/* * Copyright (C) 2023 Linux Studio Plugins Project * (C) 2023 Vladimir Sadovnikov * * This file is part of lsp-dsp-lib * Created on: 31 окт. 2023 г. * * lsp-dsp-lib is free software: you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License as published by * the Free Software Foundation, either version 3 of the License, or * any later version. * * lsp-dsp-lib is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public License * along with lsp-dsp-lib. If not, see . */ #ifndef PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_GATE_H_ #define PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_GATE_H_ #ifndef PRIVATE_DSP_ARCH_X86_AVX512_IMPL #error "This header should not be included directly" #endif /* PRIVATE_DSP_ARCH_X86_AVX512_IMPL */ #include #include namespace lsp { namespace avx512 { #pragma pack(push, 1) typedef struct gate_knee_t { float start[16]; // +0x000 float end[16]; // +0x040 float gain_start[16]; // +0x080 float gain_end[16]; // +0x0c0 float herm[64]; // +0x100 } gate_knee_t; #pragma pack(pop) static const uint32_t gate_const[] __lsp_aligned64 = { LSP_DSP_VEC16(0x7fffffff) }; #define UNPACK_GATE_KNEE(DST, SRC) \ __ASM_EMIT("vbroadcastss 0x00(%[" SRC "]), %%zmm0") \ __ASM_EMIT("vbroadcastss 0x04(%[" SRC "]), %%zmm1") \ __ASM_EMIT("vbroadcastss 0x08(%[" SRC "]), %%zmm2") \ __ASM_EMIT("vbroadcastss 0x0c(%[" SRC "]), %%zmm3") \ __ASM_EMIT("vbroadcastss 0x10(%[" SRC "]), %%zmm4") \ __ASM_EMIT("vbroadcastss 0x14(%[" SRC "]), %%zmm5") \ __ASM_EMIT("vbroadcastss 0x18(%[" SRC "]), %%zmm6") \ __ASM_EMIT("vbroadcastss 0x1c(%[" SRC "]), %%zmm7") \ __ASM_EMIT("vmovaps %%zmm0, 0x000 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm1, 0x040 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm2, 0x080 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm3, 0x0c0 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm4, 0x100 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm5, 0x140 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm6, 0x180 + %[" DST "]") \ __ASM_EMIT("vmovaps %%zmm7, 0x1c0 + %[" DST "]") #define PROCESS_KNEE_SINGLE_X32 \ /* in: zmm0 = lx0, zmm4 = lx1 */ \ __ASM_EMIT("vmovaps 0x100 + %[knee], %%zmm1") /* zmm1 = herm[0] */ \ __ASM_EMIT("vmovaps 0x100 + %[knee], %%zmm5") \ __ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \ __ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm4, %%zmm5") \ __ASM_EMIT("vfmadd213ps 0x180 + %[knee], %%zmm0, %%zmm1") /* zmm1 = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \ __ASM_EMIT("vfmadd213ps 0x180 + %[knee], %%zmm4, %%zmm5") \ __ASM_EMIT("vfmadd213ps 0x1c0 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = ((herm[0]*lx0+herm[1])*lx0+herm[2])*lx0+herm[3] */ \ __ASM_EMIT("vfmadd213ps 0x1c0 + %[knee], %%zmm5, %%zmm4") \ EXP_CORE_X32 /* zmm0 = EV = expf(KV) */ \ __ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \ __ASM_EMIT("vmovaps 0x040 + %[mem], %%zmm5") \ __ASM_EMIT("vcmpps $2, 0x00 + %[knee], %%zmm1, %%k4") /* k4 = [x0 <= start] */ \ __ASM_EMIT("vcmpps $2, 0x00 + %[knee], %%zmm5, %%k6") \ __ASM_EMIT("vcmpps $5, 0x40 + %[knee], %%zmm1, %%k5") /* k5 = [x0 >= end] */ \ __ASM_EMIT("vcmpps $5, 0x40 + %[knee], %%zmm5, %%k7") \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? gain_start : expf(KV) */ \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm4 %{%%k6%}") \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm0 %{%%k5%}") /* zmm0 = [x0 >= end] ? gain_end : [x0 <= start] ? gain_start : expf(KV) */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm4 %{%%k7%}") \ /* out: zmm0 = g0, zmm4 = g1 */ #define PROCESS_KNEE_SINGLE_X16 \ /* in: ymm0 = lx0 */ \ __ASM_EMIT("vmovaps 0x100 + %[knee], %%zmm1") /* zmm1 = herm[0] */ \ __ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%zmm0, %%zmm1") /* zmm1 = herm[0]*lx0+herm[1] */ \ __ASM_EMIT("vfmadd213ps 0x180 + %[knee], %%zmm0, %%zmm1") /* zmm1 = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \ __ASM_EMIT("vfmadd213ps 0x1c0 + %[knee], %%zmm1, %%zmm0") /* zmm0 = KV = ((herm[0]*lx0+herm[1])*lx0+herm[2])*lx0+herm[3] */ \ EXP_CORE_X16 /* zmm0 = EV = expf(KV) */ \ __ASM_EMIT("vmovaps 0x000 + %[mem], %%zmm1") /* zmm1 = x0 */ \ __ASM_EMIT("vcmpps $2, 0x00 + %[knee], %%zmm1, %%k4") /* k4 = [x0 <= start] */ \ __ASM_EMIT("vcmpps $5, 0x40 + %[knee], %%zmm1, %%k5") /* k5 = [x0 >= end] */ \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm0 %{%%k4%}") /* zmm0 = [x0 <= start] ? gain_start : expf(KV) */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm0 %{%%k5%}") /* zmm0 = [x0 >= end] ? gain_end : [x0 <= start] ? gain_start : expf(KV) */ \ /* out: ymm0 = g0 */ #define PROCESS_KNEE_SINGLE_X8 \ /* in: ymm0 = lx0 */ \ __ASM_EMIT("vmovaps 0x100 + %[knee], %%ymm1") /* ymm1 = herm[0] */ \ __ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%ymm0, %%ymm1") /* ymm1 = herm[0]*lx0+herm[1] */ \ __ASM_EMIT("vfmadd213ps 0x180 + %[knee], %%ymm0, %%ymm1") /* ymm1 = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \ __ASM_EMIT("vfmadd213ps 0x1c0 + %[knee], %%ymm1, %%ymm0") /* ymm0 = KV = ((herm[0]*lx0+herm[1])*lx0+herm[2])*lx0+herm[3] */ \ EXP_CORE_X8 /* ymm0 = EV = expf(KV) */ \ __ASM_EMIT("vmovaps 0x000 + %[mem], %%ymm1") /* ymm1 = x0 */ \ __ASM_EMIT("vcmpps $2, 0x00 + %[knee], %%ymm1, %%k4") /* k4 = [x0 <= start] */ \ __ASM_EMIT("vcmpps $5, 0x40 + %[knee], %%ymm1, %%k5") /* k5 = [x0 >= end] */ \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%ymm0 %{%%k4%}") /* ymm0 = [x0 <= start] ? gain_start : expf(KV) */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%ymm0 %{%%k5%}") /* ymm0 = [x0 >= end] ? gain_end : [x0 <= start] ? gain_start : expf(KV) */ \ /* out: ymm0 = g0 */ #define PROCESS_KNEE_SINGLE_X4 \ /* in: xmm0 = lx0 */ \ __ASM_EMIT("vmovaps 0x100 + %[knee], %%xmm1") /* xmm1 = herm[0] */ \ __ASM_EMIT("vfmadd213ps 0x140 + %[knee], %%xmm0, %%xmm1") /* xmm1 = herm[0]*lx0+herm[1] */ \ __ASM_EMIT("vfmadd213ps 0x180 + %[knee], %%xmm0, %%xmm1") /* xmm1 = (herm[0]*lx0+herm[1])*lx0+herm[2] */ \ __ASM_EMIT("vfmadd213ps 0x1c0 + %[knee], %%xmm1, %%xmm0") /* xmm0 = KV = ((herm[0]*lx0+herm[1])*lx0+herm[2])*lx0+herm[3] */ \ EXP_CORE_X4 /* xmm0 = EV = expf(KV) */ \ __ASM_EMIT("vmovaps 0x000 + %[mem], %%xmm1") /* xmm1 = x0 */ \ __ASM_EMIT("vcmpps $2, 0x00 + %[knee], %%xmm1, %%k4") /* k4 = [x0 <= start] */ \ __ASM_EMIT("vcmpps $5, 0x40 + %[knee], %%xmm1, %%k5") /* k5 = [x0 >= end] */ \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%xmm0 %{%%k4%}") /* xmm0 = [x0 <= start] ? gain_start : expf(KV) */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%xmm0 %{%%k5%}") /* xmm0 = [x0 >= end] ? gain_end : [x0 <= start] ? gain_start : expf(KV) */ \ /* out: xmm0 = g0 */ #define PROCESS_GATE_FULL_X32 \ /* in: zmm0 = x0, zmm4 = x1 */ \ __ASM_EMIT("vandps 0x00 + %[G2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \ __ASM_EMIT("vandps 0x00 + %[G2C], %%zmm4, %%zmm4") \ __ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm0, %%k4") /* k4 = [x0 > start] */ \ __ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm4, %%k6") \ __ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm0, %%k5 %{%%k4%}") /* k5 = [x0 > start] && [x0 < end] */ \ __ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm4, %%k7 %{%%k6%}") \ __ASM_EMIT("korw %%k7, %%k5, %%k5") \ __ASM_EMIT("kmovw %%k5, %k[mask]") \ __ASM_EMIT("test %k[mask], %k[mask]") \ __ASM_EMIT("jnz 100f") \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm0") /* zmm0 = gain_start */ \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm4") \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm0 %{%%k4%}") /* zmm0 = [x0 > start] ? gain_end : gain_start */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm4 %{%%k6%}") \ __ASM_EMIT("jmp 200f") \ __ASM_EMIT("100:") \ __ASM_EMIT("vmovaps %%zmm0, 0x00 + %[mem]") /* store fabsf(x0) */ \ __ASM_EMIT("vmovaps %%zmm4, 0x40 + %[mem]") \ LOGE_CORE_X32 /* ymm0 = lx0 = logf(fabsf(x0)) */ \ PROCESS_KNEE_SINGLE_X32 /* apply knee */ \ __ASM_EMIT("200:") \ /* out: zmm0 = g0, zmm4 = g1 */ #define PROCESS_GATE_FULL_X16 \ /* in: zmm0 = x */ \ __ASM_EMIT("vandps 0x00 + %[G2C], %%zmm0, %%zmm0") /* zmm0 = fabsf(x0) */ \ __ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%zmm0, %%k4") /* k4 = [x0 > start] */ \ __ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%zmm0, %%k5 %{%%k4%}")/* k5 = [x0 > start] && [x0 < end] */ \ __ASM_EMIT("kmovw %%k5, %k[mask]") \ __ASM_EMIT("test %k[mask], %k[mask]") \ __ASM_EMIT("jnz 100f") \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%zmm0") /* zmm0 = gain_start */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%zmm0 %{%%k4%}") /* zmm0 = [x0 > start] ? gain_end : gain_start */ \ __ASM_EMIT("jmp 200f") \ __ASM_EMIT("100:") \ __ASM_EMIT("vmovaps %%zmm0, 0x00 + %[mem]") /* store fabsf(x0) */ \ LOGE_CORE_X16 /* ymm0 = lx0 = logf(fabsf(x0)) */ \ PROCESS_KNEE_SINGLE_X16 /* apply knee */ \ __ASM_EMIT("200:") \ /* out: zmm0 = g */ #define PROCESS_GATE_FULL_X8 \ /* in: ymm0 = x0 */ \ __ASM_EMIT("vandps 0x00 + %[G2C], %%ymm0, %%ymm0") /* ymm0 = fabsf(x0) */ \ __ASM_EMIT("vcmpps $6, 0x000 + %[knee], %%ymm0, %%k4") /* k4 = [x0 > start] */ \ __ASM_EMIT("vcmpps $1, 0x040 + %[knee], %%ymm0, %%k5 %{%%k4%}")/* k5 = [x0 > start] && [x0 < end] */ \ __ASM_EMIT("kmovw %%k5, %k[mask]") \ __ASM_EMIT("test %k[mask], %k[mask]") \ __ASM_EMIT("jnz 100f") \ __ASM_EMIT("vmovaps 0x080 + %[knee], %%ymm0") /* ymm0 = gain_start */ \ __ASM_EMIT("vmovaps 0x0c0 + %[knee], %%ymm0 %{%%k4%}") /* ymm0 = [x0 > start] ? gain_end : gain_start */ \ __ASM_EMIT("jmp 200f") \ __ASM_EMIT("100:") \ __ASM_EMIT("vandps 0x00 + %[G2C], %%ymm0, %%ymm0") /* ymm0 = fabsf(x0) */ \ __ASM_EMIT("vmovaps %%ymm0, 0x00 + %[mem]") /* store fabsf(x0) */ \ LOGE_CORE_X8 /* ymm0 = lx0 = logf(fabsf(x0)) */ \ PROCESS_KNEE_SINGLE_X8 /* apply knee */ \ __ASM_EMIT("200:") \ /* out: ymm0 = g */ #define PROCESS_GATE_FULL_X4 \ /* in: xmm0 = x0 */ \ __ASM_EMIT("vandps 0x00 + %[G2C], %%xmm0, %%xmm0") /* xmm0 = fabsf(x0) */ \ __ASM_EMIT("vmovaps %%xmm0, 0x00 + %[mem]") /* store fabsf(x0) */ \ LOGE_CORE_X4 /* xmm0 = lx0 = logf(fabsf(x0)) */ \ PROCESS_KNEE_SINGLE_X4 /* apply knee */ \ /* out: xmm0 = g */ void gate_x1_gain(float *dst, const float *src, const dsp::gate_knee_t *c, size_t count) { IF_ARCH_X86( gate_knee_t knee __lsp_aligned64; float mem[32] __lsp_aligned64; float stub[16] __lsp_aligned64; size_t mask; ); ARCH_X86_ASM ( // Prepare stuff UNPACK_GATE_KNEE("knee", "gate") // 32x blocks __ASM_EMIT("sub $32, %[count]") __ASM_EMIT("jb 2f") __ASM_EMIT("1:") __ASM_EMIT("vmovups 0x00(%[src]), %%zmm0") __ASM_EMIT("vmovups 0x40(%[src]), %%zmm4") PROCESS_GATE_FULL_X32 __ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])") __ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])") __ASM_EMIT("add $0x80, %[src]") __ASM_EMIT("add $0x80, %[dst]") __ASM_EMIT("sub $32, %[count]") __ASM_EMIT("jae 1b") __ASM_EMIT("2:") // 16x blocks __ASM_EMIT("add $16, %[count]") __ASM_EMIT("jl 4f") __ASM_EMIT("vmovups 0x00(%[src]), %%zmm0") PROCESS_GATE_FULL_X16 __ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])") __ASM_EMIT("add $0x40, %[src]") __ASM_EMIT("add $0x40, %[dst]") __ASM_EMIT("sub $16, %[count]") __ASM_EMIT("4:") // 8x blocks __ASM_EMIT("add $8, %[count]") __ASM_EMIT("jl 6f") __ASM_EMIT("vmovups 0x00(%[src]), %%ymm0") PROCESS_GATE_FULL_X8 __ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])") __ASM_EMIT("add $0x20, %[src]") __ASM_EMIT("add $0x20, %[dst]") __ASM_EMIT("sub $8, %[count]") __ASM_EMIT("6:") // 4x blocks __ASM_EMIT("add $4, %[count]") __ASM_EMIT("jl 8f") __ASM_EMIT("vmovups 0x00(%[src]), %%xmm0") PROCESS_GATE_FULL_X4 __ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])") __ASM_EMIT("sub $4, %[count]") __ASM_EMIT("add $0x10, %[src]") __ASM_EMIT("add $0x10, %[dst]") __ASM_EMIT("8:") // Tail: 1x-3x block __ASM_EMIT("add $4, %[count]") __ASM_EMIT("jle 16f") __ASM_EMIT("test $1, %[count]") __ASM_EMIT("jz 10f") __ASM_EMIT("vmovss 0x00(%[src]), %%xmm0") __ASM_EMIT("add $4, %[src]") __ASM_EMIT("10:") __ASM_EMIT("test $2, %[count]") __ASM_EMIT("jz 12f") __ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0") __ASM_EMIT("12:") PROCESS_GATE_FULL_X4 __ASM_EMIT("test $1, %[count]") __ASM_EMIT("jz 14f") __ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])") __ASM_EMIT("add $4, %[dst]") __ASM_EMIT("14:") __ASM_EMIT("test $2, %[count]") __ASM_EMIT("jz 16f") __ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])") __ASM_EMIT("16:") : [dst] "+r" (dst), [src] "+r" (src), [count] "+r" (count), [mask] "=&r" (mask) : [gate] "r" (c), [knee] "o" (knee), [mem] "o" (mem), [stub] "o" (stub), [G2C] "o" (gate_const), [L2C] "o" (LOG2_CONST), [LOGC] "o" (LOGE_C), [E2C] "o" (EXP2_CONST), [LOG2E] "m" (EXP_LOG2E) : "cc", "memory", "%xmm0", "%xmm1", "%xmm2", "%xmm3", "%xmm4", "%xmm5", "%xmm6", "%xmm7", "%k4", "%k5", "%k6", "%k7" ); } void gate_x1_curve(float *dst, const float *src, const dsp::gate_knee_t *c, size_t count) { IF_ARCH_X86( gate_knee_t knee __lsp_aligned64; float mem[32] __lsp_aligned64; float stub[16] __lsp_aligned64; size_t mask; ); ARCH_X86_ASM ( // Prepare stuff UNPACK_GATE_KNEE("knee", "gate") // 32x blocks __ASM_EMIT("sub $32, %[count]") __ASM_EMIT("jb 2f") __ASM_EMIT("1:") __ASM_EMIT("vmovups 0x00(%[src]), %%zmm0") __ASM_EMIT("vmovups 0x40(%[src]), %%zmm4") PROCESS_GATE_FULL_X32 __ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0") __ASM_EMIT("vmulps 0x40(%[src]), %%zmm4, %%zmm4") __ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])") __ASM_EMIT("vmovups %%zmm4, 0x40(%[dst])") __ASM_EMIT("add $0x80, %[src]") __ASM_EMIT("add $0x80, %[dst]") __ASM_EMIT("sub $32, %[count]") __ASM_EMIT("jae 1b") __ASM_EMIT("2:") // 16x blocks __ASM_EMIT("add $16, %[count]") __ASM_EMIT("jl 4f") __ASM_EMIT("vmovups 0x00(%[src]), %%zmm0") PROCESS_GATE_FULL_X16 __ASM_EMIT("vmulps 0x00(%[src]), %%zmm0, %%zmm0") __ASM_EMIT("vmovups %%zmm0, 0x00(%[dst])") __ASM_EMIT("add $0x40, %[src]") __ASM_EMIT("add $0x40, %[dst]") __ASM_EMIT("sub $16, %[count]") __ASM_EMIT("4:") // 8x blocks __ASM_EMIT("add $8, %[count]") __ASM_EMIT("jl 6f") __ASM_EMIT("vmovups 0x00(%[src]), %%ymm0") PROCESS_GATE_FULL_X8 __ASM_EMIT("vmulps 0x00(%[src]), %%ymm0, %%ymm0") __ASM_EMIT("vmovups %%ymm0, 0x00(%[dst])") __ASM_EMIT("add $0x20, %[src]") __ASM_EMIT("add $0x20, %[dst]") __ASM_EMIT("sub $8, %[count]") __ASM_EMIT("6:") // 4x blocks __ASM_EMIT("add $4, %[count]") __ASM_EMIT("jl 8f") __ASM_EMIT("vmovups 0x00(%[src]), %%xmm0") PROCESS_GATE_FULL_X4 __ASM_EMIT("vmulps 0x00(%[src]), %%xmm0, %%xmm0") __ASM_EMIT("vmovups %%xmm0, 0x00(%[dst])") __ASM_EMIT("sub $4, %[count]") __ASM_EMIT("add $0x10, %[src]") __ASM_EMIT("add $0x10, %[dst]") __ASM_EMIT("8:") // Tail: 1x-3x block __ASM_EMIT("add $4, %[count]") __ASM_EMIT("jle 16f") __ASM_EMIT("test $1, %[count]") __ASM_EMIT("jz 10f") __ASM_EMIT("vmovss 0x00(%[src]), %%xmm0") __ASM_EMIT("add $4, %[src]") __ASM_EMIT("10:") __ASM_EMIT("test $2, %[count]") __ASM_EMIT("jz 12f") __ASM_EMIT("vmovhps 0x00(%[src]), %%xmm0, %%xmm0") __ASM_EMIT("12:") __ASM_EMIT("vmovaps %%xmm0, %%xmm4") PROCESS_GATE_FULL_X4 __ASM_EMIT("vmulps %%xmm4, %%xmm0, %%xmm0") __ASM_EMIT("test $1, %[count]") __ASM_EMIT("jz 14f") __ASM_EMIT("vmovss %%xmm0, 0x00(%[dst])") __ASM_EMIT("add $4, %[dst]") __ASM_EMIT("14:") __ASM_EMIT("test $2, %[count]") __ASM_EMIT("jz 16f") __ASM_EMIT("vmovhps %%xmm0, 0x00(%[dst])") __ASM_EMIT("16:") : [dst] "+r" (dst), [src] "+r" (src), [count] "+r" (count), [mask] "=&r" (mask) : [gate] "r" (c), [knee] "o" (knee), [mem] "o" (mem), [stub] "o" (stub), [G2C] "o" (gate_const), [L2C] "o" (LOG2_CONST), [LOGC] "o" (LOGE_C), [E2C] "o" (EXP2_CONST), [LOG2E] "m" (EXP_LOG2E) : "cc", "memory", "%xmm0", "%xmm1", "%xmm2", "%xmm3", "%xmm4", "%xmm5", "%xmm6", "%xmm7", "%k4", "%k5", "%k6", "%k7" ); } #undef PROCESS_GATE_FULL_X4 #undef PROCESS_GATE_FULL_X8 #undef PROCESS_GATE_FULL_X16 #undef PROCESS_GATE_FULL_X32 #undef PROCESS_KNEE_SINGLE_X4 #undef PROCESS_KNEE_SINGLE_X8 #undef PROCESS_KNEE_SINGLE_X16 #undef PROCESS_KNEE_SINGLE_X32 #undef UNPACK_GATE_KNEE } /* namespace avx512 */ } /* namespace lsp */ #endif /* PRIVATE_DSP_ARCH_X86_AVX512_DYNAMICS_GATE_H_ */