| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 6x8-minmax-neon-dup-ld64.c | 123 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64() local
129 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64()
135 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 5x8-relu-wasmsimd-splat.c | 114 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat() local
123 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
128 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
|
| D | 6x8-minmax-neonfma-dup-ld64.c | 123 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64() local
129 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64()
135 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 5x8-wasmsimd-splat.c | 114 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat() local
123 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
128 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-minmax-wasmsimd-arm-splat.c | 116 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat() local
125 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
130 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 5x8-minmax-wasmsimd-x86-splat.c | 114 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat() local
123 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
128 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 6x8-minmax-neonfma-dup-ld128.c | 125 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
131 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
137 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8-minmax-wasmsimd-arm-splat.c | 129 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat() local
139 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
145 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-neon-dup-ld128.c | 125 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128() local
131 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
137 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 6x8inc-minmax-neon-dup-ld64.c | 125 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64() local
131 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64()
137 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 6x8inc-minmax-neonfma-dup-ld64.c | 125 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64() local
131 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64()
137 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 5x8inc-minmax-wasmsimd-arm-splat.c | 118 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat() local
127 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat()
132 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 5x8inc-minmax-wasmsimd-x86-splat.c | 116 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat() local
125 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat()
130 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 6x8inc-minmax-neon-dup-ld128.c | 127 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128() local
133 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
139 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8inc-minmax-neonfma-dup-ld128.c | 127 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128() local
133 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
139 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8inc-minmax-wasmsimd-x86-splat.c | 129 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat() local
139 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
145 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
|
| D | 6x8inc-minmax-wasmsimd-arm-splat.c | 131 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat() local
141 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
147 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 5x8-wasmsimd-splat.c | 139 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat() local
148 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
153 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
|
| D | 6x8-minmax-neonfma-dup-ld64.c | 151 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64() local
157 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64()
163 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 6x8-minmax-neon-dup-ld64.c | 151 const float32x4_t va4c1 = vdupq_lane_f32(va4, 1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64() local
157 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64()
163 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 5x8-relu-wasmsimd-splat.c | 139 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat() local
148 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat()
153 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-minmax-wasmsimd-x86-splat.c | 139 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat() local
148 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
153 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 5x8-minmax-wasmsimd-arm-splat.c | 141 const v128_t va4c1 = wasm_v32x4_shuffle(va4, va4, 1, 1, 1, 1); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat() local
150 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c1, vb0123c1)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
155 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c1, vb4567c1)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-neon-dup-ld128.c | 153 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128() local
159 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
165 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8-minmax-neonfma-dup-ld128.c | 153 const float32x4_t va4c1 = vdupq_lane_f32(vget_low_f32(va4), 1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
159 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c1, vb0123c1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
165 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c1, vb4567c1); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|