Commit Graph

1849 Commits

Author SHA1 Message Date
dzaima
9826201343 move neon.singeli type checks to w64/w128 2023-04-01 22:53:38 +03:00
dzaima
a62c21733e SSE2 equal.singeli 2023-04-01 22:08:47 +03:00
dzaima
96ee73f644 enable Singeli monadic arith & comparisons on SSE2 2023-04-01 20:40:29 +03:00
dzaima
3a45eac1aa separate SSE2, implement SSE2 dyarith & bitsel
also separate maskstore to homogeneous & top bit mask variants, and make blend require equal element widths
2023-04-01 20:40:23 +03:00
dzaima
13eebb1759 move all singeli builds to build.bqn, add has=
also:
include commit-based version by default for build/build
expand things included in --version
don't require Singeli for BMI2 usage in bitarr/bitarr
2023-03-31 19:15:23 +03:00
dzaima
dc2ac37e0a arch-independent Singeli builds 2023-03-31 16:45:29 +03:00
dzaima
0180a54207 quiet gcc warning 2023-03-31 16:34:08 +03:00
dzaima
c513724fd4 hide clangd warning
and remove whitespace which was previously used for alignment
2023-03-31 14:27:43 +03:00
dzaima
baa1281efe don't assume select returns refcount 1 array
that probably would be a safe assumption right now, but who knows about the future; cpyWithShape should return back its argument directly if it already has refcount 1 anyway
2023-03-31 14:14:13 +03:00
dzaima
413b49de30 don't assume sizeof(ur)==1 2023-03-31 13:44:32 +03:00
Marshall Lochbaum
69e2f6c690 Convert ⍉˘ with atom or no left argument to dyadic ⍉ 2023-03-30 21:25:21 -04:00
Marshall Lochbaum
9fe6a2e7b2 Under for invertible Reorder Axes 2023-03-30 19:07:09 -04:00
Marshall Lochbaum
d53f3dbd72 Native dyadic ⍉⁼ 2023-03-30 19:02:45 -04:00
Marshall Lochbaum
89016fc88d Update Reorder Axes implementation comments 2023-03-29 20:23:16 -04:00
Marshall Lochbaum
597e25af4d Full simplification pass for Reorder Axes 2023-03-29 20:08:36 -04:00
Marshall Lochbaum
d492cd0865 And strided height, so all orders are fast if axes are long 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
e923a71881 Use strided width in transpose-based Reorder Axes 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
814a677676 Make transpose movement functions strided 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
de18fb996b Save a shape object in Reorder Axes 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
7fea4ca2ad Combine transp_c2 temp allocations into one TALLOC 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
5b17ee44d6 SIMD kernel for Reorder Axes transposing last two axes 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
8413461074 Fast bottom-level loop for Reorder Axes loop structure 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
bbedbbc585 Transpose large cells with mut_copyG 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
b829076271 Put CPU-size transpose functions in a table 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
410d51501d Check for AVX2 in Singeli transpose so it can build on non-x86 architectures 2023-03-29 12:37:15 -04:00
Marshall Lochbaum
bf04b3aab0 Move w128i and similar from sse3 and avx.singeli to base 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
ebbddf9dcf Move height/width 2 transposes to Singeli 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
37f481ea0d Separate out transpose data movement function and share with monadic ⍉⁼ 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
3c32844fe0 Separate out transpose_move function and use for el_B 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
968bfd2fc6 Separate transpose el_B case and fix dropped fill for shape 2‿n 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
dea063cb7b Easy optimizations: (1-˜=𝕩)⍉𝕩 to ⍉𝕩 and detect trailing unchanged axes 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
b555e3c035 Reorder Axes translated from runtime version 2023-03-28 16:11:46 -04:00
Marshall Lochbaum
432b4eaaa6 Transpose implementation comments 2023-03-28 16:11:46 -04:00
dzaima
21b9ad5c3c don't include debug.singeli 2023-03-27 17:53:01 +03:00
dzaima
9780cacacd commented-out heap alignment 2023-03-27 17:38:26 +03:00
dzaima
f175bc9e19 fix printf of sample count 2023-03-26 15:57:09 +03:00
dzaima
1f2456042c clean up calls.h includes 2023-03-26 15:54:27 +03:00
dzaima
1f70c08aa2 fillarr NOGC 2023-03-26 15:43:59 +03:00
dzaima
709c1896d3 better selfElType 2023-03-26 14:30:49 +03:00
dzaima
37a32eb15a fix native non-Singeli builds 2023-03-24 23:51:18 +02:00
dzaima
830c752e87 instruction pointer storing profiling 2023-03-24 18:07:48 +02:00
dzaima
e4c95dce22 fix profiler buffer size usage 2023-03-24 13:47:33 +02:00
dzaima
98f4df88d0 whitespace 2023-03-23 18:35:55 +02:00
dzaima
fa8437ca25 don't use _mm256_loadu2_m128i
doesn't exist on old gcc
2023-03-23 18:21:34 +02:00
Marshall Lochbaum
699faea69c Dedicated width-overflow loop for SIMD transpose 2023-03-22 17:34:37 -04:00
Marshall Lochbaum
5a2bc15f2a Overlap SIMD transpose on width too 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
61eefe0776 Fix writing past end in i16 transpose with overlapped halves 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
8ff40cb2e5 Overlap SIMD transpose to handle uneven heights (not widths yet) 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
fad7f3aa8b Simplify 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
dd141add3f Merge i16 SIMD transpose with the rest of them 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
9673771187 Combine code for the different transpose kernel geometries 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
e40feaa81c Make transpose kernel code type-generic 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
923c485cc2 Use packed-halves transpose kernel for 8x8 i16 2023-03-22 15:14:09 -04:00
Marshall Lochbaum
2a8e6e9c8a Separate most transpose kernel logic from loops and move i8 case into main 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
7d800bf349 SIMD transpose on 1-byte elements 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
5fccf4cda3 Refactoring; check for small transpose width/height in Singeli, not C 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
ff6361e363 Move tranpose to its own file 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
0e2e315e51 Allow an 8x8 row for 2-byte transpose 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
239c765998 SIMD transpose on 2-byte elements 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
f6c6e72661 More transpose kernel refactoring 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
c0aaa6f615 SIMD transpose on 8-byte elements 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
a0e85db702 Refactor vtranspose with more general code 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
ca44d41daa Faster i32 transpose when height is a multiple of 16 by writing cache lines 2023-03-22 15:14:08 -04:00
Marshall Lochbaum
bfb2519aed Remove transpose_rec (was unused): the whole cache-oblivious thing seems to be useless 2023-03-22 15:14:08 -04:00
dzaima
5c6075c29d support namespace a.b in native compiler
and use for •file.At in bootstrap.bqn
2023-03-22 19:58:01 +02:00
dzaima
dc583d754c separate FLDG for gid-based field lookup 2023-03-22 19:49:32 +02:00
dzaima
9337de5a0e fix LST_/ARM_ comments in vm.h 2023-03-21 23:04:39 +02:00
dzaima
d9ebdf060f )t:1s
also fix )profile
2023-03-21 04:06:53 +02:00
dzaima
fe36c74080 support exponential for )t 2023-03-21 02:56:39 +02:00
dzaima
1ec37410c6 faster constint|intarr 2023-03-21 02:45:42 +02:00
dzaima
d333902f74 fast integer 𝕨|𝕩 2023-03-21 02:06:45 +02:00
dzaima
e14a3629f6 don't try lowercasing × 2023-03-20 18:18:16 +02:00
dzaima
b0201131e2 array type support for •FFI 2023-03-20 16:44:58 +02:00
dzaima
ca0c100b73 name normalization for •ns.Get & •ns.Has
also fix •ns.Has error message & rename some function declarations
2023-03-19 23:27:05 +02:00
dzaima
0b32ef96ce inline → inplace in existing-scope-mutating function names 2023-03-19 22:47:00 +02:00
dzaima
4a6877a87a Rank checks 2023-03-19 15:42:11 +02:00
dzaima
a3d74c6af2 MUT_APPEND 2023-03-16 22:27:44 +02:00
dzaima
390ee503fb extract function for a+↕b 2023-03-16 22:05:44 +02:00
dzaima
a9b83ca70a V_POS, V_DEPTH 2023-03-16 22:03:08 +02:00
dzaima
db8a4d3af0 don't use def as a variable name 2023-03-12 00:11:30 +02:00
dzaima
1178e5df56 new Singeli syntax 2023-03-11 23:17:30 +02:00
dzaima
8194f51b30 MAKE_MUT_INIT, move mut_init to mut.c 2023-03-11 16:52:09 +02:00
dzaima
3f0bc57503 move diagnostics coloring option outside of hash 2023-03-11 16:51:59 +02:00
dzaima
01c457b77b make all cpy.*Arr function results Arr* 2023-03-11 16:51:13 +02:00
dzaima
92763fa8b2 proper hash values for derived functions 2023-03-06 20:29:35 +02:00
dzaima
ab5b8f92b4 fast F˜˝˘ and F˜˝ and F˜⌜ for pervasive F 2023-03-05 16:48:10 +02:00
dzaima
680d0f8601 use mut_copy for building fold_rows result 2023-03-05 15:20:12 +02:00
dzaima
1c0b797456
Merge pull request #69 from mlochbaum/transfold
Blocked transpose implementation for fold-cells
2023-03-05 15:10:26 +02:00
dzaima
98c963453b •internal.GC 2023-03-05 14:16:38 +02:00
dzaima
4b6dd1bf81 immediately mark&sweep unknown reference objects instead of storing 2023-03-05 14:05:08 +02:00
dzaima
c5aedaa414 switch to GCv2, various fixes/cleanup
includes:
  fixing some refcounts for FOR_BUILD build •FFI
  moving DEBUG_VM prints to stderr
  fixing heapverify reports of loose shape objects of fillarrs
  regular ol' adding-of-semicolon-after-label for old gcc
  move GC enable to after all initialization, not load_init
2023-03-05 01:23:53 +02:00
Marshall Lochbaum
f4d2e26e58 Don't do transposing fold unless height is greater than width 2023-03-04 14:42:00 -05:00
Marshall Lochbaum
176152b445 Handle fold-cells on row length 1 with ⊏˘ 2023-03-04 14:28:22 -05:00
Marshall Lochbaum
4b35acb82b Transpose-based fold-cells implementation 2023-03-04 14:24:56 -05:00
dzaima
16274e5952 bucket coalescing on GC 2023-03-04 20:09:12 +02:00
dzaima
ead637b135 don't leak c1fn/c2fn arguments from array function 2023-03-04 18:58:07 +02:00
dzaima
717074a2f8 fast transpose on shape n‿2 𝕩 2023-03-03 19:13:15 +02:00
dzaima
76d26db4c2 faster F˘´n‿2⥊𝕩 for pervasive F 2023-03-03 17:45:16 +02:00
dzaima
fd72d416db manual prefix character typing detecting 2023-03-03 16:11:55 +02:00
dzaima
093beda394 faster path for 𝕨/𝕩 with non-boolean 𝕨 with sum 0 2023-03-03 13:41:20 +02:00