diff options
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp')
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp | 269 |
1 files changed, 269 insertions, 0 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp b/src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp new file mode 100644 index 0000000000..5f4fcac690 --- /dev/null +++ b/src/core/NEON/kernels/arm_gemm/kernels/a64_ffinterleaved_bf16fp32_dot_8x12/generic.cpp @@ -0,0 +1,269 @@ +/* + * Copyright (c) 2022-2023 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifdef __aarch64__ + +#include <cstddef> +#include "../../bfloat.hpp" + +namespace arm_gemm { + +void a64_ffinterleaved_bf16fp32_dot_8x12( + const bfloat16 *Apanel, + const bfloat16 *Bpanel, + size_t B_stride, + float *Cpanel, + int ablocks, + size_t N, + int K) { + + struct KernelArgs { + size_t K = {}; + const bfloat16 *Bpanel = {}; + size_t N = {}; + size_t B_stride = {}; + const bfloat16 *cur_B_ptr = {}; + } ka; + + ka.K = (K/2) - 1; + ka.Bpanel = Bpanel; + ka.N = N; + ka.B_stride = B_stride; + + __asm__ __volatile__( + "1:" // Height loop + "ldr x20, [%x[args_ptr], %[offsetof_Bpanel]]\n" + "ldr x25, [%x[args_ptr], %[offsetof_N]]\n" + "str x20, [%x[args_ptr], %[offsetof_cur_B_ptr]]\n" + "mov x24, %x[Apanel]\n" + "2:" // Width loop + "ldr x23, [%x[args_ptr], %[offsetof_cur_B_ptr]]\n" + "ldr x20, [%x[args_ptr], %[offsetof_B_stride]]\n" + "add x22, x23, x20, LSL #1\n" + "add x21, x22, x20, LSL #1\n" + "add x20, x21, x20, LSL #1\n" + "str x20, [%x[args_ptr], %[offsetof_cur_B_ptr]]\n" + "cmp x25, #0x8\n" + "mov %x[Apanel], x24\n" + "bgt 3f\n" + "cmp x25, #0x4\n" + "mov x21, x23\n" + "bgt 3f\n" + "mov x22, x23\n" + "3:" // B setup done + "ldr q0, [%x[Apanel], #0x0]\n" + "ldr q1, [%x[Apanel], #0x10]\n" + "movi v8.16b, #0x0\n" + "ldr q4, [x23, #0x0]\n" + "ldr q5, [x22, #0x0]\n" + "movi v9.16b, #0x0\n" + "ldr q6, [x21, #0x0]\n" + "ldr x20, [%x[args_ptr], %[offsetof_K]]\n" + "cmp x20, #0x2\n" + "movi v10.16b, #0x0\n" + "movi v11.16b, #0x0\n" + "movi v12.16b, #0x0\n" + "movi v13.16b, #0x0\n" + "movi v14.16b, #0x0\n" + "movi v15.16b, #0x0\n" + "movi v16.16b, #0x0\n" + "movi v17.16b, #0x0\n" + "movi v18.16b, #0x0\n" + "movi v19.16b, #0x0\n" + "movi v20.16b, #0x0\n" + "movi v21.16b, #0x0\n" + "movi v22.16b, #0x0\n" + "movi v23.16b, #0x0\n" + "movi v24.16b, #0x0\n" + "movi v25.16b, #0x0\n" + "movi v26.16b, #0x0\n" + "movi v27.16b, #0x0\n" + "movi v28.16b, #0x0\n" + "movi v29.16b, #0x0\n" + "movi v30.16b, #0x0\n" + "movi v31.16b, #0x0\n" + "blt 5f\n" + "4:" // main loop head + "ldr q3, [%x[Apanel], #0x20]\n" + "ldr q7, [%x[Apanel], #0x30]\n" + ".inst 0x4f40f088 // bfdot v8.4s, v4.8h, v0.h[0]\n" + ".inst 0x4f60f08b // bfdot v11.4s, v4.8h, v0.h[1]\n" + ".inst 0x4f40f88e // bfdot v14.4s, v4.8h, v0.h[2]\n" + "sub x20, x20, #0x2\n" + ".inst 0x4f60f891 // bfdot v17.4s, v4.8h, v0.h[3]\n" + ".inst 0x4f41f094 // bfdot v20.4s, v4.8h, v1.h[0]\n" + "cmp x20, #0x2\n" + ".inst 0x4f61f097 // bfdot v23.4s, v4.8h, v1.h[1]\n" + ".inst 0x4f41f89a // bfdot v26.4s, v4.8h, v1.h[2]\n" + "add %x[Apanel], %x[Apanel], #0x40\n" + ".inst 0x4f61f89d // bfdot v29.4s, v4.8h, v1.h[3]\n" + "ldr q4, [x23, #0x10]\n" + ".inst 0x4f40f0a9 // bfdot v9.4s, v5.8h, v0.h[0]\n" + ".inst 0x4f60f0ac // bfdot v12.4s, v5.8h, v0.h[1]\n" + ".inst 0x4f40f8af // bfdot v15.4s, v5.8h, v0.h[2]\n" + "add x23, x23, #0x20\n" + ".inst 0x4f60f8b2 // bfdot v18.4s, v5.8h, v0.h[3]\n" + ".inst 0x4f41f0b5 // bfdot v21.4s, v5.8h, v1.h[0]\n" + ".inst 0x4f61f0b8 // bfdot v24.4s, v5.8h, v1.h[1]\n" + ".inst 0x4f41f8bb // bfdot v27.4s, v5.8h, v1.h[2]\n" + ".inst 0x4f61f8be // bfdot v30.4s, v5.8h, v1.h[3]\n" + "ldr q5, [x22, #0x10]\n" + ".inst 0x4f40f0ca // bfdot v10.4s, v6.8h, v0.h[0]\n" + ".inst 0x4f60f0cd // bfdot v13.4s, v6.8h, v0.h[1]\n" + ".inst 0x4f40f8d0 // bfdot v16.4s, v6.8h, v0.h[2]\n" + "add x22, x22, #0x20\n" + ".inst 0x4f60f8d3 // bfdot v19.4s, v6.8h, v0.h[3]\n" + "ldr q0, [%x[Apanel], #0x0]\n" + ".inst 0x4f41f0d6 // bfdot v22.4s, v6.8h, v1.h[0]\n" + ".inst 0x4f61f0d9 // bfdot v25.4s, v6.8h, v1.h[1]\n" + ".inst 0x4f41f8dc // bfdot v28.4s, v6.8h, v1.h[2]\n" + ".inst 0x4f61f8df // bfdot v31.4s, v6.8h, v1.h[3]\n" + "ldr q2, [x21, #0x10]\n" + "ldr q1, [%x[Apanel], #0x10]\n" + "add x21, x21, #0x20\n" + ".inst 0x4f43f088 // bfdot v8.4s, v4.8h, v3.h[0]\n" + ".inst 0x4f63f08b // bfdot v11.4s, v4.8h, v3.h[1]\n" + ".inst 0x4f43f88e // bfdot v14.4s, v4.8h, v3.h[2]\n" + ".inst 0x4f63f891 // bfdot v17.4s, v4.8h, v3.h[3]\n" + ".inst 0x4f47f094 // bfdot v20.4s, v4.8h, v7.h[0]\n" + ".inst 0x4f67f097 // bfdot v23.4s, v4.8h, v7.h[1]\n" + ".inst 0x4f47f89a // bfdot v26.4s, v4.8h, v7.h[2]\n" + ".inst 0x4f67f89d // bfdot v29.4s, v4.8h, v7.h[3]\n" + "ldr q4, [x23, #0x0]\n" + ".inst 0x4f43f0a9 // bfdot v9.4s, v5.8h, v3.h[0]\n" + ".inst 0x4f63f0ac // bfdot v12.4s, v5.8h, v3.h[1]\n" + ".inst 0x4f43f8af // bfdot v15.4s, v5.8h, v3.h[2]\n" + ".inst 0x4f63f8b2 // bfdot v18.4s, v5.8h, v3.h[3]\n" + ".inst 0x4f47f0b5 // bfdot v21.4s, v5.8h, v7.h[0]\n" + ".inst 0x4f67f0b8 // bfdot v24.4s, v5.8h, v7.h[1]\n" + ".inst 0x4f47f8bb // bfdot v27.4s, v5.8h, v7.h[2]\n" + ".inst 0x4f67f8be // bfdot v30.4s, v5.8h, v7.h[3]\n" + "ldr q5, [x22, #0x0]\n" + ".inst 0x4f43f04a // bfdot v10.4s, v2.8h, v3.h[0]\n" + ".inst 0x4f63f04d // bfdot v13.4s, v2.8h, v3.h[1]\n" + ".inst 0x4f43f850 // bfdot v16.4s, v2.8h, v3.h[2]\n" + ".inst 0x4f63f853 // bfdot v19.4s, v2.8h, v3.h[3]\n" + ".inst 0x4f47f056 // bfdot v22.4s, v2.8h, v7.h[0]\n" + ".inst 0x4f67f059 // bfdot v25.4s, v2.8h, v7.h[1]\n" + ".inst 0x4f47f85c // bfdot v28.4s, v2.8h, v7.h[2]\n" + ".inst 0x4f67f85f // bfdot v31.4s, v2.8h, v7.h[3]\n" + "ldr q6, [x21, #0x0]\n" + "bge 4b\n" + "5:" // main loop skip + ".inst 0x4f40f088 // bfdot v8.4s, v4.8h, v0.h[0]\n" + ".inst 0x4f60f08b // bfdot v11.4s, v4.8h, v0.h[1]\n" + "add %x[Apanel], %x[Apanel], #0x20\n" + ".inst 0x4f40f88e // bfdot v14.4s, v4.8h, v0.h[2]\n" + ".inst 0x4f60f891 // bfdot v17.4s, v4.8h, v0.h[3]\n" + "add x23, x23, #0x10\n" + ".inst 0x4f41f094 // bfdot v20.4s, v4.8h, v1.h[0]\n" + ".inst 0x4f61f097 // bfdot v23.4s, v4.8h, v1.h[1]\n" + "add x22, x22, #0x10\n" + ".inst 0x4f41f89a // bfdot v26.4s, v4.8h, v1.h[2]\n" + ".inst 0x4f61f89d // bfdot v29.4s, v4.8h, v1.h[3]\n" + "add x21, x21, #0x10\n" + ".inst 0x4f40f0a9 // bfdot v9.4s, v5.8h, v0.h[0]\n" + ".inst 0x4f60f0ac // bfdot v12.4s, v5.8h, v0.h[1]\n" + ".inst 0x4f40f8af // bfdot v15.4s, v5.8h, v0.h[2]\n" + ".inst 0x4f60f8b2 // bfdot v18.4s, v5.8h, v0.h[3]\n" + ".inst 0x4f41f0b5 // bfdot v21.4s, v5.8h, v1.h[0]\n" + ".inst 0x4f61f0b8 // bfdot v24.4s, v5.8h, v1.h[1]\n" + ".inst 0x4f41f8bb // bfdot v27.4s, v5.8h, v1.h[2]\n" + ".inst 0x4f61f8be // bfdot v30.4s, v5.8h, v1.h[3]\n" + ".inst 0x4f40f0ca // bfdot v10.4s, v6.8h, v0.h[0]\n" + ".inst 0x4f60f0cd // bfdot v13.4s, v6.8h, v0.h[1]\n" + ".inst 0x4f40f8d0 // bfdot v16.4s, v6.8h, v0.h[2]\n" + ".inst 0x4f60f8d3 // bfdot v19.4s, v6.8h, v0.h[3]\n" + ".inst 0x4f41f0d6 // bfdot v22.4s, v6.8h, v1.h[0]\n" + ".inst 0x4f61f0d9 // bfdot v25.4s, v6.8h, v1.h[1]\n" + ".inst 0x4f41f8dc // bfdot v28.4s, v6.8h, v1.h[2]\n" + ".inst 0x4f61f8df // bfdot v31.4s, v6.8h, v1.h[3]\n" + "cbz x20, 6f\n" + "ldr q4, [%x[Apanel], #0x0]\n" + "ldr q3, [%x[Apanel], #0x10]\n" + "add %x[Apanel], %x[Apanel], #0x20\n" + "ldr q2, [x23, #0x0]\n" + "ldr q1, [x22, #0x0]\n" + ".inst 0x4f44f048 // bfdot v8.4s, v2.8h, v4.h[0]\n" + "ldr q0, [x21, #0x0]\n" + ".inst 0x4f64f04b // bfdot v11.4s, v2.8h, v4.h[1]\n" + ".inst 0x4f44f84e // bfdot v14.4s, v2.8h, v4.h[2]\n" + ".inst 0x4f64f851 // bfdot v17.4s, v2.8h, v4.h[3]\n" + ".inst 0x4f43f054 // bfdot v20.4s, v2.8h, v3.h[0]\n" + ".inst 0x4f63f057 // bfdot v23.4s, v2.8h, v3.h[1]\n" + ".inst 0x4f43f85a // bfdot v26.4s, v2.8h, v3.h[2]\n" + ".inst 0x4f63f85d // bfdot v29.4s, v2.8h, v3.h[3]\n" + ".inst 0x4f44f029 // bfdot v9.4s, v1.8h, v4.h[0]\n" + ".inst 0x4f64f02c // bfdot v12.4s, v1.8h, v4.h[1]\n" + ".inst 0x4f44f82f // bfdot v15.4s, v1.8h, v4.h[2]\n" + ".inst 0x4f64f832 // bfdot v18.4s, v1.8h, v4.h[3]\n" + ".inst 0x4f43f035 // bfdot v21.4s, v1.8h, v3.h[0]\n" + ".inst 0x4f63f038 // bfdot v24.4s, v1.8h, v3.h[1]\n" + ".inst 0x4f43f83b // bfdot v27.4s, v1.8h, v3.h[2]\n" + ".inst 0x4f63f83e // bfdot v30.4s, v1.8h, v3.h[3]\n" + ".inst 0x4f44f00a // bfdot v10.4s, v0.8h, v4.h[0]\n" + ".inst 0x4f64f00d // bfdot v13.4s, v0.8h, v4.h[1]\n" + ".inst 0x4f44f810 // bfdot v16.4s, v0.8h, v4.h[2]\n" + ".inst 0x4f64f813 // bfdot v19.4s, v0.8h, v4.h[3]\n" + ".inst 0x4f43f016 // bfdot v22.4s, v0.8h, v3.h[0]\n" + ".inst 0x4f63f019 // bfdot v25.4s, v0.8h, v3.h[1]\n" + ".inst 0x4f43f81c // bfdot v28.4s, v0.8h, v3.h[2]\n" + ".inst 0x4f63f81f // bfdot v31.4s, v0.8h, v3.h[3]\n" + "6:" // multiply loop done + "subs x25, x25, #0xc\n" + "str q8, [%x[Cpanel], #0x0]\n" + "str q9, [%x[Cpanel], #0x10]\n" + "str q10, [%x[Cpanel], #0x20]\n" + "str q11, [%x[Cpanel], #0x30]\n" + "str q12, [%x[Cpanel], #0x40]\n" + "str q13, [%x[Cpanel], #0x50]\n" + "str q14, [%x[Cpanel], #0x60]\n" + "str q15, [%x[Cpanel], #0x70]\n" + "str q16, [%x[Cpanel], #0x80]\n" + "str q17, [%x[Cpanel], #0x90]\n" + "str q18, [%x[Cpanel], #0xa0]\n" + "str q19, [%x[Cpanel], #0xb0]\n" + "str q20, [%x[Cpanel], #0xc0]\n" + "str q21, [%x[Cpanel], #0xd0]\n" + "str q22, [%x[Cpanel], #0xe0]\n" + "str q23, [%x[Cpanel], #0xf0]\n" + "str q24, [%x[Cpanel], #0x100]\n" + "str q25, [%x[Cpanel], #0x110]\n" + "str q26, [%x[Cpanel], #0x120]\n" + "str q27, [%x[Cpanel], #0x130]\n" + "str q28, [%x[Cpanel], #0x140]\n" + "str q29, [%x[Cpanel], #0x150]\n" + "str q30, [%x[Cpanel], #0x160]\n" + "str q31, [%x[Cpanel], #0x170]\n" + "add %x[Cpanel], %x[Cpanel], #0x180\n" + "bgt 2b\n" + "subs %x[ablocks], %x[ablocks], #0x1\n" + "bne 1b\n" + : [Apanel] "+&r" (Apanel), [Cpanel] "+&r" (Cpanel), [ablocks] "+&r" (ablocks) + : [args_ptr] "r" (&ka), [offsetof_B_stride] "I" (offsetof(KernelArgs, B_stride)), [offsetof_Bpanel] "I" (offsetof(KernelArgs, Bpanel)), [offsetof_K] "I" (offsetof(KernelArgs, K)), [offsetof_N] "I" (offsetof(KernelArgs, N)), [offsetof_cur_B_ptr] "I" (offsetof(KernelArgs, cur_B_ptr)) + : "cc", "memory", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8", "v9", "v10", "v11", "v12", "v13", "v14", "v15", "v16", "v17", "v18", "v19", "v20", "v21", "v22", "v23", "v24", "v25", "v26", "v27", "v28", "v29", "v30", "v31", "x20", "x21", "x22", "x23", "x24", "x25" + ); +} + +} // namespace arm_gemm +#endif // __aarch64__ |