uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1622126199
> However, I am not sure why many of the tests are failing, even though the
tests for Pforutil and forutil are passing. I will take a closer look at the
specific reasons when I h
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1622121123
> I agree. There are more complications: DataInput does not have a read
method for int[], only one for float[] and long[]. So changing this is a bigger
task.
I just notic
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1620744388
I agree. There are more complications: DataInput does not have a read method
for int[], only one for float[] and long[]. So changing this is a bigger task.
I tend to think that w
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1618683855
Here is the example what I did with your code: #12412
It should demonstrate how I inlcuded it. Most interesting is the cleanup
inside the lucene90 codec. I also defined the
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1618548308
Hi,
the bug in the generator is fixed. I was able to include the ForUtil
generator into a local branch and made everything working.
What I am wondering about: The genera
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1618261123
Here is the bugfix for the generator. This bug caused a lot trouble here
because I did not understand why the script wasn't executed at all.
--
This is an automated message fro
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1618230924
Hi, I just noticed that the forUtil.gradle had a bug in Lucene 9. It does
not regenerate anything because the actual python script is never executed.
I will povide a separa
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1617817314
One small suggestion: Please stay with classical switch statement, as
backporting the python code and java code to Java 11 is getting harder (please
remember: also the vectorized
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1617809930
I am also not sure if we really need to unroll all that loops and have
different impls for each size.
--
This is an automated message from the Apache Git Service.
To respond to
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1617805484
> What about changing the format now, as part of this change? This would
mean that users who do not enable the Parama vector API would get a bit worse
performance compared to tod
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1617754157
The framework to plug in in custom formats into
`org.apache.lucene.internal.vectorization` was merged, so stage is open to
implement something.
Hints:
- Both the scalar
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1617678206
P.S.: I checked: ForUtil in 9.x is different than in 8.x because the byte
order and some other slight changes. So I would only vectorize the core one
(9.x) and leave out the back
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1616745892
Here is my PR about the general setup to easily plug in different interfaces
to support vectorization at other places (not only VectorUtil). This uses a
non-exported package and
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1616432561
Great! I am in weekend mode bit I have a rewrite of the provider lookup
almost ready.
I will check how to include it.
--
This is an automated message from the Apache Git Ser
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1615899286
I had some ideas in mind that the general lookup part just makes sure theres
is a paname impl available and then delegates to the panama provider to decide
which implementation t
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1615890843
I like the idea to put everything under internal, we have that apckage
already.
But I would keep the implementations all under one package and let codes use
them (specific
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1615852602
One we have some code here ready, I will submit a PR with the refactoring as
preparation.
--
This is an automated message from the Apache Git Service.
To respond to the message
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1615851386
Hey, please for not let me do the integration. I have a plan already because
with the current version it wont work.
We have no secrets at moment, so the implementation can
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1611856694
The above would be a separate PR to cleanup the adhoc internal
implementation of the Panama Integration a bit. The implementation devloped
here could then be added to the new Luc
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1611841983
You can take the current python script as "basis" and work from there.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to G
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1611811043
Hi,
if you look at the first line of `ForUtil.java`, you will see the following
comment:
```java
// This file has been automatically generated, DO NOT EDIT
```
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1606834703
In my opinion, the best would be to proceed here without integrating in
Lucene and I can have a look at the code. The provider interfaces in Lucene are
still package private, so
uschindler commented on issue #12396:
URL: https://github.com/apache/lucene/issues/12396#issuecomment-1606799756
Hi this is a good idea, ForUtil is next to PackedInts a good option for SIMD.
If you have something implemented, open a PR, we will think of the best way
ho to include it w
23 matches
Mail list logo
