Jim Apple has posted comments on this change. Change subject: IMPALA-2809: Improve ByteSwap with builtin function or SSE or AVX2. ......................................................................
Patch Set 13: (4 comments) http://gerrit.cloudera.org:8080/#/c/3081/13/be/src/benchmarks/bswap-benchmark.cc File be/src/benchmarks/bswap-benchmark.cc: Line 50: // SIMD 57.73 2.18X I think we should pause this code review while you figure out why SIMD is so much slower than the two SIMD methods. Line 250: /// For example, if FIXED_LEN_SIZE == 2, we are going to deal with the first I think it would be inappropriate to check this code in if Zuo Wang is the only one who understands why anyone would bother to have different FIXED_LEN_SIZE. If it can't be justified, please remove it. Line 255: /// we can get these constants by following definitions of corresponding AVX2 intrinsics: This is way too sparse. Please explain in more detail. Consider using one example to demonstrate. Line 499: MACRO_TEST_AVX2_CASE(4, 1) Please fix this duplication, too. -- To view, visit http://gerrit.cloudera.org:8080/3081 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I392ed5a8d5683f30f161282c228c1aedd7b648c1 Gerrit-PatchSet: 13 Gerrit-Project: Impala Gerrit-Branch: cdh5-trunk Gerrit-Owner: Youwei Wang <[email protected]> Gerrit-Reviewer: Alex Behm <[email protected]> Gerrit-Reviewer: Jim Apple <[email protected]> Gerrit-Reviewer: Marcel Kornacker <[email protected]> Gerrit-Reviewer: Mostafa Mokhtar <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-Reviewer: Youwei Wang <[email protected]> Gerrit-HasComments: Yes
