| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 4x8-minmax-neonfma-dup-ld64.c | 81 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld64() local
85 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld64()
89 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld64()
|
| D | 4x8-minmax-neon-dup-ld64.c | 81 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld64() local
85 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld64()
89 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld64()
|
| D | 3x8-minmax-wasmsimd-arm-splat.c | 74 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
81 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
84 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 72 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
79 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
82 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-wasmsimd-splat.c | 82 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local
90 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
94 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neon-dup-ld128.c | 82 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local
86 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
90 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8-relu-wasmsimd-splat.c | 82 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() local
90 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
94 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 82 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
86 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
90 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x2-minmax-neonfma-lane-ld64.c | 82 const float32x2_t va1c0 = vdup_lane_f32(va1, 0); in xnn_f32_gemm_minmax_ukernel_4x2__neonfma_lane_ld64() local
86 vacc1x01 = vfma_f32(vacc1x01, va1c0, vb01c0); in xnn_f32_gemm_minmax_ukernel_4x2__neonfma_lane_ld64()
|
| D | 4x8-minmax-wasmsimd-arm-splat.c | 84 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local
92 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
96 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 4x8-minmax-neon-dup-ld64.c | 103 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld64() local
107 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld64()
111 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld64()
|
| D | 4x8-minmax-neonfma-dup-ld64.c | 103 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld64() local
107 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld64()
111 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld64()
|
| D | 3x8-minmax-wasmsimd-arm-splat.c | 93 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local
100 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
103 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 3x8-minmax-wasmsimd-x86-splat.c | 91 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local
98 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
101 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8-relu-wasmsimd-splat.c | 104 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() local
112 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
116 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
|
| D | 4x8-wasmsimd-splat.c | 104 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat() local
112 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
116 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
|
| D | 4x2-minmax-neonfma-lane-ld64.c | 104 const float32x2_t va1c0 = vdup_lane_f32(va1, 0); in xnn_f32_igemm_minmax_ukernel_4x2__neonfma_lane_ld64() local
108 vacc1x01 = vfma_f32(vacc1x01, va1c0, vb01c0); in xnn_f32_igemm_minmax_ukernel_4x2__neonfma_lane_ld64()
|
| D | 4x8-minmax-neonfma-dup-ld128.c | 104 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128() local
108 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
112 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
| D | 4x8-minmax-neon-dup-ld128.c | 104 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128() local
108 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
112 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_igemm_minmax_ukernel_4x8__neon_dup_ld128()
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 4x8inc-minmax-neon-dup-ld64.c | 83 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld64() local
87 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld64()
91 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld64()
|
| D | 3x8inc-minmax-wasmsimd-arm-splat.c | 76 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() local
83 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
86 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
| D | 4x8inc-minmax-neonfma-dup-ld64.c | 83 const float32x4_t va1c0 = vdupq_lane_f32(va1, 0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld64() local
87 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld64()
91 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld64()
|
| D | 3x8inc-minmax-wasmsimd-x86-splat.c | 74 const v128_t va1c0 = wasm_v32x4_shuffle(va1, va1, 0, 0, 0, 0); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() local
81 vacc1x0123 = wasm_f32x4_add(vacc1x0123, wasm_f32x4_mul(va1c0, vb0123c0)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
84 vacc1x4567 = wasm_f32x4_add(vacc1x4567, wasm_f32x4_mul(va1c0, vb4567c0)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
| D | 4x8inc-minmax-neon-dup-ld128.c | 84 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128() local
88 vacc1x0123 = vmlaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
92 vacc1x4567 = vmlaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
|
| D | 4x8inc-minmax-neonfma-dup-ld128.c | 84 const float32x4_t va1c0 = vdupq_lane_f32(vget_low_f32(va1), 0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128() local
88 vacc1x0123 = vfmaq_f32(vacc1x0123, va1c0, vb0123c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
92 vacc1x4567 = vfmaq_f32(vacc1x4567, va1c0, vb4567c0); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
|