:58Z

Going to push as commit fc3d3d9.
Since your change was applied there have been 669 commits pushed to the master branch:

a3f9e22: 8358218: Problemlist jdk/incubator/vector/PreferredSpeciesTest.java#id0
84002d1: 8228773: URLClassLoader constructors should include API note warning that the parent should not be null
3a3ea7e: 8357598: Toolkit.removeAWTEventListener should handle null listener in AWTEventListenerProxy
... and 666 more: https://git.openjdk.org/jdk/compare/2ec61f0f34d2fa313ae3ed5d6aa0881ec091b347...master

Your commit was automatically rebased without conflicts.

openjdk · 2025-05-31T23:08:06Z

@sviswa7 @vamsi-parasa Pushed as commit fc3d3d9.

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

8351994: Enable Extended EVEX to REX2/REX demotion when src and dst a…

6a7ca6b

…re the same

openjdk bot added the rfr Pull request is ready for review label Apr 4, 2025

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Apr 4, 2025

jatin-bhateja reviewed Apr 4, 2025

View reviewed changes

src/hotspot/cpu/x86/assembler_x86.cpp Show resolved Hide resolved

sviswa7 reviewed Apr 9, 2025

View reviewed changes

Disable demotion for esetzucc and cleanup code

bd84608

sviswa7 reviewed Apr 11, 2025

View reviewed changes

src/hotspot/cpu/x86/assembler_x86.cpp Outdated Show resolved Hide resolved

vamsi-parasa added 2 commits April 14, 2025 11:21

refactor esetzucc to do encoding without a demote flag

09d3c3a

refactor RRM instructions to avoid explicit demotion

bc69f32

openjdk bot added the merge-conflict Pull request has merge conflict with target branch label Apr 16, 2025

vamsi-parasa added 2 commits April 16, 2025 16:24

RRM refactoring works for exorb and exorw

1e666ba

RRM refactorted to use a unified evex_ndd_and_int8 function

3b5d985

vamsi-parasa added 5 commits April 17, 2025 14:56

refactor APX NDD shift instrcutions to do demotion internally

ca750e3

refactor imul instructions to fold demotion logic inside

1fa0fbe

fix emit_arith_ndd discrepancy

c2cd9c7

rename demote flag to optimize_rax_dst in emit_arith

e065333

merge master

a494e42

openjdk bot removed the merge-conflict Pull request has merge conflict with target branch label Apr 23, 2025

vamsi-parasa added 3 commits April 23, 2025 18:08

WIP: cleanup refactoring

ac53cd1

refactoring done for ecmov RRR and others

2768cc5

cleanup ecmov, eorw and other refactoring

6a01e74

sviswa7 reviewed Apr 29, 2025

View reviewed changes

refactor emit_arith and evex_prefix_int8_operand

9291083

vamsi-parasa and others added 2 commits May 22, 2025 12:48

Update src/hotspot/cpu/x86/assembler_x86.cpp

131f11a

Co-authored-by: Jatin Bhateja <jatin.bhateja@intel.com>

Update src/hotspot/cpu/x86/assembler_x86.cpp

c6718a1

Co-authored-by: Jatin Bhateja <jatin.bhateja@intel.com>

openjdk bot removed the ready Pull request is ready to be integrated label May 22, 2025

refactor is_P6_or_later and remove cpu_family==18

3378eae

sviswa7 approved these changes May 22, 2025

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label May 22, 2025

jatin-bhateja approved these changes May 23, 2025

View reviewed changes

eme64 approved these changes May 31, 2025

View reviewed changes

openjdk bot added the sponsor Pull request is ready to be sponsored label May 31, 2025

openjdk bot added the integrated Pull request has been integrated label May 31, 2025

openjdk bot closed this May 31, 2025

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review sponsor Pull request is ready to be sponsored labels May 31, 2025

jatin-bhateja mentioned this pull request Aug 4, 2025

8351016: RA support for EVEX to REX/REX2 demotion to optimize NDD instructions #26283

Open

3 tasks

8351994: Enable Extended EVEX to REX2/REX demotion when src and dst are the same #24431

8351994: Enable Extended EVEX to REX2/REX demotion when src and dst are the same #24431

Uh oh!

Conversation

vamsi-parasa commented Apr 4, 2025 • 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 Apr 4, 2025

Uh oh!

openjdk bot commented Apr 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

openjdk bot commented Apr 4, 2025

Uh oh!

mlbridge bot commented Apr 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

openjdk bot commented Apr 16, 2025

Uh oh!

jatin-bhateja commented Apr 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

vamsi-parasa commented Apr 28, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

vamsi-parasa commented May 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sviswa7 left a comment

Choose a reason for hiding this comment

Uh oh!

vamsi-parasa commented May 23, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eme64 commented May 26, 2025

Uh oh!

vamsi-parasa commented May 26, 2025

Uh oh!

eme64 commented May 27, 2025

Uh oh!

vamsi-parasa commented May 28, 2025

Uh oh!

sviswa7 commented May 30, 2025

Uh oh!

eme64 commented May 31, 2025

Uh oh!

eme64 left a comment

Choose a reason for hiding this comment

Uh oh!

sviswa7 commented May 31, 2025

Uh oh!

vamsi-parasa commented May 31, 2025

vamsi-parasa commented Apr 4, 2025 •

edited by openjdk bot

Loading

openjdk bot commented Apr 4, 2025 •

edited

Loading

mlbridge bot commented Apr 4, 2025 •

edited

Loading

jatin-bhateja commented Apr 17, 2025 •

edited

Loading

vamsi-parasa commented May 22, 2025 •

edited

Loading

vamsi-parasa commented May 23, 2025 •

edited

Loading