| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 3x8inc-minmax-wasmsimd-arm-splat.c | 116 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() local
123 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
126 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8inc-minmax-wasmsimd-x86-splat.c | 114 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() local
121 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
124 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8inc-minmax-neon-dup-ld128.c | 133 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128() local
137 vacc2x0123 = vmlaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
141 vacc2x4567 = vmlaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8inc-minmax-neonfma-dup-ld128.c | 133 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128() local
137 vacc2x0123 = vfmaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
141 vacc2x4567 = vfmaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8inc-minmax-wasmsimd-arm-splat.c | 135 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat() local
143 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat()
147 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 4x8inc-minmax-wasmsimd-x86-splat.c | 133 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat() local
141 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat()
145 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 3x8-minmax-wasmsimd-arm-splat.c | 114 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
121 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
124 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 112 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
119 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
122 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-wasmsimd-splat.c | 131 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local
139 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
143 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 131 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local
135 vacc2x0123 = vmlaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
139 vacc2x4567 = vmlaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-relu-wasmsimd-splat.c | 131 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() local
139 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
143 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 131 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
135 vacc2x0123 = vfmaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
139 vacc2x4567 = vfmaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8-minmax-wasmsimd-x86-splat.c | 131 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local
139 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
143 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| D | 4x8-minmax-wasmsimd-arm-splat.c | 133 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local
141 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
145 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 5x8-wasmsimd-splat.c | 150 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat() local
159 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
164 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-relu-wasmsimd-splat.c | 150 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat() local
159 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
164 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 3x8-minmax-wasmsimd-arm-splat.c | 133 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
140 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
143 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 131 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
138 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
141 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-relu-wasmsimd-splat.c | 153 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() local
161 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
165 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-wasmsimd-splat.c | 153 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat() local
161 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
165 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 153 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128() local
157 vacc2x0123 = vmlaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
161 vacc2x4567 = vmlaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 153 const float32x4_t va2c3 = vdupq_lane_f32(vget_high_f32(va2), 1); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
157 vacc2x0123 = vfmaq_f32(vacc2x0123, va2c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
161 vacc2x4567 = vfmaq_f32(vacc2x4567, va2c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8-minmax-wasmsimd-arm-splat.c | 155 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local
163 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
167 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 4x8-minmax-wasmsimd-x86-splat.c | 153 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local
161 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
165 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| D | 5x8-wasmsimd-splat.c | 175 const v128_t va2c3 = wasm_v32x4_shuffle(va2, va2, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat() local
184 vacc2x0123 = wasm_f32x4_add(vacc2x0123, wasm_f32x4_mul(va2c3, vb0123c3)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
189 vacc2x4567 = wasm_f32x4_add(vacc2x4567, wasm_f32x4_mul(va2c3, vb4567c3)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
|