| /external/XNNPACK/src/f32-velu/gen/ |
| D | velu-wasmsimd-x86-rr2-p6-x8.c | 55 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x8() local
67 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x8()
88 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x8()
89 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x8()
96 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x8()
|
| D | velu-wasmsimd-x86-rr2-p6-x12.c | 58 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x12() local
74 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x12()
103 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x12()
104 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x12()
114 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x12()
|
| D | velu-wasmsimd-x86-rr2-lut16-p3-x8.c | 76 v128_t vs0123 = wasm_i32x4_add(vl0123, ven0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x8() local
86 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x8()
98 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x8()
99 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x8()
106 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x8()
|
| D | velu-neonfma-rr1-p6-x8.c | 53 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x8() local
76 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x8()
77 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x8()
84 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x8()
|
| D | velu-wasmsimd-x86-rr2-lut16-p3-x12.c | 88 v128_t vs0123 = wasm_i32x4_add(vl0123, ven0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x12() local
102 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x12()
119 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x12()
120 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x12()
130 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x12()
|
| D | velu-wasmsimd-x86-rr2-p6-x16.c | 61 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x16() local
81 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x16()
118 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x16()
119 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x16()
132 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x16()
|
| D | velu-neon-rr2-p6-x8.c | 54 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neon_rr2_p6_x8() local
80 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neon_rr2_p6_x8()
81 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neon_rr2_p6_x8()
88 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neon_rr2_p6_x8()
|
| D | velu-sse2-rr2-p6-x8.c | 55 __m128 vs0123 = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(vn0123), 23)); in xnn_f32_velu_ukernel__sse2_rr2_p6_x8() local
82 vt0123 = _mm_mul_ps(vt0123, vs0123); in xnn_f32_velu_ukernel__sse2_rr2_p6_x8()
83 vs0123 = _mm_sub_ps(vs0123, vone); in xnn_f32_velu_ukernel__sse2_rr2_p6_x8()
90 const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__sse2_rr2_p6_x8()
|
| D | velu-sse41-rr2-p6-x8.c | 55 __m128 vs0123 = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(vn0123), 23)); in xnn_f32_velu_ukernel__sse41_rr2_p6_x8() local
82 vt0123 = _mm_mul_ps(vt0123, vs0123); in xnn_f32_velu_ukernel__sse41_rr2_p6_x8()
83 vs0123 = _mm_sub_ps(vs0123, vone); in xnn_f32_velu_ukernel__sse41_rr2_p6_x8()
90 const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__sse41_rr2_p6_x8()
|
| D | velu-wasmsimd-arm-rr2-p6-x8.c | 55 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x8() local
82 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x8()
83 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x8()
90 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x8()
|
| D | velu-wasmsimd-x86-rr2-p6-x20.c | 64 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x20() local
88 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x20()
133 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x20()
134 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x20()
150 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x20()
|
| D | velu-wasmsimd-x86-rr2-lut16-p3-x16.c | 100 v128_t vs0123 = wasm_i32x4_add(vl0123, ven0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x16() local
118 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x16()
140 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x16()
141 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x16()
154 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x16()
|
| D | velu-neon-rr2-p6-x12.c | 57 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neon_rr2_p6_x12() local
92 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neon_rr2_p6_x12()
93 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neon_rr2_p6_x12()
103 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neon_rr2_p6_x12()
|
| D | velu-wasmsimd-arm-rr2-p6-x12.c | 58 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x12() local
94 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x12()
95 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x12()
105 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_p6_x12()
|
| D | velu-sse41-rr2-p6-x12.c | 58 __m128 vs0123 = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(vn0123), 23)); in xnn_f32_velu_ukernel__sse41_rr2_p6_x12() local
94 vt0123 = _mm_mul_ps(vt0123, vs0123); in xnn_f32_velu_ukernel__sse41_rr2_p6_x12()
95 vs0123 = _mm_sub_ps(vs0123, vone); in xnn_f32_velu_ukernel__sse41_rr2_p6_x12()
105 const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__sse41_rr2_p6_x12()
|
| D | velu-neonfma-rr1-p6-x12.c | 56 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x12() local
87 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x12()
88 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x12()
98 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x12()
|
| D | velu-wasmsimd-arm-rr2-lut16-p3-x8.c | 76 v128_t vs0123 = wasm_i32x4_add(vl0123, ven0123); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_lut16_p3_x8() local
92 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_lut16_p3_x8()
93 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_lut16_p3_x8()
100 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_arm_rr2_lut16_p3_x8()
|
| D | velu-neonfma-rr1-lut16-p3-x8.c | 74 float32x4_t vs0123 = vreinterpretq_f32_s32(vaddq_s32(vl0123, ven0123)); in xnn_f32_velu_ukernel__neonfma_rr1_lut16_p3_x8() local
87 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neonfma_rr1_lut16_p3_x8()
88 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neonfma_rr1_lut16_p3_x8()
95 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neonfma_rr1_lut16_p3_x8()
|
| D | velu-neonfma-rr1-p6-x16.c | 59 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x16() local
98 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x16()
99 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x16()
112 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neonfma_rr1_p6_x16()
|
| D | velu-neon-rr2-lut16-p3-x8.c | 75 float32x4_t vs0123 = vreinterpretq_f32_s32(vaddq_s32(vl0123, ven0123)); in xnn_f32_velu_ukernel__neon_rr2_lut16_p3_x8() local
91 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neon_rr2_lut16_p3_x8()
92 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neon_rr2_lut16_p3_x8()
99 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neon_rr2_lut16_p3_x8()
|
| D | velu-sse2-rr2-p6-x12.c | 58 __m128 vs0123 = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(vn0123), 23)); in xnn_f32_velu_ukernel__sse2_rr2_p6_x12() local
94 vt0123 = _mm_mul_ps(vt0123, vs0123); in xnn_f32_velu_ukernel__sse2_rr2_p6_x12()
95 vs0123 = _mm_sub_ps(vs0123, vone); in xnn_f32_velu_ukernel__sse2_rr2_p6_x12()
105 const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__sse2_rr2_p6_x12()
|
| D | velu-wasmsimd-x86-rr2-p6-x24.c | 67 v128_t vs0123 = wasm_i32x4_shl(vn0123, 23); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x24() local
95 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x24()
148 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x24()
149 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x24()
168 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_p6_x24()
|
| D | velu-wasmsimd-x86-rr2-lut16-p3-x20.c | 112 v128_t vs0123 = wasm_i32x4_add(vl0123, ven0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x20() local
134 vs0123 = wasm_v128_andnot(vs0123, vsatm0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x20()
161 vt0123 = wasm_f32x4_mul(vt0123, vs0123); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x20()
162 vs0123 = wasm_f32x4_sub(vs0123, vone); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x20()
178 const v128_t ve0123 = wasm_f32x4_mul(wasm_f32x4_add(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__wasmsimd_x86_rr2_lut16_p3_x20()
|
| D | velu-neon-rr2-p6-x16.c | 60 float32x4_t vs0123 = vreinterpretq_f32_s32(vshlq_n_s32(vreinterpretq_s32_f32(vn0123), 23)); in xnn_f32_velu_ukernel__neon_rr2_p6_x16() local
104 vt0123 = vmulq_f32(vt0123, vs0123); in xnn_f32_velu_ukernel__neon_rr2_p6_x16()
105 vs0123 = vsubq_f32(vs0123, vone); in xnn_f32_velu_ukernel__neon_rr2_p6_x16()
118 const float32x4_t ve0123 = vmulq_f32(vaddq_f32(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__neon_rr2_p6_x16()
|
| D | velu-sse41-rr2-p6-x16.c | 61 __m128 vs0123 = _mm_castsi128_ps(_mm_slli_epi32(_mm_castps_si128(vn0123), 23)); in xnn_f32_velu_ukernel__sse41_rr2_p6_x16() local
106 vt0123 = _mm_mul_ps(vt0123, vs0123); in xnn_f32_velu_ukernel__sse41_rr2_p6_x16()
107 vs0123 = _mm_sub_ps(vs0123, vone); in xnn_f32_velu_ukernel__sse41_rr2_p6_x16()
120 const __m128 ve0123 = _mm_mul_ps(_mm_add_ps(vp0123, vs0123), valpha); in xnn_f32_velu_ukernel__sse41_rr2_p6_x16()
|