Commit Graph

  • b62e7cd820 fix generic arch build dzaima 2024-09-02 00:04:32 +0300
  • 858dd5bc99 fix bitwiden aarch64 check dzaima 2024-08-31 18:28:40 +0300
  • c8e0bf653b fast path for ∊allEqual dzaima 2024-08-27 20:20:52 +0300
  • 0106f04495 make getPageSize return ux dzaima 2024-08-27 20:20:34 +0300
  • 559ccae9cc fix select_rows_direct dzaima 2024-08-23 04:24:42 +0300
  • 79b0bda0d3 fix •file.MapBytes dzaima 2024-08-23 04:20:44 +0300
  • 09bdeb7fd8 more usz overflow checking dzaima 2024-08-23 03:59:51 +0300
  • 06689960a5 native ≍⁼⍟k 𝕩 dzaima 2024-08-23 02:44:49 +0300
  • 5ef4fe828b use customizeShape in transpose dzaima 2024-08-23 02:21:59 +0300
  • 6ead1baa11 native ≍⁼ dzaima 2024-08-23 02:21:48 +0300
  • 413fb8893b
    Merge pull request #117 from mlochbaum/adaptive-minmax-scan dzaima 2024-08-20 23:40:54 +0300
  • 56cb10d6e5 Min/max scan that can skip unused argument vectors sometimes Marshall Lochbaum 2024-08-19 14:12:40 -0400
  • d185d32d9e remove debug include dzaima 2024-08-20 17:22:10 +0300
  • 357eb1a27a expanded comment of supported things in the native compiler dzaima 2024-08-20 00:17:29 +0300
  • a654bd88d2 fix aarch64 build dzaima 2024-08-17 20:13:38 +0300
  • 0467e1c894 use *arrv_ptr more dzaima 2024-08-15 17:34:52 +0300
  • 597fe07838 another incG dzaima 2024-08-15 17:08:27 +0300
  • 37a3694df7 restore usage of bitarr_ptr where appropriate dzaima 2024-08-15 17:03:03 +0300
  • 5098781ffc replace all bitarr_ptr usages with bitany_ptr dzaima 2024-08-15 17:07:04 +0300
  • 88de336516 introduce bitany_ptr dzaima 2024-08-15 17:28:25 +0300
  • 00a64101e9 better bitp_set dzaima 2024-08-15 05:59:29 +0300
  • 3b2b9bba1a AVX-512 bitwiden dzaima 2024-08-14 22:13:49 +0300
  • 67626d1e49 a couple more AVX-512 defs dzaima 2024-08-14 23:43:46 +0300
  • 6b72139d75 include avx512 in base dzaima 2024-08-14 22:12:58 +0300
  • ab4d3565df more --replxx-read-only dzaima 2024-08-14 21:17:29 +0300
  • 60380e6270 fix out-of-bounds indexing in factors dzaima 2024-08-14 22:10:40 +0300
  • 34ac49fd21
    Merge pull request #116 from mlochbaum/repbit dzaima 2024-08-14 22:09:18 +0300
  • b2566c8f3a NEON k/bool support Marshall Lochbaum 2024-08-14 14:08:35 -0400
  • b801c7c186 Move to dedicated k/bool loops for k=5,7 and at least SSSE3 Marshall Lochbaum 2024-08-14 10:20:21 -0400
  • b2758d355c k/bool cleanup and implementation comments Marshall Lochbaum 2024-08-14 09:34:19 -0400
  • bf69705c82 Factor (8×k)/bool as bit-replicate, then byte-replicate Marshall Lochbaum 2024-08-13 20:52:30 -0400
  • fc1265b3f2 Enable (non-boolean) constrep on SSSE3 Marshall Lochbaum 2024-08-13 20:12:19 -0400
  • 048529740b SSE2 k/bool where k divides 8 implementations Marshall Lochbaum 2024-08-13 15:59:48 -0400
  • ee27717c97 SSE2 odd/bool implementation using [4]u32 shuffles Marshall Lochbaum 2024-08-13 11:42:12 -0400
  • 36e9ca5814 Extend odd/bool SSSE3 code to support AVX2 Marshall Lochbaum 2024-08-11 17:24:19 -0400
  • 2be022921e Combine generic and SSSE3 general-case odd/bool, so SSSE3 uses mask unpacking Marshall Lochbaum 2024-08-10 14:20:22 -0400
  • 3e5dbdbf8d Dedicated generic k/bool for divisors of 8; factor; remove pdep emulation Marshall Lochbaum 2024-08-09 14:44:33 -0400
  • 13ec029d9f Get odd k/bool carry by shifting in the unified register, not after masking Marshall Lochbaum 2024-08-09 08:29:43 -0400
  • 1621c7d07c Generic-arch k/bool algorithm for odd k with mask unpacking Marshall Lochbaum 2024-08-08 18:08:49 -0400
  • 7d7d36b354 Extend SSSE3 k/bool to k<32 using pairwise swaps Marshall Lochbaum 2024-08-07 22:03:55 -0400
  • 696a23af9e Mask off last k/bool vector properly Marshall Lochbaum 2024-08-07 17:38:57 -0400
  • 1a4cada0cb AVX2 support in rep_const_bool_ssse3_div8 Marshall Lochbaum 2024-08-07 10:35:21 -0400
  • b4d84041bc Extend odd k/bool SSSE3 algorithm to k<16, factor even k≤32 Marshall Lochbaum 2024-08-06 22:16:20 -0400
  • a439a0a430 SSSE3 k/bool for 2≤k≤8, k≠6 Marshall Lochbaum 2024-08-06 10:02:27 -0400
  • a4b6d8d827 SSSE3 3/bool, faster than BMI2 Marshall Lochbaum 2024-08-05 15:27:36 -0400
  • 7f6c401eb3 Generic-architecture k/bool for small 8<k Marshall Lochbaum 2024-08-05 11:06:14 -0400
  • 0602927b17 Move pdep-based k/bool code to Singeli Marshall Lochbaum 2024-08-05 09:57:05 -0400
  • de5bebb239 fix bitnarrow on aarch64 dzaima 2024-08-14 16:41:58 +0300
  • e4005e9d7c add SINGELI_SSSE3 dzaima 2024-08-14 15:47:48 +0300
  • f83f63b22d add has=ssse3, has=sse4.1, and has=sse4.2 dzaima 2024-08-14 15:47:02 +0300
  • 565a0a8458 inline zeroPadToCellBits0 dzaima 2024-08-14 15:21:42 +0300
  • fc800ee191 remove debug include dzaima 2024-08-14 15:06:43 +0300
  • ef1ea4ad0f make gcc not complain about break dzaima 2024-08-14 05:23:05 +0300
  • 2ca2c52c16 use direct bitwiden & bitnarrow in select-rows dzaima 2024-08-14 05:00:31 +0300
  • d95a792230 add bitwiden & bitnarrow to header dzaima 2024-08-14 04:45:03 +0300
  • d0f30bd717 extract direct pointer version of bitnarrow dzaima 2024-08-14 04:43:15 +0300
  • d676663f45 better unaligned load option for bitwiden dzaima 2024-08-14 04:19:55 +0300
  • 6e8c6d2e10 extract direct pointer version of bitwiden dzaima 2024-08-14 04:12:26 +0300
  • 9672f07c1b add select-bit-cells test dzaima 2024-08-14 04:30:28 +0300
  • 7eb2b97e29 use COPY_TO_FROM in cpyU16Bits & cpyU8Bits dzaima 2024-08-14 02:52:29 +0300
  • c445f24aae COPY_TO_FROM dzaima 2024-08-14 02:31:51 +0300
  • 9b515c3fad reverse pointer arguments of copy.singeli exports dzaima 2024-08-14 01:12:13 +0300
  • 6d7648c947 more things that could be select-rows'd dzaima 2024-08-10 21:52:16 +0300
  • f6c52ef3b3 disable Singeli bitwiden on SSE2 dzaima 2024-08-14 02:16:17 +0300
  • f356ba0f36 more definitions in base.singeli dzaima 2024-08-09 05:56:09 +0300
  • 3fc92eceaa include util/tup in base.singeli dzaima 2024-08-09 05:44:51 +0300
  • a9460da19c more missing NEON comments dzaima 2024-08-09 05:17:04 +0300
  • eccbac37ab aarch64 8→n bitnarrow dzaima 2024-08-14 02:21:13 +0300
  • c72ed51149 Singeli n→8 bitwiden dzaima 2024-08-14 02:35:31 +0300
  • f36cefc9ac more Singeli utils dzaima 2024-08-08 23:31:37 +0300
  • 5748833060 fast inds⊸⊏˘bits for ≤8-bit input & output cells dzaima 2024-08-08 23:28:43 +0300
  • 920a89f019 native ⋈⁼ & ⌾⋈ dzaima 2024-08-14 05:17:18 +0300
  • ae0d392d43 tyarrv_free dzaima 2024-08-14 05:00:14 +0300
  • 46b63125bc ARBITRARY & GUARANTEED dzaima 2024-08-14 01:57:40 +0300
  • 2c36a3e399 fix ⟨1‿2⟩⊸⊏˘ mat dzaima 2024-08-09 00:36:07 +0300
  • 3716162e69 use q_fbit more dzaima 2024-08-08 23:30:02 +0300
  • e0facc54f8 minor src/README.md cleanup dzaima 2024-08-08 23:29:00 +0300
  • 8e49fc04bb --replxx-read-only dzaima 2024-07-31 03:54:42 +0300
  • ced082cc82 use custom valgrind pdep/pext everywhere dzaima 2024-07-31 03:54:34 +0300
  • d1469df96b fix out-of-bounds load on empty replxx line dzaima 2024-07-31 03:03:11 +0300
  • bdb5781ef7 make bitp_get & bitp_set load/store u8 instead of u64 dzaima 2024-07-29 04:47:32 +0300
  • 64b69700c5 include last power of two in fast-path ⌽˘ dzaima 2024-07-29 02:16:49 +0300
  • 6e085a024b update Singeli submodule dzaima 2024-08-03 22:33:31 +0300
  • 315ab8563a separate fuzz test files dzaima 2024-07-29 01:30:33 +0300
  • bc81b26a03 use fast path for inds⊸⊏˘el_B and ⌽˘el_B dzaima 2024-07-29 01:19:44 +0300
  • 6900ae4b68 fix ⌽˘ 0‿10⥊2 and (↕0)⊸⊏˘ 10‿10⥊2 dzaima 2024-07-29 00:36:57 +0300
  • 52f36c61c0 fix 0‿1⊸⊏˘3‿1⥊2; accept any non-empty array x in select_rows_typed→select_rows_direct dzaima 2024-07-29 00:18:48 +0300
  • 3a722b552f ⌽⎉1 x via (⌽↕¯1⊑≢x)⊸⊏⎉1 x dzaima 2024-07-28 21:43:10 +0300
  • 081b928cb4 fix zip_halves being weird dzaima 2024-07-28 05:07:18 +0300
  • ffc250eb0e clearer aarch64 zip_halves mzip usage dzaima 2024-07-28 04:01:34 +0300
  • 39f4a66ae0 cleanup dzaima 2024-07-28 03:01:23 +0300
  • bd3789963c these are actually good dzaima 2024-07-28 02:51:23 +0300
  • 6f25e76e5d tune NEON inds⊸⊏˘f64arr dzaima 2024-07-28 02:46:08 +0300
  • 8fd522cb6a a bit more inds⊸⊏⎉k testing dzaima 2024-07-28 00:55:37 +0300
  • d1d8e8be03 fix assertion on F⌾(⟨⟩⊸⊏) x dzaima 2024-07-26 23:24:42 +0300
  • 42b8420af6 ignore unused label warning dzaima 2024-07-26 23:13:08 +0300
  • 93fca7c628 fix overflow in index widening dzaima 2024-07-26 20:56:58 +0300
  • 7cc1031f70 use outside-of-loop index widening on aarch64 dzaima 2024-07-26 05:13:18 +0300
  • a486e87451 select-cells comments dzaima 2024-07-26 21:51:34 +0300
  • 69ca524251 widen inds outside select-cells loops dzaima 2024-07-26 22:16:52 +0300