You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by "AntoinePrv (via GitHub)" <gi...@apache.org> on 2026/07/23 08:32:14 UTC
[PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
AntoinePrv opened a new pull request, #50611:
URL: https://github.com/apache/arrow/pull/50611
### Rationale for this change
The SVE128 code path has conflict with the SVE256 that we do not yet manage properly.
- There was first the ODR violation in GH-49921
- Now it seems that there may also be an issue with LTO
Anyhow, after we fixed the inlining issue in Neon, the SVE128 had no clear advantages over Neon as expected, os this was due to be removed anyways.
### What changes are included in this PR?
Remove SVE128 unpack
### Are these changes tested?
In CI.
### Are there any user-facing changes?
No
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5071361845
Everything that relies on bit-unpacking, i.e. some Parquet reading benchmarks.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "raulcd (via GitHub)" <gi...@apache.org>.
raulcd commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056929705
The wheel can be downloaded from the summary page artifacts on the job:
https://github.com/ursacomputing/crossbow/actions/runs/29994419195
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou merged PR #50611:
URL: https://github.com/apache/arrow/pull/50611
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "raulcd (via GitHub)" <gi...@apache.org>.
raulcd commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056202903
@github-actions crossbow submit wheel-manylinux-2-28-cp313-cp313-arm64
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5131182768
Benchmark reports are ok, the so-called regressions seem unrelated to these changes.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on code in PR #50611:
URL: https://github.com/apache/arrow/pull/50611#discussion_r3665470596
##########
cpp/src/arrow/CMakeLists.txt:
##########
@@ -346,21 +346,42 @@ endmacro()
macro(append_runtime_sve128_src SRCS SRC)
if(ARROW_HAVE_RUNTIME_SVE128)
list(APPEND ${SRCS} ${SRC})
- set_source_files_properties(${SRC} PROPERTIES COMPILE_OPTIONS "${ARROW_SVE128_FLAGS}")
+ set(_flags ${ARROW_SVE128_FLAGS})
Review Comment:
@kou Do these CMake changes look ok?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5072455959
@AntoinePrv could you push an empty commit so conbench can be started again? (It allows only one benchmark run per commit).
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5060500505
> The best hypothesis I have goes as follows (this was heavily AI assisted):
>
> We had a similar problem with the ODR violation, but the cause is different here (latest xsimd has the ODR fix). The issue arise during LTO where the fags `-msve-vector-bits=128` and `-msve-vector-bits=256` get conflated, resulting in the 256 version making 128 bits assumptions.
We should find out why. Can we enable `CMAKE_VERBOSE_MAKEFILE` on the manylinux build to see all the compiler options for each file? Normally only one `-msve-vector-bits` is passed for each compilation unit.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5073492783
Commit e84d38f17b19ea94a3cfab59a46ec4730fa20a48 already has scheduled benchmark runs.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "raulcd (via GitHub)" <gi...@apache.org>.
raulcd commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5071913575
The regression for mac-arm seems to be quite significant.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "conbench-apache-arrow[bot] (via GitHub)" <gi...@apache.org>.
conbench-apache-arrow[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5076113434
Thanks for your patience. Conbench analyzed the 4 benchmarking runs that have been run so far on PR commit e84d38f17b19ea94a3cfab59a46ec4730fa20a48.
There was 1 benchmark result indicating a performance regression:
- Pull Request Run on `arm64-t4g-2xlarge-linux` at [2026-07-24 19:58:18Z](https://conbench.arrow-dev.org/compare/runs/a3dcfd3183924c59b5894dd4f7bbbd15...fc77a2c37f064442af49e4e89096046a/)
- [`TouchArea` (C++) with params=size:65536/real_time/threads:2, source=cpp-micro, suite=arrow-memory-pool-benchmark](https://conbench.arrow-dev.org/compare/benchmarks/06a61974ad46734980001eb7345d27b2...06a63c31f787794a8000a89949d57775)
The [full Conbench report](https://github.com/apache/arrow/runs/89628319180) has more details.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "kou (via GitHub)" <gi...@apache.org>.
kou commented on code in PR #50611:
URL: https://github.com/apache/arrow/pull/50611#discussion_r3680679539
##########
cpp/src/arrow/CMakeLists.txt:
##########
@@ -346,21 +346,42 @@ endmacro()
macro(append_runtime_sve128_src SRCS SRC)
if(ARROW_HAVE_RUNTIME_SVE128)
list(APPEND ${SRCS} ${SRC})
- set_source_files_properties(${SRC} PROPERTIES COMPILE_OPTIONS "${ARROW_SVE128_FLAGS}")
+ set(_flags ${ARROW_SVE128_FLAGS})
Review Comment:
Yes.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5103624659
I added not no-lto 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
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5078037765
No more macOS regressions, I think it's just noisy and TouchArea similarly so.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056597534
Revision: 3458a70d02c81cf134b2a9eab2108d84f3872ab8
Submitted crossbow builds: [ursacomputing/crossbow @ actions-b71ab30610](https://github.com/ursacomputing/crossbow/branches/all?query=actions-b71ab30610)
|Task|Status|
|----|------|
|wheel-manylinux-2-28-cp313-cp313-arm64|[](https://github.com/ursacomputing/crossbow/actions/runs/29994419195/job/89164580205)|
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5060625746
Looking at the [original build logs](https://github.com/ursacomputing/crossbow/actions/runs/29994419195/job/89164580205#step:8:790), `bpacking_simd_128.cc` is compiled inside a unity build compilation unit, which doesn't seem right. It should be compiled independently, just like `bpacking_simd_256.cc`.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5103174587
> @pitrou what so you think, should we merge? One more data point is that the each `unpack` implementation adds some significant size to the binary.
Yes, I think this is ok as workaround. But more generally we should also think about disabling LTO on SVE translation units as done in https://github.com/apache/arrow/pull/50618/, what do you think @cyb70289 ?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "cyb70289 (via GitHub)" <gi...@apache.org>.
cyb70289 commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5103553370
LGTM to remove this sve128 code with no performance benefit.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5071193645
Do we have SVE benchmarks?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "conbench-apache-arrow[bot] (via GitHub)" <gi...@apache.org>.
conbench-apache-arrow[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5071475393
Thanks for your patience. Conbench analyzed the 4 benchmarking runs that have been run so far on PR commit 3458a70d02c81cf134b2a9eab2108d84f3872ab8.
There were 4 benchmark results indicating a performance regression:
- Pull Request Run on `test-mac-arm` at [2026-07-24 15:06:52Z](https://conbench.arrow-dev.org/compare/runs/ca7fc31804be444baeb7a61afc339d31...0991b0b447e64938a2bf17fbd19d8a13/)
- [`BM_UnpackUint16` (C++) with params=DynamicAligned/8/512, source=cpp-micro, suite=arrow-bpacking-benchmark](https://conbench.arrow-dev.org/compare/benchmarks/06a6324ae7257d96800011733af3d0e4...06a6379f48027fd68000cf7f0be6da4a)
- [`BM_UnpackUint16` (C++) with params=DynamicUnaligned/8/512, source=cpp-micro, suite=arrow-bpacking-benchmark](https://conbench.arrow-dev.org/compare/benchmarks/06a6324b257479b68000f75aa64c7583...06a6379f88cf75128000567e0e15dff6)
- and 2 more (see the report linked below)
The [full Conbench report](https://github.com/apache/arrow/runs/89520016618) has more details.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5072191860
Commit 3458a70d02c81cf134b2a9eab2108d84f3872ab8 already has scheduled benchmark runs.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5073135312
> > The regression for mac-arm seems to be quite significant.
>
> It appears similar regression occurs occasionally. I'll restart the benchmark.
Yes, MacOS has no reason to be affected by this (it cannot even run the code we removed), so this one is definitely spurious.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5058211598
The best hypothesis I have goes as follows (this was heavily AI assisted):
We had a similar problem with the ODR violation, but the cause is different here (latest xsimd has the ODR fix).
The issue arise during LTO where the fags `-msve-vector-bits=128` and `-msve-vector-bits=256` get conflated, resulting in the 256 version making 128 bits assumptions.
The supporting evidence is the presence of `ptrue p7.b, vl16` in `unpack_sve256`: this sets a predicate (all true since we don't use them) only to 16 bytes (128 bits).
While there is some uncertainty in this diagnosis, the presence of both `sve128` and `sve256` is an extremely likely enabler of the issue. Given I had plan to remove the SVE128 because it did not bring much over Neon, landing this fix has no drawbacks.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5067124824
https://github.com/apache/arrow/pull/50618 removes LTO for SVE-enabled translation units and works fine as per @cyb70289
Let's see whether the benchmarks are affected.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5068108204
Benchmark runs are scheduled for commit 3458a70d02c81cf134b2a9eab2108d84f3872ab8. Watch https://buildkite.com/apache-arrow and https://conbench.arrow-dev.org for updates. A comment will be posted here when the runs are complete.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5127750733
Benchmark runs are scheduled for commit 504bd8e703a29b61f3915209fa4bb00b609a122b. Watch https://buildkite.com/apache-arrow and https://conbench.arrow-dev.org for updates. A comment will be posted here when the runs are complete.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5073124771
@ursabot please benchmark
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "conbench-apache-arrow[bot] (via GitHub)" <gi...@apache.org>.
conbench-apache-arrow[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5155019181
After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 51131e28434bf87d634cb18c2b8457ce101331e0.
There was 1 benchmark result indicating a performance regression:
- Commit Run on `amd64-m5-4xlarge-linux` at [2026-07-30 17:02:43Z](https://conbench.arrow-dev.org/compare/runs/d56a14b2cab54fc3b8fe5ce2b09e9e0e...2d217bd8e4d54c7ea6b0bc9acf13da09/)
- [`csv-read` (Python) with compression=uncompressed, dataset=fanniemae_2016Q4, output_format=arrow_table, streaming=streaming](https://conbench.arrow-dev.org/compare/benchmarks/06a6b73256e57586800064c59d1bdc72...06a6b800daa37d3080008a62fa4ff0a7)
The [full Conbench report](https://github.com/apache/arrow/runs/91451231048) has more details. It also includes information about 33 possible false positives for unstable benchmarks that are known to sometimes produce them.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5060761864
Ok, I think the issue is https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121412 .
And specifically:
> That said, -fno-lto for "mismatched" TUs is the workaroud that, unfortunately, has to be applied on the users side at this point.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5072186639
> The regression for mac-arm seems to be quite significant.
It appears similar regression occurs occasionally. I'll restart the benchmark.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5072188266
@ursabot please benchmark
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5127749871
@ursabot please benchmark lang=C++
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056223681
Revision: 84f82095cc5f4c6c604bc485e41edf3149202cc6
Submitted crossbow builds: [ursacomputing/crossbow @ actions-8af012a39a](https://github.com/ursacomputing/crossbow/branches/all?query=actions-8af012a39a)
|Task|Status|
|----|------|
|wheel-manylinux-2-28-cp313-cp313-arm64|[](https://github.com/ursacomputing/crossbow/tree/actions-8af012a39a-github-wheel-manylinux-2-28-cp313-cp313-arm64)|
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5057474136
Can we try to understand some happens? For example look for symbols with `sve` in them but no `sve128` or `sve256`.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5058335216
> should we try building a wheel without LTO and validate?
We can, though it would be worst fix IMHO.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "raulcd (via GitHub)" <gi...@apache.org>.
raulcd commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5058317248
should try building a wheel without LTO and validate?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "conbench-apache-arrow[bot] (via GitHub)" <gi...@apache.org>.
conbench-apache-arrow[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5130623120
Thanks for your patience. Conbench analyzed the 4 benchmarking runs that have been run so far on PR commit 504bd8e703a29b61f3915209fa4bb00b609a122b.
There were 35 benchmark results indicating a performance regression:
- Pull Request Run on `amd64-c6a-4xlarge-linux` at [2026-07-30 08:17:23Z](https://conbench.arrow-dev.org/compare/runs/539f0a44b79f48e597394387a8ca0b8b...1c92b306b0a2492cb47582822356ef57/)
- [`ConstructFromEmptyStdVector` (C++) with params=<STATIC_VECTOR(std::string)>, source=cpp-micro, suite=arrow-small-vector-benchmark](https://conbench.arrow-dev.org/compare/benchmarks/06a61904fa42746780002465b95bbbe1...06a6b0906ccb76648000cde4b167db8b)
- [`DoubleColumnMajorMatrixConversionFixture` (C++) with params=<Int32Type>/ConvertToSparseCSCMatrixInt32, source=cpp-micro, suite=arrow-tensor-conversion-benchmark](https://conbench.arrow-dev.org/compare/benchmarks/06a6190a9bd078b38000ced729ece5da...06a6b0926e0573d08000f661086e5b29)
- and 33 more (see the report linked below)
The [full Conbench report](https://github.com/apache/arrow/runs/90866624379) has more details.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5102170711
@pitrou what so you think, should we merge?
One more data point is that the each `unpack` implementation adds some significant size to the binary.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "pitrou (via GitHub)" <gi...@apache.org>.
pitrou commented on code in PR #50611:
URL: https://github.com/apache/arrow/pull/50611#discussion_r3664870568
##########
cpp/src/arrow/util/bpacking.cc:
##########
@@ -34,7 +34,6 @@ struct UnpackDynamicFunction {
return std::array{
ARROW_DISPATCH_TARGET_NONE(&bpacking::unpack_scalar<Uint>) //
ARROW_DISPATCH_TARGET_NEON(&bpacking::unpack_neon<Uint>) //
- ARROW_DISPATCH_TARGET_SVE128(&bpacking::unpack_sve128<Uint>) //
Review Comment:
Can you add a comment explaining why we don't include SVE128 here?
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "github-actions[bot] (via GitHub)" <gi...@apache.org>.
github-actions[bot] commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056171651
:warning: GitHub issue #50503 **has been automatically assigned in GitHub** to PR creator.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "AntoinePrv (via GitHub)" <gi...@apache.org>.
AntoinePrv commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5056562171
@github-actions crossbow submit wheel-manylinux-2-28-cp313-cp313-arm64
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5073125729
Benchmark runs are scheduled for commit e84d38f17b19ea94a3cfab59a46ec4730fa20a48. Watch https://buildkite.com/apache-arrow and https://conbench.arrow-dev.org for updates. A comment will be posted here when the runs are complete.
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "rok (via GitHub)" <gi...@apache.org>.
rok commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5073492417
@ursabot please benchmark
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] GH-50503: [Parquet] Remove SVE128 unpack [arrow]
Posted by "raulcd (via GitHub)" <gi...@apache.org>.
raulcd commented on PR #50611:
URL: https://github.com/apache/arrow/pull/50611#issuecomment-5068107419
@ursabot please benchmark
--
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: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org