msfroh commented on PR #13521:
URL: https://github.com/apache/lucene/pull/13521#issuecomment-2313906920
I tried modifying the loop to process 4 longs per iteration and noticed no
difference on my Xeon host, which is unsurprising since there was no difference
between 1 and 3.
I also t
github-actions[bot] commented on PR #13072:
URL: https://github.com/apache/lucene/pull/13072#issuecomment-2313794696
This PR has not had activity in the past 2 weeks, labeling it as stale. If
the PR is waiting for review, notify the d...@lucene.apache.org list. Thank you
for your contributi
github-actions[bot] commented on PR #13253:
URL: https://github.com/apache/lucene/pull/13253#issuecomment-2313794562
This PR has not had activity in the past 2 weeks, labeling it as stale. If
the PR is waiting for review, notify the d...@lucene.apache.org list. Thank you
for your contributi
gsmiller commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1733650732
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -212,13 +213,74 @@ static void prefixSum(long[] buffer, int count, long
gsmiller commented on code in PR #13693:
URL: https://github.com/apache/lucene/pull/13693#discussion_r1733631990
##
lucene/core/src/java/org/apache/lucene/document/SortedSetDocValuesRangeQuery.java:
##
@@ -161,66 +156,8 @@ public Scorer get(long leadCost) throws IOException {
gsmiller commented on code in PR #13693:
URL: https://github.com/apache/lucene/pull/13693#discussion_r1733632247
##
lucene/core/src/java/org/apache/lucene/document/SortedSetDocValuesRangeQuery.java:
##
@@ -236,69 +173,4 @@ public boolean isCacheable(LeafReaderContext ctx) {
gsmiller opened a new pull request, #13693:
URL: https://github.com/apache/lucene/pull/13693
### Description
This is a small optimization that treats "slow" multi-term queries (e.g.,
TermInSet, RegexpQuery, etc.) as ordinal ranges when the query terms create a
contiguous range. This
msfroh commented on PR #13521:
URL: https://github.com/apache/lucene/pull/13521#issuecomment-2313628836
The approach is pretty neat.
I'm wondering if `Bit21With3StepsEncoder` does better on aarch64 because of
the explicitly unrolled loop? If so, I'm wondering if unrolling to a multip
jpountz commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1733487895
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException {
jpountz commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1733472210
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException {
owaiskazi19 commented on PR #13691:
URL: https://github.com/apache/lucene/pull/13691#issuecomment-231364
> Thank you @owaiskazi19 -- I backported to 9.12 as well.
Thanks @mikemccand for the help here.
--
This is an automated message from the Apache Git Service.
To respond to the
mikemccand commented on PR #13691:
URL: https://github.com/apache/lucene/pull/13691#issuecomment-2313434759
Thank you @owaiskazi19 -- I backported to 9.12 as well.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
UR
mikemccand merged PR #13691:
URL: https://github.com/apache/lucene/pull/13691
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@lucene.
gsmiller commented on issue #13671:
URL: https://github.com/apache/lucene/issues/13671#issuecomment-2313193985
Cool, sounds like we've got a good plan. I'll just add that I've been
noodling on the idea for a while of seeing if we could actually get rid of
`DrillSideways` and fold what it do
msokolov commented on PR #13683:
URL: https://github.com/apache/lucene/pull/13683#issuecomment-2313146569
I'd like to get this tool merged before moving on to thinking about
different index encodings. To that end I plan to ensure that it works with all
the vector similarities, and with byte
msokolov commented on issue #13687:
URL: https://github.com/apache/lucene/issues/13687#issuecomment-2313000464
In graph theory "strongly connected" means every node in the graph can be
reached from every other node. In a bidirectional (undirected) graph its the
same as "connected". But in a
mikemccand commented on issue #13671:
URL: https://github.com/apache/lucene/issues/13671#issuecomment-2312947485
I like Option 1, first, as well. We can later explore flavors of 2 and 3.
Thanks @epotyom and @gsmiller.
--
This is an automated message from the Apache Git Service.
To respo
epotyom commented on issue #13671:
URL: https://github.com/apache/lucene/issues/13671#issuecomment-2312932669
@gsmiller , I agree. I'll create a PR for the 1st option then in the next
couple of days. Thanks!
--
This is an automated message from the Apache Git Service.
To respond to the me
mikemccand commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1733130032
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException
pierwill commented on issue #13687:
URL: https://github.com/apache/lucene/issues/13687#issuecomment-2312910112
> But it is still an open question the extent to which this kind of check
could potentially improve graph connectivity and thus recall
I only understand the broad ideas here.
jpountz commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1733116688
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException {
mikemccand commented on issue #13647:
URL: https://github.com/apache/lucene/issues/13647#issuecomment-2312879477
I also aliased (CNAMEd)
[benchmarks.mikemccandless.com](https://benchmarks.mikemccandless.com/) --
GitHub pages makes this simple-ish, yay.
--
This is an automated message fro
mikemccand commented on issue #13647:
URL: https://github.com/apache/lucene/issues/13647#issuecomment-2312800530
> > FYI: I clicked on a few random links and found a 404
https://mikemccand.github.io/luceneutil/analyzers.html although this page does
seem to exist on the current site
>
gsmiller commented on issue #13671:
URL: https://github.com/apache/lucene/issues/13671#issuecomment-2312763368
Thanks @epotyom for the detailed thoughts! If I'm understanding this
correctly, it sounds like "option 1" would let us remove the need for
`CollectorOwner` along with the new API i
mikemccand commented on issue #13647:
URL: https://github.com/apache/lucene/issues/13647#issuecomment-2312740634
Phew, OK, I think nightly benchy is now successfully publishing
automatically to https://mikemccand.github.io/lucenenightly (using GitHub
pages). Last night's run "just worked".
jpountz commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1732939828
##
lucene/benchmark-jmh/src/java/org/apache/lucene/benchmark/jmh/AdvanceBenchmark.java:
##
@@ -0,0 +1,376 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) u
madrob commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1732934065
##
lucene/benchmark-jmh/src/java/org/apache/lucene/benchmark/jmh/AdvanceBenchmark.java:
##
@@ -0,0 +1,376 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) un
jpountz commented on PR #13692:
URL: https://github.com/apache/lucene/pull/13692#issuecomment-2312622096
> This is actually quite an impressive speedup!
To be honest, I believe that this is the only run where I got more than 10%,
other runs were in the 5%-10% range for `CountAndHighHi
mikemccand commented on PR #13692:
URL: https://github.com/apache/lucene/pull/13692#issuecomment-2312618083
> CountAndHighHigh 48.56 (1.5%) 53.62 (1.0%) 10.4% (
7% - 13%) 0.000
This is actually quite an impressive speedup!
--
This is an automated message
jpountz commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1732885915
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException {
mikemccand commented on code in PR #13692:
URL: https://github.com/apache/lucene/pull/13692#discussion_r1732819218
##
lucene/core/src/java/org/apache/lucene/codecs/lucene912/Lucene912PostingsReader.java:
##
@@ -509,15 +547,14 @@ private void refillRemainder() throws IOException
javanna commented on code in PR #13542:
URL: https://github.com/apache/lucene/pull/13542#discussion_r1732759669
##
lucene/core/src/java/org/apache/lucene/search/IndexSearcher.java:
##
@@ -890,11 +945,70 @@ public static class LeafSlice {
*
* @lucene.experimental
jpountz commented on PR #13692:
URL: https://github.com/apache/lucene/pull/13692#issuecomment-2311845595
Here is what the `AdvanceBenchmark` reports. The branchless binary search is
`binarySearch5`, which performs much faster than a regular binary search but
still slower than a linear searc
jpountz opened a new pull request, #13692:
URL: https://github.com/apache/lucene/pull/13692
Advancing within a block consists of finding the first index within an array
of 128 values whose value is greater than or equal a target. Given the small
size, it's not obvious whether it's better to
owaiskazi19 opened a new pull request, #13691:
URL: https://github.com/apache/lucene/pull/13691
### Description
Coming from https://github.com/opensearch-project/OpenSearch/issues/14921.
This PR fixes the correct value of exponent from pivot in the Sigmoid function.
--
This is
35 matches
Mail list logo
