8214761: Bug in parallel Kahan summation implementation #4674

igraves · 2021-07-02T20:12:39Z

Progress

Change must not contain extraneous whitespace
Commit message must refer to an issue
Change must be properly reviewed

Issue

JDK-8214761: Bug in parallel Kahan summation implementation

Reviewers

Joe Darcy (@jddarcy - Reviewer) ⚠️ Review applies to 10b8dcd

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/4674/head:pull/4674
$ git checkout pull/4674

Update a local copy of the PR:
$ git checkout pull/4674
$ git pull https://git.openjdk.java.net/jdk pull/4674/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 4674

View PR using the GUI difftool:
$ git pr show -t 4674

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/4674.diff

bridgekeeper · 2021-07-02T20:13:51Z

👋 Welcome back igraves! 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 · 2021-07-02T20:16:00Z

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

core-libs

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 · 2021-07-02T20:20:59Z

Webrevs

theRealAph · 2021-07-02T20:30:24Z

Crikey, how did we get that wrong?
It'd be nice if we had a regression test for this. Can you provide one, please?

stefan-zobel · 2021-07-02T21:30:56Z

src/java.base/share/classes/java/util/DoubleSummaryStatistics.java

-        sumWithCompensation(other.sumCompensation);
+
+        //Negating this value because low-order bits are in negated form
+        sumWithCompensation(-other.sumCompensation);


Wouldn't that be double tmp = sum - sumCompensation; in getSum() in line 246 too?

Good catch will review and make the change.

stefan-zobel · 2021-07-02T22:20:39Z

What about Collectors.computeFinalSum() - should this be double tmp = summands[0] + summands[1]; or double tmp = summands[0] - summands[1]; ?

stefan-zobel · 2021-07-02T22:39:13Z

Crikey, how did we get that wrong?
It'd be nice if we had a regression test for this. Can you provide one, please?

I found this: https://mail.openjdk.java.net/pipermail/core-libs-dev/2018-December/057239.html
Ivan Gerasimov already tackled this back then. His webrev still exists and it contains a simple regression test.

igraves · 2021-07-22T20:58:24Z

Circling back on this. I've worked in the test from Ivan Gerasimov's email back when. It includes some additional comparisons to prior approaches to assert improvements in error.

bridgekeeper · 2021-08-19T22:40:13Z

@igraves This pull request has been inactive for more than 4 weeks and will be automatically closed if another 4 weeks passes without any activity. To avoid this, simply add a new comment to the pull request. Feel free to ask for assistance if you need help with progressing this pull request towards integration!

jddarcy

The code changes look fine, but IMO the comments should be re-worded a bit.

Rather text like

// Negating this value because low-order bits are in negated form

I suggest something like

// Subtract compensation bits

The main compensation loop also subtracts the compensation bits. A comment like "subtract compensation bits" makes the corrected handling of them seem less anomalous.

openjdk · 2021-08-30T21:17:09Z

@igraves 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:

8214761: Bug in parallel Kahan summation implementation

Reviewed-by: darcy

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 14 new commits pushed to the master branch:

9689f61: 8272788: Nonleaf ranked locks should not be safepoint_check_never
4ee0dac: 8273248: ProblemList java/lang/instrument/BootClassPath/BootClassPathTest.sh on all configs
655ea6d: 8270489: Support archived heap objects in EpsilonGC
dacd197: 8273217: Make ParHeapInspectTask _safepoint_check_never
02822e1: 8272377: assert preconditions that are ensured when created in add_final_edges
a58cf16: 8272563: assert(is_double_stack() && !is_virtual()) failed: type check
0e14bf7: 8273176: handle latest VS2019 in abstract_vm_version
f1c5e26: 8273206: jdk/jfr/event/gc/collection/TestG1ParallelPhases.java fails after JDK-8159979
e600fe1: 8272618: Unnecessary Attr.visitIdent.noOuterThisPath
2fce7cb: 8272963: Update the java manpage markdown source
... and 4 more: https://git.openjdk.java.net/jdk/compare/683e30db79789ee44b3cc0b44c085de4615bca7b...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.

igraves · 2021-09-03T00:48:45Z

/integrate

openjdk · 2021-09-03T00:50:23Z

Going to push as commit dd87181.
Since your change was applied there have been 29 commits pushed to the master branch:

7fff22a: 8272805: Avoid looking up standard charsets
92b05fe: 8273251: Call check_possible_safepoint() from SafepointMechanism::process_if_requested()
29e0f13: 8272385: Enforce ECPrivateKey d value to be in the range [1, n-1] for SunEC provider
aaa6f69: 8273250: Address javadoc issues in Deflater::setDictionationary
5ee5dd9: 8272914: Create hotspot:tier2 and hotspot:tier3 test groups
5245c1c: 8273147: Update and restructure TestGCLogMessages log message list
632a7e0: 8273165: GraphKit::combine_exception_states fails with "matching stack sizes" assert
c2e015c: 8273229: Update OS detection code to recognize Windows Server 2022
0c1b16b: 8273243: Fix indentations in java.net.InetAddress methods
152e669: 8273140: Replace usages of Enum.class.getEnumConstants() with Enum.values() where possible
... and 19 more: https://git.openjdk.java.net/jdk/compare/683e30db79789ee44b3cc0b44c085de4615bca7b...master

Your commit was automatically rebased without conflicts.

openjdk · 2021-09-03T00:50:41Z

@igraves Pushed as commit dd87181.

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

8214761: Bug in parallel Kahan summation implementation

d7fc394

openjdk bot added core-libs core-libs-dev@openjdk.org rfr Pull request is ready for review labels Jul 2, 2021

stefan-zobel reviewed Jul 2, 2021

View reviewed changes

igraves added 3 commits July 12, 2021 15:43

Stashing

a35e6db

stashing

c13313c

Updates with more test coverage

10b8dcd

jddarcy approved these changes Aug 30, 2021

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Aug 30, 2021

igraves added 3 commits August 31, 2021 16:54

Merge branch 'master' into kahan-summation-bug

b2f4b35

Changing some comments.

905450d

Fixing compensation test

722826f

openjdk bot closed this Sep 3, 2021

openjdk bot added integrated Pull request has been integrated and removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Sep 3, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

8214761: Bug in parallel Kahan summation implementation #4674

8214761: Bug in parallel Kahan summation implementation #4674

igraves commented Jul 2, 2021 •

edited by openjdk bot

Loading

bridgekeeper bot commented Jul 2, 2021

openjdk bot commented Jul 2, 2021

mlbridge bot commented Jul 2, 2021 •

edited

Loading

theRealAph commented Jul 2, 2021

stefan-zobel Jul 2, 2021 •

edited

Loading

igraves Jul 6, 2021

stefan-zobel commented Jul 2, 2021

stefan-zobel commented Jul 2, 2021

igraves commented Jul 22, 2021

bridgekeeper bot commented Aug 19, 2021

jddarcy left a comment

openjdk bot commented Aug 30, 2021 •

edited

Loading

igraves commented Sep 3, 2021

openjdk bot commented Sep 3, 2021

openjdk bot commented Sep 3, 2021

8214761: Bug in parallel Kahan summation implementation #4674

8214761: Bug in parallel Kahan summation implementation #4674

Conversation

igraves commented Jul 2, 2021 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Jul 2, 2021

openjdk bot commented Jul 2, 2021

mlbridge bot commented Jul 2, 2021 • edited Loading

Webrevs

theRealAph commented Jul 2, 2021

stefan-zobel Jul 2, 2021 • edited Loading

Choose a reason for hiding this comment

igraves Jul 6, 2021

Choose a reason for hiding this comment

stefan-zobel commented Jul 2, 2021

stefan-zobel commented Jul 2, 2021

igraves commented Jul 22, 2021

bridgekeeper bot commented Aug 19, 2021

jddarcy left a comment

Choose a reason for hiding this comment

openjdk bot commented Aug 30, 2021 • edited Loading

igraves commented Sep 3, 2021

openjdk bot commented Sep 3, 2021

openjdk bot commented Sep 3, 2021

igraves commented Jul 2, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Jul 2, 2021 •

edited

Loading

stefan-zobel Jul 2, 2021 •

edited

Loading

openjdk bot commented Aug 30, 2021 •

edited

Loading