Impala Public Jenkins has posted comments on this change. ( http://gerrit.cloudera.org:8080/15531 )
Change subject: IMPALA-9544 Replace Intel's SSE instructions with ARM's NEON instructions ...................................................................... Patch Set 13: (39 comments) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bit-util.cc File be/src/util/bit-util.cc: http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bit-util.cc@237 PS13, Line 237: #ifndef __aarch64__ line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bit-util.cc@249 PS13, Line 249: #endif line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bloom-filter.h File be/src/util/bloom-filter.h: http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bloom-filter.h@212 PS13, Line 212: line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bloom-filter.h@221 PS13, Line 221: line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bloom-filter.cc File be/src/util/bloom-filter.cc: http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/bloom-filter.cc@20 PS13, Line 20: #include "sse2neon.h" line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse-util.h File be/src/util/sse-util.h: http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse-util.h@126 PS13, Line 126: } line has trailing whitespace http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h File be/src/util/sse2neon.h: http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@210 PS13, Line 210: // https://msdn.microsoft.com/en-us/library/bb514059%28v=vs.120%29.aspx?f=255&MSPPError=-2147217396 line too long (99 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@403 PS13, Line 403: // https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2010/whtfzhzk(v=vs.100) line too long (104 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@410 PS13, Line 410: // https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_set1_epi64x&expand=4961 line too long (97 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@544 PS13, Line 544: return vreinterpretq_m128i_s32(vcombine_s32(vld1_s32((int32_t const *) p), vcreate_s32(0))); line too long (96 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@1025 PS13, Line 1025: // https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_shuffle_epi8&expand=5146 line too long (98 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@1174 PS13, Line 1174: // https://docs.microsoft.com/en-us/previous-versions/visualstudio/visual-studio-2010/y41dkk37(v=vs.100) line too long (104 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@1513 PS13, Line 1513: // Since the widening shifts would be rather confusing to follow in little endian, everything line too long (97 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@1514 PS13, Line 1514: // will be illustrated in big endian order instead. This has a different result - the bits line too long (94 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@1601 PS13, Line 1601: // https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_test_all_zeros&expand=5871 line too long (100 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3048 PS13, Line 3048: return vreinterpretq_m128i_s8(vzip1q_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b))); line too long (99 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3073 PS13, Line 3073: return vreinterpretq_m128i_s16(vzip1q_s16(vreinterpretq_s16_m128i(a), vreinterpretq_s16_m128i(b))); line too long (103 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3094 PS13, Line 3094: return vreinterpretq_m128i_s32(vzip1q_s32(vreinterpretq_s32_m128i(a), vreinterpretq_s32_m128i(b))); line too long (103 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3122 PS13, Line 3122: return vreinterpretq_m128_f32(vzip1q_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b))); line too long (100 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3143 PS13, Line 3143: return vreinterpretq_m128_f32(vzip2q_f32(vreinterpretq_f32_m128(a), vreinterpretq_f32_m128(b))); line too long (100 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3167 PS13, Line 3167: return vreinterpretq_m128i_s8(vzip2q_s8(vreinterpretq_s8_m128i(a), vreinterpretq_s8_m128i(b))); line too long (99 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3194 PS13, Line 3194: return vreinterpretq_m128i_s16(vzip2q_s16(vreinterpretq_s16_m128i(a), vreinterpretq_s16_m128i(b))); line too long (103 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3209 PS13, Line 3209: return vreinterpretq_m128i_s32(vzip2q_s32(vreinterpretq_s32_m128i(a), vreinterpretq_s32_m128i(b))); line too long (103 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3234 PS13, Line 3234: #define _mm_alignr_epi8(a, b, c) ((__m128i) vextq_s8((int8x16_t) (b), (int8x16_t) (a), (c))) line too long (92 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3334 PS13, Line 3334: uint8x16_t k48_32 = vcombine_u8(vcreate_u8(0x0000ffffffffffff), vcreate_u8(0x00000000ffffffff)); line too long (100 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3335 PS13, Line 3335: uint8x16_t k16_00 = vcombine_u8(vcreate_u8(0x000000000000ffff), vcreate_u8(0x0000000000000000)); line too long (100 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3354 PS13, Line 3354: uint8x16_t lm_p0 = vreinterpretq_u8_u64(vzip1q_u64(vreinterpretq_u64_u8(l), vreinterpretq_u64_u8(m))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3355 PS13, Line 3355: uint8x16_t lm_p1 = vreinterpretq_u8_u64(vzip2q_u64(vreinterpretq_u64_u8(l), vreinterpretq_u64_u8(m))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3356 PS13, Line 3356: uint8x16_t nk_p0 = vreinterpretq_u8_u64(vzip1q_u64(vreinterpretq_u64_u8(n), vreinterpretq_u64_u8(k))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3357 PS13, Line 3357: uint8x16_t nk_p1 = vreinterpretq_u8_u64(vzip2q_u64(vreinterpretq_u64_u8(n), vreinterpretq_u64_u8(k))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3378 PS13, Line 3378: uint8x16_t t0 = vreinterpretq_u8_u64(vuzp1q_u64(vreinterpretq_u64_u8(t0t1_l), vreinterpretq_u64_u8(t0t1_h))); line too long (113 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3379 PS13, Line 3379: uint8x16_t t1 = vreinterpretq_u8_u64(vuzp2q_u64(vreinterpretq_u64_u8(t0t1_l), vreinterpretq_u64_u8(t0t1_h))); line too long (113 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3380 PS13, Line 3380: uint8x16_t t2 = vreinterpretq_u8_u64(vuzp1q_u64(vreinterpretq_u64_u8(t2t3_l), vreinterpretq_u64_u8(t2t3_h))); line too long (113 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3381 PS13, Line 3381: uint8x16_t t3 = vreinterpretq_u8_u64(vuzp2q_u64(vreinterpretq_u64_u8(t2t3_l), vreinterpretq_u64_u8(t2t3_h))); line too long (113 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3408 PS13, Line 3408: case 0x00: return vreinterpretq_m128i_u64(_sse2neon_vmull_p64(vget_low_u64(a), vget_low_u64(b))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3409 PS13, Line 3409: case 0x01: return vreinterpretq_m128i_u64(_sse2neon_vmull_p64(vget_high_u64(a), vget_low_u64(b))); line too long (106 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3410 PS13, Line 3410: case 0x10: return vreinterpretq_m128i_u64(_sse2neon_vmull_p64(vget_low_u64(a), vget_high_u64(b))); line too long (107 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3411 PS13, Line 3411: case 0x11: return vreinterpretq_m128i_u64(_sse2neon_vmull_p64(vget_high_u64(a), vget_high_u64(b))); line too long (107 > 90) http://gerrit.cloudera.org:8080/#/c/15531/13/be/src/util/sse2neon.h@3421 PS13, Line 3421: // https://github.com/ColinIanKing/linux-next-mirror/blob/b5f466091e130caaf0735976648f72bd5e09aa84/crypto/aegis128-neon-inner.c#L52 line too long (131 > 90) -- To view, visit http://gerrit.cloudera.org:8080/15531 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Id7dfe17125b2910ece54e7dd18b4e4b25d7de8b9 Gerrit-Change-Number: 15531 Gerrit-PatchSet: 13 Gerrit-Owner: Anonymous Coward <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Comment-Date: Tue, 07 Apr 2020 03:44:17 +0000 Gerrit-HasComments: Yes
