| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 3x8inc-minmax-wasmsimd-arm-splat.c | 115 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() local
122 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
125 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8inc-minmax-wasmsimd-x86-splat.c | 113 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() local
120 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
123 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8inc-minmax-neon-dup-ld128.c | 132 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128() local
136 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
140 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8inc-minmax-neonfma-dup-ld128.c | 132 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128() local
136 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
140 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8inc-minmax-wasmsimd-arm-splat.c | 134 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat() local
142 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat()
146 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 4x8inc-minmax-wasmsimd-x86-splat.c | 132 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat() local
140 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat()
144 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 3x8-minmax-wasmsimd-arm-splat.c | 113 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
120 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
123 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 111 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
118 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
121 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-wasmsimd-splat.c | 130 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local
138 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
142 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 130 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local
134 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
138 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-relu-wasmsimd-splat.c | 130 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() local
138 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
142 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 130 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
134 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
138 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8-minmax-wasmsimd-x86-splat.c | 130 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local
138 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
142 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| D | 4x8-minmax-wasmsimd-arm-splat.c | 132 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local
140 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
144 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 5x8-wasmsimd-splat.c | 149 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat() local
158 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
163 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-relu-wasmsimd-splat.c | 149 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat() local
158 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
163 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 3x8-minmax-wasmsimd-arm-splat.c | 132 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
139 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
142 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 130 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
137 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
140 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-relu-wasmsimd-splat.c | 152 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() local
160 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
164 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-wasmsimd-splat.c | 152 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat() local
160 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
164 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 152 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128() local
156 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
160 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 152 const float32x4_t va1c3 = vdupq_lane_f32(vget_high_f32(va1), 1); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
156 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
160 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8-minmax-wasmsimd-arm-splat.c | 154 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local
162 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
166 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| D | 4x8-minmax-wasmsimd-x86-splat.c | 152 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local
160 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
164 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
| D | 5x8-wasmsimd-splat.c | 174 const v128_t va1c3 = wasm_v32x4_shuffle(va1, va1, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat() local
183 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c3, vb0123c3)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
188 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c3, vb4567c3)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
|