> 8214761: Bug in parallel Kahan summation implementation
Ian Graves has updated the pull request incrementally with one additional
commit since the last revision:
Fixing compensation test
-
Changes:
- all: https://git.openjdk.java.net/jdk/pull/4674/files
- new: https://git.o
> 8214761: Bug in parallel Kahan summation implementation
Ian Graves has updated the pull request with a new target base due to a merge
or a rebase. The incremental webrev excludes the unrelated changes brought in
by the merge/rebase. The pull request contains six additional commits since the
l
On Wed, 21 Jul 2021 20:19:31 GMT, Ian Graves wrote:
>> 8214761: Bug in parallel Kahan summation implementation
>
> Ian Graves has updated the pull request incrementally with three additional
> commits since the last revision:
>
> - Updates with more test coverage
> - stashing
> - Stashing
T
On Wed, 21 Jul 2021 20:19:31 GMT, Ian Graves wrote:
>> 8214761: Bug in parallel Kahan summation implementation
>
> Ian Graves has updated the pull request incrementally with three additional
> commits since the last revision:
>
> - Updates with more test coverage
> - stashing
> - Stashing
C
> 8214761: Bug in parallel Kahan summation implementation
Ian Graves has updated the pull request incrementally with three additional
commits since the last revision:
- Updates with more test coverage
- stashing
- Stashing
-
Changes:
- all: https://git.openjdk.java.net/jdk/pul
On Fri, 2 Jul 2021 21:30:56 GMT, stefan-zobel
wrote:
>> 8214761: Bug in parallel Kahan summation implementation
>
> src/java.base/share/classes/java/util/DoubleSummaryStatistics.java line 161:
>
>> 159:
>> 160: //Negating this value because low-order bits are in negated form
>> 161:
On Fri, 2 Jul 2021 20:30:24 GMT, Andrew Haley wrote:
> 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 tac
On Fri, 2 Jul 2021 20:12:39 GMT, Ian Graves wrote:
> 8214761: Bug in parallel Kahan summation implementation
What about `Collectors.computeFinalSum()` - should this be `double tmp =
summands[0] + summands[1];` or `double tmp = summands[0] - summands[1];` ?
-
PR: https://git.openjd
On Fri, 2 Jul 2021 20:12:39 GMT, Ian Graves wrote:
> 8214761: Bug in parallel Kahan summation implementation
src/java.base/share/classes/java/util/DoubleSummaryStatistics.java line 161:
> 159:
> 160: //Negating this value because low-order bits are in negated form
> 161: sumWit
On Fri, 2 Jul 2021 20:12:39 GMT, Ian Graves wrote:
> 8214761: Bug in parallel Kahan summation implementation
Crikey, how did we get that wrong?
It'd be nice if we had a regression test for this. Can you provide one, please?
-
PR: https://git.openjdk.java.net/jdk/pull/4674
On Mon, 12 Apr 2021 19:45:24 GMT, Ian Graves wrote:
> Fixes a bug where the compensated sum should be negated when added together
> in the merge step of a given collector. This impacts accuracy of parallel
> summations with Double streams and creates larger deviations from a standard
> sequent
Hello,
Thanks for the nudge; I'll get to this sometime after the Skara
transition for jdk/jdk.
Cheers,
-Joe
On 9/3/2020 9:29 AM, Chris Dennis wrote:
In case there is a need for extra motivation here:
import java.util.stream.DoubleStream;
public class Test8214761 {
public static void ma
In case there is a need for extra motivation here:
import java.util.stream.DoubleStream;
public class Test8214761 {
public static void main(String[] args) {
double a = Double.valueOf(args[0]);
if (Math.ulp(a) <= Math.ulp(Math.nextAfter(a, 0))) {
System.out.println("Stable addition
Bump...
I've run in to this while running tests that check computation results against
the expected bounds of a Kahan summation. Any chance that this gets picked up
in the near future?
Thanks,
Chris
On 12/13/18, 6:16 PM, "core-libs-dev on behalf of Ivan Gerasimov"
wrote:
Gentle ping.
Gentle ping.
On 12/9/18 7:37 PM, Ivan Gerasimov wrote:
Hello!
DoubleSummaryStatistics takes advantage of Kahan summation algorithm
to reduce the error of the total sum.
Internally it maintains a field double sumCompensation, which keeps
lower bits (which were rounded off) of the last addit
15 matches
Mail list logo
