| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 6x8-minmax-neon-dup-ld64.c | 102 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64() local
108 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64()
114 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 5x8-relu-wasmsimd-splat.c | 95 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat() local
104 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
109 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemm_relu_ukernel_5x8__wasmsimd_splat()
|
| D | 6x8-minmax-neonfma-dup-ld64.c | 102 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64() local
108 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64()
114 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 5x8-wasmsimd-splat.c | 95 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat() local
104 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
109 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemm_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-minmax-wasmsimd-arm-splat.c | 97 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat() local
106 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
111 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 5x8-minmax-wasmsimd-x86-splat.c | 95 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat() local
104 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
109 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 6x8-minmax-neonfma-dup-ld128.c | 103 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
109 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
115 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8-minmax-wasmsimd-arm-splat.c | 107 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat() local
117 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
123 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-neon-dup-ld128.c | 103 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128() local
109 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
115 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 6x8inc-minmax-neon-dup-ld64.c | 104 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64() local
110 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64()
116 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 6x8inc-minmax-neonfma-dup-ld64.c | 104 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64() local
110 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64()
116 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 5x8inc-minmax-wasmsimd-arm-splat.c | 99 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat() local
108 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat()
113 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 5x8inc-minmax-wasmsimd-x86-splat.c | 97 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat() local
106 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat()
111 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 6x8inc-minmax-neon-dup-ld128.c | 105 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128() local
111 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
117 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8inc-minmax-neonfma-dup-ld128.c | 105 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128() local
111 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
117 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8inc-minmax-wasmsimd-x86-splat.c | 107 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat() local
117 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
123 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
|
| D | 6x8inc-minmax-wasmsimd-arm-splat.c | 109 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat() local
119 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
125 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 5x8-wasmsimd-splat.c | 120 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat() local
129 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
134 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_igemm_ukernel_5x8__wasmsimd_splat()
|
| D | 6x8-minmax-neonfma-dup-ld64.c | 130 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64() local
136 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64()
142 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld64()
|
| D | 6x8-minmax-neon-dup-ld64.c | 130 const float32x4_t va4c0 = vdupq_lane_f32(va4, 0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64() local
136 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64()
142 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld64()
|
| D | 5x8-relu-wasmsimd-splat.c | 120 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat() local
129 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat()
134 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_igemm_relu_ukernel_5x8__wasmsimd_splat()
|
| D | 5x8-minmax-wasmsimd-x86-splat.c | 120 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat() local
129 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
134 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_x86_splat()
|
| D | 5x8-minmax-wasmsimd-arm-splat.c | 122 const v128_t va4c0 = wasm_v32x4_shuffle(va4, va4, 0, 0, 0, 0); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat() local
131 vacc4x0123 = wasm_f32x4_add(vacc4x0123, wasm_f32x4_mul(va4c0, vb0123c0)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
136 vacc4x4567 = wasm_f32x4_add(vacc4x4567, wasm_f32x4_mul(va4c0, vb4567c0)); in xnn_f32_igemm_minmax_ukernel_5x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-neon-dup-ld128.c | 131 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128() local
137 vacc4x0123 = vmlaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
143 vacc4x4567 = vmlaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8-minmax-neonfma-dup-ld128.c | 131 const float32x4_t va4c0 = vdupq_lane_f32(vget_low_f32(va4), 0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
137 vacc4x0123 = vfmaq_f32(vacc4x0123, va4c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
143 vacc4x4567 = vfmaq_f32(vacc4x4567, va4c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|