1 /*
2 * Copyright (c) 2022 Arm Limited.
3 *
4 * SPDX-License-Identifier: MIT
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
21 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
22 * IN THE SOFTWARE.
23 */
24
25 #if defined(__ARM_FEATURE_SVE)
26
27 template <>
interleave_block(int8_t * & out,const int8_t * const * in,size_t width,size_t height,size_t row_offset,bool first)28 void interleave_block<4, 4, VLType::SME, false>(
29 int8_t * &out, const int8_t * const *in,
30 size_t width, size_t height, size_t row_offset, bool first
31 )
32 {
33 ARM_COMPUTE_UNUSED(first);
34
35 __asm__ __volatile__(
36 ".inst 0xd503477f // SMSTART ZA\n"
37 "cntw x15\n"
38 "cntw x14\n"
39 "cntw x13, ALL, MUL #2\n"
40 "cntw x11, ALL, MUL #3\n"
41 "cmp %x[height], x15\n"
42 "csel x15, %x[height], x15, LT\n"
43 "whilelt p12.b, XZR, %x[height]\n"
44 "whilelt p10.b, x14, %x[height]\n"
45 "whilelt p9.b, x13, %x[height]\n"
46 "whilelt p8.b, x11, %x[height]\n"
47 "zip1 p12.b, p12.b, p9.b\n"
48 "zip1 p10.b, p10.b, p8.b\n"
49 "mov x10, #0x0\n"
50 "cntb x9\n"
51 "ptrue p11.s\n"
52 "sub x15, x15, #0x1\n"
53 "zip1 p10.b, p12.b, p10.b\n"
54 "mov x28, %x[row_offset]\n"
55 "mov x27, %x[out]\n"
56 "whilelt p9.b, x10, %x[width]\n"
57 "whilelt p8.b, x10, %x[width]\n"
58 "1:" // Width loop
59 "add x26, %x[in], XZR, LSL #3\n"
60 "add x25, %x[in], x14, LSL #3\n"
61 "ldr x24, [x26], #0x8\n"
62 "add x23, %x[in], x13, LSL #3\n"
63 "add x22, %x[in], x11, LSL #3\n"
64 "ldr x19, [x25], #0x8\n"
65 "mov x12, #0x0\n"
66 "ldr x21, [x23], #0x8\n"
67 "ldr x20, [x22], #0x8\n"
68 "cbz x15, 3f\n"
69 "2:" // Loads: Loop
70 ".inst 0x25246140 // psel p0.b, p8.b/Z, p10.b[w12]\n"
71 ".inst 0xe01c0300 // ld1b { za0h.b[x12] }, p0/Z, [x24, x28]\n"
72 ".inst 0x252c6140 // psel p0.b, p8.b/Z, p10.b[w12, #1]\n"
73 "ldr x24, [x26], #0x8\n"
74 ".inst 0xe01c0261 // ld1b { za0h.b[x12, #1] }, p0/Z, [x19, x28]\n"
75 ".inst 0x25346141 // psel p1.b, p8.b/Z, p10.b[w12, #2]\n"
76 ".inst 0x253c6140 // psel p0.b, p8.b/Z, p10.b[w12, #3]\n"
77 "ldr x19, [x25], #0x8\n"
78 ".inst 0xe01c06a2 // ld1b { za0h.b[x12, #2] }, p1/Z, [x21, x28]\n"
79 "ldr x21, [x23], #0x8\n"
80 ".inst 0xe01c0283 // ld1b { za0h.b[x12, #3] }, p0/Z, [x20, x28]\n"
81 "add x12, x12, #0x4\n"
82 "cmp x12, x15, LSL #2\n"
83 "ldr x20, [x22], #0x8\n"
84 "blt 2b\n"
85 "3:" // Loads: Tail
86 ".inst 0x25246140 // psel p0.b, p8.b/Z, p10.b[w12]\n"
87 ".inst 0xe01c0300 // ld1b { za0h.b[x12] }, p0/Z, [x24, x28]\n"
88 ".inst 0x252c6140 // psel p0.b, p8.b/Z, p10.b[w12, #1]\n"
89 ".inst 0xe01c0261 // ld1b { za0h.b[x12, #1] }, p0/Z, [x19, x28]\n"
90 ".inst 0x25346140 // psel p0.b, p8.b/Z, p10.b[w12, #2]\n"
91 "sub x19, %x[width], x10\n"
92 ".inst 0xe01c02a2 // ld1b { za0h.b[x12, #2] }, p0/Z, [x21, x28]\n"
93 "cmp x19, x9\n"
94 "csel x19, x19, x9, LT\n"
95 ".inst 0x253c6140 // psel p0.b, p8.b/Z, p10.b[w12, #3]\n"
96 "add x19, x19, #0x3\n"
97 ".inst 0xe01c0283 // ld1b { za0h.b[x12, #3] }, p0/Z, [x20, x28]\n"
98 "mov x12, #0x0\n"
99 "lsr x19, x19, #0x2\n"
100 "4:" // Stores: Loop
101 ".inst 0x25306d20 // psel p0.s, p11.s/Z, p9.s[w12]\n"
102 ".inst 0xe0bf8360 // st1w { za0v.s[x12] }, p0/Z, [x27, XZR, LSL #2]\n"
103 ".inst 0x25306d20 // psel p0.s, p11.s/Z, p9.s[w12]\n"
104 ".inst 0xe0ae8364 // st1w { za1v.s[x12] }, p0/Z, [x27, x14, LSL #2]\n"
105 ".inst 0x25306d21 // psel p1.s, p11.s/Z, p9.s[w12]\n"
106 ".inst 0x25306d20 // psel p0.s, p11.s/Z, p9.s[w12]\n"
107 ".inst 0xe0ad8768 // st1w { za2v.s[x12] }, p1/Z, [x27, x13, LSL #2]\n"
108 ".inst 0xe0ab836c // st1w { za3v.s[x12] }, p0/Z, [x27, x11, LSL #2]\n"
109 "add x12, x12, #0x1\n"
110 "cmp x12, x19\n"
111 "addvl x27, x27, #4\n"
112 "blt 4b\n"
113 "incb x10\n"
114 "whilelt p9.b, x10, %x[width]\n"
115 "whilelt p8.b, x10, %x[width]\n"
116 "incb x28\n"
117 "b.any 1b\n"
118 "mov %x[out], x27\n"
119 ".inst 0xd503467f // SMSTOP\n"
120 : [out] "+&r" (out)
121 : [height] "r" (height), [in] "r" (in), [row_offset] "r" (row_offset), [width] "r" (width)
122 : "cc", "memory", "p0", "p1", "p2", "p3", "p4", "p5", "p6", "p7", "p8", "p9", "p10", "p11", "p12", "p13", "p14", "p15", "x9", "x10", "x11", "x12", "x13", "x14", "x15", "x19", "x20", "x21", "x22", "x23", "x24", "x25", "x26", "x27", "x28", "z0", "z1", "z2", "z3", "z4", "z5", "z6", "z7", "z8", "z9", "z10", "z11", "z12", "z13", "z14", "z15", "z16", "z17", "z18", "z19", "z20", "z21", "z22", "z23", "z24", "z25", "z26", "z27", "z28", "z29", "z30", "z31"
123 );
124 }
125
126 #endif // defined(__ARM_FEATURE_SVE)
127