[jira] [Assigned] (ARROW-10321) [C++] Building AVX512 code when we should not

[ https://issues.apache.org/jira/browse/ARROW-10321?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du reassigned ARROW-10321: Assignee: Frank Du > [C++] Building AVX512 code when we should not >

[jira] [Commented] (ARROW-10321) [C++] Building AVX512 code when we should not

[ https://issues.apache.org/jira/browse/ARROW-10321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17215140#comment-17215140 ] Frank Du commented on ARROW-10321: -- For the second issue, yes, it should be better fixed in cmake side,

[jira] [Commented] (ARROW-10026) [C++] Improve kernel performance on small batches

[ https://issues.apache.org/jira/browse/ARROW-10026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17198197#comment-17198197 ] Frank Du commented on ARROW-10026: -- Seems most batch for ArrayData now is exactly a copy of current,

[jira] [Created] (ARROW-10022) [C++] [Compute] core dumped on some scalar-arithmetic-benchmark

Frank Du created ARROW-10022: Summary: [C++] [Compute] core dumped on some scalar-arithmetic-benchmark Key: ARROW-10022 URL: https://issues.apache.org/jira/browse/ARROW-10022 Project: Apache Arrow

[jira] [Created] (ARROW-9904) [C++] Unroll the loop manually for CountSetBits

Frank Du created ARROW-9904: --- Summary: [C++] Unroll the loop manually for CountSetBits Key: ARROW-9904 URL: https://issues.apache.org/jira/browse/ARROW-9904 Project: Apache Arrow Issue Type:

[jira] [Commented] (ARROW-9794) [C++] Add functionality to cpu_info to discriminate between Intel vs AMD x86

[ https://issues.apache.org/jira/browse/ARROW-9794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17188914#comment-17188914 ] Frank Du commented on ARROW-9794: -  Sure, I will take a look > [C++] Add functionality to cpu_info to

[jira] [Assigned] (ARROW-9794) [C++] Add functionality to cpu_info to discriminate between Intel vs AMD x86

[ https://issues.apache.org/jira/browse/ARROW-9794?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du reassigned ARROW-9794: --- Assignee: Frank Du > [C++] Add functionality to cpu_info to discriminate between Intel vs AMD x86 >

[jira] [Commented] (ARROW-9877) [C++][CI] homebrew-cpp fails due to avx512

[ https://issues.apache.org/jira/browse/ARROW-9877?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17186383#comment-17186383 ] Frank Du commented on ARROW-9877: - Apple clang doesn't extract "-march=skylake-avx512" to include

[jira] [Commented] (ARROW-9877) [C++][CI] homebrew-cpp fails due to avx512

[ https://issues.apache.org/jira/browse/ARROW-9877?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17186188#comment-17186188 ] Frank Du commented on ARROW-9877: - Yes. But how can I trigger a homebrew-cpp build?

[jira] [Created] (ARROW-9872) [C++] Move byte_stream_split intrinsic code to runtime path

Frank Du created ARROW-9872: --- Summary: [C++] Move byte_stream_split intrinsic code to runtime path Key: ARROW-9872 URL: https://issues.apache.org/jira/browse/ARROW-9872 Project: Apache Arrow

[jira] [Created] (ARROW-9871) [C++] Add uppercase support to ARROW_USER_SIMD_LEVEL.

Frank Du created ARROW-9871: --- Summary: [C++] Add uppercase support to ARROW_USER_SIMD_LEVEL. Key: ARROW-9871 URL: https://issues.apache.org/jira/browse/ARROW-9871 Project: Apache Arrow Issue Type:

[jira] [Commented] (ARROW-9643) [C++] Illegal instruction on haswell cpu

[ https://issues.apache.org/jira/browse/ARROW-9643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17176692#comment-17176692 ] Frank Du commented on ARROW-9643: - Yes, this is a bug introduced from 

[jira] [Created] (ARROW-9717) [C++] Add a facility helper for SIMD runtime dispatch.

Frank Du created ARROW-9717: --- Summary: [C++] Add a facility helper for SIMD runtime dispatch. Key: ARROW-9717 URL: https://issues.apache.org/jira/browse/ARROW-9717 Project: Apache Arrow Issue

[jira] [Created] (ARROW-9702) [C++] Move bpacking simd to runtime path

Frank Du created ARROW-9702: --- Summary: [C++] Move bpacking simd to runtime path Key: ARROW-9702 URL: https://issues.apache.org/jira/browse/ARROW-9702 Project: Apache Arrow Issue Type: Improvement

[jira] [Closed] (ARROW-8579) [C++] AVX512 part for SIMD operations of DecodeSpaced/EncodeSpaced

[ https://issues.apache.org/jira/browse/ARROW-8579?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du closed ARROW-8579. --- Resolution: Won't Fix > [C++] AVX512 part for SIMD operations of DecodeSpaced/EncodeSpaced >

[jira] [Commented] (ARROW-8579) [C++] AVX512 part for SIMD operations of DecodeSpaced/EncodeSpaced

[ https://issues.apache.org/jira/browse/ARROW-8579?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17176077#comment-17176077 ] Frank Du commented on ARROW-8579: - No need now. The typical 0.01% is optimized  by 

[jira] [Assigned] (ARROW-9605) [C++] Optimize performance for aggregate min/max compute kernels

[ https://issues.apache.org/jira/browse/ARROW-9605?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du reassigned ARROW-9605: --- Assignee: Frank Du > [C++] Optimize performance for aggregate min/max compute kernels >

[jira] [Commented] (ARROW-9643) [C++] Illegal instruction on haswell cpu

[ https://issues.apache.org/jira/browse/ARROW-9643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17171353#comment-17171353 ] Frank Du commented on ARROW-9643: - I can reproduce the issue on gcc-7.5 + ubuntu18.04 with a AVX2 device, 

[jira] [Commented] (ARROW-9643) [C++] Illegal instruction on haswell cpu

[ https://issues.apache.org/jira/browse/ARROW-9643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17171326#comment-17171326 ] Frank Du commented on ARROW-9643: - Just upload a PR which call the SIMD func register if it's capable, it

[jira] [Commented] (ARROW-9643) [C++] Illegal instruction on haswell cpu

[ https://issues.apache.org/jira/browse/ARROW-9643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17171313#comment-17171313 ] Frank Du commented on ARROW-9643: - Thanks for the quick findings. Sorry for that, I has a AVX2 device but

[jira] [Created] (ARROW-9605) [C++] Optimize performance for aggregate min/max compute kernels

Frank Du created ARROW-9605: --- Summary: [C++] Optimize performance for aggregate min/max compute kernels Key: ARROW-9605 URL: https://issues.apache.org/jira/browse/ARROW-9605 Project: Apache Arrow

[jira] [Created] (ARROW-9604) [C++] Add benchmark for aggregate min/max compute kernels

Frank Du created ARROW-9604: --- Summary: [C++] Add benchmark for aggregate min/max compute kernels Key: ARROW-9604 URL: https://issues.apache.org/jira/browse/ARROW-9604 Project: Apache Arrow Issue

[jira] [Created] (ARROW-9398) [C++] Register the SIMD sum variants under function instance instead a SIMD function

Frank Du created ARROW-9398: --- Summary: [C++] Register the SIMD sum variants under function  instance instead a SIMD function Key: ARROW-9398 URL: https://issues.apache.org/jira/browse/ARROW-9398 Project:

[jira] [Updated] (ARROW-8996) [C++] Runtime SIMD path for Aggregate Sum/Mean kernel

[ https://issues.apache.org/jira/browse/ARROW-8996?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du updated ARROW-8996: Summary: [C++] Runtime SIMD path for Aggregate Sum/Mean kernel (was: [C++] Runtime SIMD path for

[jira] [Updated] (ARROW-8996) [C++] Runtime SIMD path for Aggregate Sum/Mean kernel

[ https://issues.apache.org/jira/browse/ARROW-8996?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du updated ARROW-8996: Description: Thanks to the new kernel framework, now it can override a SIMD kernel version at runtime.  

[jira] [Updated] (ARROW-8996) [C++] Runtime SIMD path for Aggregate Sum kernel of dense

[ https://issues.apache.org/jira/browse/ARROW-8996?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du updated ARROW-8996: Summary: [C++] Runtime SIMD path for Aggregate Sum kernel of dense (was: [C++] Runtime SSE path for

[jira] [Created] (ARROW-9263) [C++] Benchmark: promote RegressionSetArgs size to L2

Frank Du created ARROW-9263: --- Summary: [C++] Benchmark: promote RegressionSetArgs size to L2 Key: ARROW-9263 URL: https://issues.apache.org/jira/browse/ARROW-9263 Project: Apache Arrow Issue Type:

[jira] [Created] (ARROW-9217) [C++] Cover 0.01% null for the plain spaced encoding/decoding benchmark

Frank Du created ARROW-9217: --- Summary: [C++] Cover 0.01% null for the plain spaced encoding/decoding benchmark Key: ARROW-9217 URL: https://issues.apache.org/jira/browse/ARROW-9217 Project: Apache Arrow

[jira] [Updated] (ARROW-9216) [C++] Use BitBlockCounter for plain spaced encoding/decoding

[ https://issues.apache.org/jira/browse/ARROW-9216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Frank Du updated ARROW-9216: Description: Speedup the typical use case which most datas are true values > [C++] Use BitBlockCounter for

[jira] [Created] (ARROW-9216) [C++] Use BitBlockCounter for plain spaced encoding/decoding

Frank Du created ARROW-9216: --- Summary: [C++] Use BitBlockCounter for plain spaced encoding/decoding Key: ARROW-9216 URL: https://issues.apache.org/jira/browse/ARROW-9216 Project: Apache Arrow