blob: e2a1960bd038f0bdb318642de1cf2e08731b47b5 [file] [log] [blame]
Marat Dukhanf9480682020-07-31 14:50:24 -07001#!/bin/sh
2# Copyright 2020 Google LLC
3#
4# This source code is licensed under the BSD-style license found in the
5# LICENSE file in the root directory of this source tree.
6
Marat Dukhan27203da2020-08-05 15:19:03 -07007################################## WAsm SIMD ##################################
Marat Dukhanef4ce312020-09-10 12:29:08 -07008### C8 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -07009tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=1 -D VARIANT=LD64 -o src/qs8-igemm/gen/1x4c8-minmax-wasmsimd-ld64.c
10tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=2 -D VARIANT=LD64 -o src/qs8-igemm/gen/2x4c8-minmax-wasmsimd-ld64.c
11tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=3 -D VARIANT=LD64 -o src/qs8-igemm/gen/3x4c8-minmax-wasmsimd-ld64.c
Marat Dukhan27203da2020-08-05 15:19:03 -070012
Marat Dukhan66ccf642020-09-28 16:23:42 -070013tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=1 -D VARIANT=LD128 -o src/qs8-igemm/gen/1x4c8-minmax-wasmsimd-ld128.c
14tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=2 -D VARIANT=LD128 -o src/qs8-igemm/gen/2x4c8-minmax-wasmsimd-ld128.c
15tools/xngen src/qs8-igemm/MRx4c8-wasmsimd.c.in -D MR=3 -D VARIANT=LD128 -o src/qs8-igemm/gen/3x4c8-minmax-wasmsimd-ld128.c
Marat Dukhan27203da2020-08-05 15:19:03 -070016
Marat Dukhan40bbafe2020-08-04 02:04:22 -070017################################### ARM NEON ##################################
Frank Barchardda78da12021-03-02 14:28:00 -080018tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=1 -D NR=8 -o src/qs8-igemm/gen/1x8-minmax-neon-mlal-lane.c
19tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=2 -D NR=8 -o src/qs8-igemm/gen/2x8-minmax-neon-mlal-lane.c
20tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=3 -D NR=8 -o src/qs8-igemm/gen/3x8-minmax-neon-mlal-lane.c
21tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=4 -D NR=8 -o src/qs8-igemm/gen/4x8-minmax-neon-mlal-lane.c
22
Marat Dukhan66ccf642020-09-28 16:23:42 -070023tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=1 -D NR=16 -o src/qs8-igemm/gen/1x16-minmax-neon-mlal-lane.c
24tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=2 -D NR=16 -o src/qs8-igemm/gen/2x16-minmax-neon-mlal-lane.c
Frank Barchardec0bf142021-01-15 10:14:05 -080025tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=3 -D NR=16 -o src/qs8-igemm/gen/3x16-minmax-neon-mlal-lane.c
26tools/xngen src/qs8-igemm/neon-mlal-lane.c.in -D MR=4 -D NR=16 -o src/qs8-igemm/gen/4x16-minmax-neon-mlal-lane.c
Marat Dukhan40bbafe2020-08-04 02:04:22 -070027
Frank Barchardda78da12021-03-02 14:28:00 -080028tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=1 -D NR=8 -o src/qs8-igemm/gen/1x8-minmax-neon-mull-addw-dup.c
29tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=2 -D NR=8 -o src/qs8-igemm/gen/2x8-minmax-neon-mull-addw-dup.c
30tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=3 -D NR=8 -o src/qs8-igemm/gen/3x8-minmax-neon-mull-addw-dup.c
31tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=4 -D NR=8 -o src/qs8-igemm/gen/4x8-minmax-neon-mull-addw-dup.c
32
33tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=1 -D NR=16 -o src/qs8-igemm/gen/1x16-minmax-neon-mull-addw-dup.c
34tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=2 -D NR=16 -o src/qs8-igemm/gen/2x16-minmax-neon-mull-addw-dup.c
35tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=3 -D NR=16 -o src/qs8-igemm/gen/3x16-minmax-neon-mull-addw-dup.c
36tools/xngen src/qs8-igemm/neon-mull-addw-dup.c.in -D MR=4 -D NR=16 -o src/qs8-igemm/gen/4x16-minmax-neon-mull-addw-dup.c
37
38### C2 micro-kernels
39tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=1 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/1x8c2-minmax-neon-mull-padal-dup.c
40tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=2 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/2x8c2-minmax-neon-mull-padal-dup.c
41tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=3 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/3x8c2-minmax-neon-mull-padal-dup.c
42tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=4 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/4x8c2-minmax-neon-mull-padal-dup.c
43tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=1 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/1x16c2-minmax-neon-mull-padal-dup.c
44tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=2 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/2x16c2-minmax-neon-mull-padal-dup.c
45tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=3 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/3x16c2-minmax-neon-mull-padal-dup.c
46tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=4 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/4x16c2-minmax-neon-mull-padal-dup.c
47
48tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=1 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/1x8c2-minmax-neon-mlal-padal-dup.c
49tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=2 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/2x8c2-minmax-neon-mlal-padal-dup.c
50tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=3 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/3x8c2-minmax-neon-mlal-padal-dup.c
51tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=4 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/4x8c2-minmax-neon-mlal-padal-dup.c
52tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=1 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/1x16c2-minmax-neon-mlal-padal-dup.c
53tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=2 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/2x16c2-minmax-neon-mlal-padal-dup.c
54tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=3 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/3x16c2-minmax-neon-mlal-padal-dup.c
55tools/xngen src/qs8-igemm/c2-neon-mull-padal-dup.c.in -D MR=4 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/4x16c2-minmax-neon-mlal-padal-dup.c
56
Frank Barchardd79391d2021-02-15 14:22:32 -080057### C8 micro-kernels
Frank Barchardda78da12021-03-02 14:28:00 -080058tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=1 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/1x8c8-minmax-neon-mull-padal.c
59tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=2 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/2x8c8-minmax-neon-mull-padal.c
60tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=3 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/3x8c8-minmax-neon-mull-padal.c
61tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=4 -D NR=8 -D MLA=0 -o src/qs8-igemm/gen/4x8c8-minmax-neon-mull-padal.c
62tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=1 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/1x16c8-minmax-neon-mull-padal.c
63tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=2 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/2x16c8-minmax-neon-mull-padal.c
64tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=3 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/3x16c8-minmax-neon-mull-padal.c
65tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=4 -D NR=16 -D MLA=0 -o src/qs8-igemm/gen/4x16c8-minmax-neon-mull-padal.c
66
67tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=1 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/1x8c8-minmax-neon-mlal-padal.c
68tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=2 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/2x8c8-minmax-neon-mlal-padal.c
69tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=3 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/3x8c8-minmax-neon-mlal-padal.c
70tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=4 -D NR=8 -D MLA=1 -o src/qs8-igemm/gen/4x8c8-minmax-neon-mlal-padal.c
71tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=1 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/1x16c8-minmax-neon-mlal-padal.c
72tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=2 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/2x16c8-minmax-neon-mlal-padal.c
73tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=3 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/3x16c8-minmax-neon-mlal-padal.c
74tools/xngen src/qs8-igemm/c8-neon-mull-padal.c.in -D MR=4 -D NR=16 -D MLA=1 -o src/qs8-igemm/gen/4x16c8-minmax-neon-mlal-padal.c
Frank Barchardd79391d2021-02-15 14:22:32 -080075
Frank Barchard36f95cf2021-02-23 03:12:52 -080076### C16 micro-kernels
77tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=1 -D NR=8 -o src/qs8-igemm/gen/1x8c16-minmax-neon-mlal-padal.c
78tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=2 -D NR=8 -o src/qs8-igemm/gen/2x8c16-minmax-neon-mlal-padal.c
79tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=3 -D NR=8 -o src/qs8-igemm/gen/3x8c16-minmax-neon-mlal-padal.c
80tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=4 -D NR=8 -o src/qs8-igemm/gen/4x8c16-minmax-neon-mlal-padal.c
81tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=1 -D NR=16 -o src/qs8-igemm/gen/1x16c16-minmax-neon-mlal-padal.c
82tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=2 -D NR=16 -o src/qs8-igemm/gen/2x16c16-minmax-neon-mlal-padal.c
83tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=3 -D NR=16 -o src/qs8-igemm/gen/3x16c16-minmax-neon-mlal-padal.c
84tools/xngen src/qs8-igemm/c16-neon-mlal-padal.c.in -D MR=4 -D NR=16 -o src/qs8-igemm/gen/4x16c16-minmax-neon-mlal-padal.c
85
Benoit Jacoba05487f2020-08-14 13:37:21 -070086### C4 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -070087tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=1 -D NR=8 -o src/qs8-igemm/gen/1x8c4-minmax-neondot.c
88tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=4 -D NR=8 -o src/qs8-igemm/gen/4x8c4-minmax-neondot.c
89tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=6 -D NR=8 -o src/qs8-igemm/gen/6x8c4-minmax-neondot.c
90tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=8 -D NR=8 -o src/qs8-igemm/gen/8x8c4-minmax-neondot.c
Marat Dukhan66ccf642020-09-28 16:23:42 -070091tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=1 -D NR=16 -o src/qs8-igemm/gen/1x16c4-minmax-neondot.c
92tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=4 -D NR=16 -o src/qs8-igemm/gen/4x16c4-minmax-neondot.c
93tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=6 -D NR=16 -o src/qs8-igemm/gen/6x16c4-minmax-neondot.c
94tools/xngen src/qs8-igemm/MRxNRc4-neondot.c.in -D MR=8 -D NR=16 -o src/qs8-igemm/gen/8x16c4-minmax-neondot.c
95
Marat Dukhanf9480682020-07-31 14:50:24 -070096################################### x86 SSE ###################################
97### C2 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -070098tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=2 -D LD128=0 -o src/qs8-igemm/gen/1x4c2-minmax-sse2-ld64.c
99tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=2 -D LD128=0 -o src/qs8-igemm/gen/4x4c2-minmax-sse2-ld64.c
Marat Dukhanf9480682020-07-31 14:50:24 -0700100
Marat Dukhan66ccf642020-09-28 16:23:42 -0700101tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=3 -D LD128=0 -o src/qs8-igemm/gen/1x4c2-minmax-ssse3-ld64.c
102tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=3 -D LD128=0 -o src/qs8-igemm/gen/4x4c2-minmax-ssse3-ld64.c
Marat Dukhanf9480682020-07-31 14:50:24 -0700103
Marat Dukhan66ccf642020-09-28 16:23:42 -0700104tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=4 -D LD128=0 -o src/qs8-igemm/gen/1x4c2-minmax-sse41-ld64.c
105tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=4 -D LD128=0 -o src/qs8-igemm/gen/4x4c2-minmax-sse41-ld64.c
Marat Dukhanf9480682020-07-31 14:50:24 -0700106
Marat Dukhan66ccf642020-09-28 16:23:42 -0700107tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=5 -D LD128=0 -o src/qs8-igemm/gen/1x4c2-minmax-xop-ld64.c
108tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=5 -D LD128=0 -o src/qs8-igemm/gen/4x4c2-minmax-xop-ld64.c
Marat Dukhan1566fee2020-08-02 21:55:41 -0700109
Marat Dukhan66ccf642020-09-28 16:23:42 -0700110tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=2 -D LD128=1 -o src/qs8-igemm/gen/1x4c2-minmax-sse2-ld128.c
111tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=2 -D LD128=1 -o src/qs8-igemm/gen/4x4c2-minmax-sse2-ld128.c
Marat Dukhandee732b2020-07-31 17:32:04 -0700112
Marat Dukhan66ccf642020-09-28 16:23:42 -0700113tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=3 -D LD128=1 -o src/qs8-igemm/gen/1x4c2-minmax-ssse3-ld128.c
114tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=3 -D LD128=1 -o src/qs8-igemm/gen/4x4c2-minmax-ssse3-ld128.c
Marat Dukhandee732b2020-07-31 17:32:04 -0700115
Marat Dukhan66ccf642020-09-28 16:23:42 -0700116tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=4 -D LD128=1 -o src/qs8-igemm/gen/1x4c2-minmax-sse41-ld128.c
117tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=4 -D LD128=1 -o src/qs8-igemm/gen/4x4c2-minmax-sse41-ld128.c
Marat Dukhandee732b2020-07-31 17:32:04 -0700118
Marat Dukhan66ccf642020-09-28 16:23:42 -0700119tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=1 -D SSE=5 -D LD128=1 -o src/qs8-igemm/gen/1x4c2-minmax-xop-ld128.c
120tools/xngen src/qs8-igemm/MRx4c2-sse.c.in -D MR=4 -D SSE=5 -D LD128=1 -o src/qs8-igemm/gen/4x4c2-minmax-xop-ld128.c
Marat Dukhan1566fee2020-08-02 21:55:41 -0700121
Marat Dukhan07bd2522020-07-31 19:12:39 -0700122### C8 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -0700123tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=2 -D LD128=0 -o src/qs8-igemm/gen/1x4c8-minmax-sse2-ld64.c
124tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=2 -D LD128=0 -o src/qs8-igemm/gen/2x4c8-minmax-sse2-ld64.c
125tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=2 -D LD128=0 -o src/qs8-igemm/gen/3x4c8-minmax-sse2-ld64.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700126
Marat Dukhan66ccf642020-09-28 16:23:42 -0700127tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=3 -D LD128=0 -o src/qs8-igemm/gen/1x4c8-minmax-ssse3-ld64.c
128tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=3 -D LD128=0 -o src/qs8-igemm/gen/2x4c8-minmax-ssse3-ld64.c
129tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=3 -D LD128=0 -o src/qs8-igemm/gen/3x4c8-minmax-ssse3-ld64.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700130
Marat Dukhan66ccf642020-09-28 16:23:42 -0700131tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=4 -D LD128=0 -o src/qs8-igemm/gen/1x4c8-minmax-sse41-ld64.c
132tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=4 -D LD128=0 -o src/qs8-igemm/gen/2x4c8-minmax-sse41-ld64.c
133tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=4 -D LD128=0 -o src/qs8-igemm/gen/3x4c8-minmax-sse41-ld64.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700134
Marat Dukhan66ccf642020-09-28 16:23:42 -0700135tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=5 -D LD128=0 -o src/qs8-igemm/gen/1x4c8-minmax-xop-ld64.c
136tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=5 -D LD128=0 -o src/qs8-igemm/gen/2x4c8-minmax-xop-ld64.c
137tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=5 -D LD128=0 -o src/qs8-igemm/gen/3x4c8-minmax-xop-ld64.c
Marat Dukhan1566fee2020-08-02 21:55:41 -0700138
Marat Dukhan66ccf642020-09-28 16:23:42 -0700139tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=2 -D LD128=1 -o src/qs8-igemm/gen/1x4c8-minmax-sse2-ld128.c
140tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=2 -D LD128=1 -o src/qs8-igemm/gen/2x4c8-minmax-sse2-ld128.c
141tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=2 -D LD128=1 -o src/qs8-igemm/gen/3x4c8-minmax-sse2-ld128.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700142
Marat Dukhan66ccf642020-09-28 16:23:42 -0700143tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=3 -D LD128=1 -o src/qs8-igemm/gen/1x4c8-minmax-ssse3-ld128.c
144tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=3 -D LD128=1 -o src/qs8-igemm/gen/2x4c8-minmax-ssse3-ld128.c
145tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=3 -D LD128=1 -o src/qs8-igemm/gen/3x4c8-minmax-ssse3-ld128.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700146
Marat Dukhan66ccf642020-09-28 16:23:42 -0700147tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=4 -D LD128=1 -o src/qs8-igemm/gen/1x4c8-minmax-sse41-ld128.c
148tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=4 -D LD128=1 -o src/qs8-igemm/gen/2x4c8-minmax-sse41-ld128.c
149tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=4 -D LD128=1 -o src/qs8-igemm/gen/3x4c8-minmax-sse41-ld128.c
Marat Dukhan07bd2522020-07-31 19:12:39 -0700150
Marat Dukhan66ccf642020-09-28 16:23:42 -0700151tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=1 -D SSE=5 -D LD128=1 -o src/qs8-igemm/gen/1x4c8-minmax-xop-ld128.c
152tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=2 -D SSE=5 -D LD128=1 -o src/qs8-igemm/gen/2x4c8-minmax-xop-ld128.c
153tools/xngen src/qs8-igemm/MRx4c8-sse.c.in -D MR=3 -D SSE=5 -D LD128=1 -o src/qs8-igemm/gen/3x4c8-minmax-xop-ld128.c
Marat Dukhan1566fee2020-08-02 21:55:41 -0700154
Marat Dukhan12809522020-08-02 22:23:51 -0700155################################### x86 AVX2 ##################################
156### C8 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -0700157tools/xngen src/qs8-igemm/MRx8c8-avx2.c.in -D MR=1 -o src/qs8-igemm/gen/1x8c8-minmax-avx2.c
158tools/xngen src/qs8-igemm/MRx8c8-avx2.c.in -D MR=2 -o src/qs8-igemm/gen/2x8c8-minmax-avx2.c
159tools/xngen src/qs8-igemm/MRx8c8-avx2.c.in -D MR=3 -o src/qs8-igemm/gen/3x8c8-minmax-avx2.c
Marat Dukhan12809522020-08-02 22:23:51 -0700160
Marat Dukhanbb00b1d2020-08-10 11:37:23 -0700161################################## x86 AVX512 #################################
162### C8 micro-kernels
Marat Dukhan66ccf642020-09-28 16:23:42 -0700163tools/xngen src/qs8-igemm/MRx16c8-avx512skx.c.in -D MR=1 -D VARIANT=LD256 -o src/qs8-igemm/gen/1x16c8-minmax-avx512skx.c
164tools/xngen src/qs8-igemm/MRx16c8-avx512skx.c.in -D MR=2 -D VARIANT=LD256 -o src/qs8-igemm/gen/2x16c8-minmax-avx512skx.c
165tools/xngen src/qs8-igemm/MRx16c8-avx512skx.c.in -D MR=3 -D VARIANT=LD256 -o src/qs8-igemm/gen/3x16c8-minmax-avx512skx.c
166tools/xngen src/qs8-igemm/MRx16c8-avx512skx.c.in -D MR=4 -D VARIANT=LD256 -o src/qs8-igemm/gen/4x16c8-minmax-avx512skx.c
Marat Dukhanbb00b1d2020-08-10 11:37:23 -0700167
Marat Dukhanf9480682020-07-31 14:50:24 -0700168################################## Unit tests #################################
169tools/generate-gemm-test.py --spec test/qs8-igemm-minmax.yaml --output test/qs8-igemm-minmax.cc