/external/XNNPACK/src/f32-igemm/gen/ |
D | 4x8-wasmrelaxedsimd-fma-splat.c | 154 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 162 vacc3x0123 = __builtin_wasm_fma_f32x4(vacc3x0123, va3c3, vb0123c3); in xnn_f32_igemm_ukernel_4x8__wasmrelaxedsimd_fma_splat() 166 vacc3x4567 = __builtin_wasm_fma_f32x4(vacc3x4567, va3c3, vb4567c3); in xnn_f32_igemm_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-relu-wasmrelaxedsimd-fma-splat.c | 154 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 162 vacc3x0123 = __builtin_wasm_fma_f32x4(vacc3x0123, va3c3, vb0123c3); in xnn_f32_igemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat() 166 vacc3x4567 = __builtin_wasm_fma_f32x4(vacc3x4567, va3c3, vb4567c3); in xnn_f32_igemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-relu-wasmsimd-splat.c | 154 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() local 162 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() 166 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
|
D | 4x8-minmax-wasmrelaxedsimd-fma-splat.c | 156 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 164 vacc3x0123 = __builtin_wasm_fma_f32x4(vacc3x0123, va3c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat() 168 vacc3x4567 = __builtin_wasm_fma_f32x4(vacc3x4567, va3c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-wasmsimd-splat.c | 154 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat() local 162 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat() 166 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_igemm_ukernel_4x8__wasmsimd_splat()
|
D | 4x8-minmax-wasmsimd-arm-splat.c | 156 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local 164 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat() 168 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
D | 4x8-minmax-wasmrelaxedsimd-splat.c | 156 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_splat() local 164 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_splat() 168 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmrelaxedsimd_splat()
|
D | 4x8-minmax-neonfma-dup-ld128.c | 154 const float32x4_t va3c3 = vdupq_lane_f32(vget_high_f32(va3), 1); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128() local 158 vacc3x0123 = vfmaq_f32(vacc3x0123, va3c3, vb0123c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128() 162 vacc3x4567 = vfmaq_f32(vacc3x4567, va3c3, vb4567c3); in xnn_f32_igemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8-minmax-wasmsimd-x86-splat.c | 156 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local 164 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat() 168 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
/external/XNNPACK/src/f32-gemm/gen/ |
D | 4x8-wasmsimd-splat.c | 132 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local 140 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() 144 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
D | 4x8-wasmrelaxedsimd-fma-splat.c | 132 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 140 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_ukernel_4x8__wasmrelaxedsimd_fma_splat() 144 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-minmax-neon-dup-ld128.c | 132 const float32x4_t va3c3 = vdupq_lane_f32(vget_high_f32(va3), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local 136 vacc3x0123 = vmlaq_f32(vacc3x0123, va3c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() 140 vacc3x4567 = vmlaq_f32(vacc3x4567, va3c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
D | 4x8-relu-wasmrelaxedsimd-fma-splat.c | 132 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 140 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat() 144 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-minmax-neonfma-dup-ld128.c | 132 const float32x4_t va3c3 = vdupq_lane_f32(vget_high_f32(va3), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128() local 136 vacc3x0123 = vfmaq_f32(vacc3x0123, va3c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128() 140 vacc3x4567 = vfmaq_f32(vacc3x4567, va3c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8-minmax-wasmrelaxedsimd-fma-splat.c | 134 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat() local 142 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat() 146 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-relu-wasmsimd-splat.c | 132 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() local 140 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat() 144 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_4x8__wasmsimd_splat()
|
D | 4x8-minmax-wasmsimd-arm-splat.c | 134 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat() local 142 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat() 146 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_arm_splat()
|
D | 5x8-wasmrelaxedsimd-fma-splat.c | 151 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_5x8__wasmrelaxedsimd_fma_splat() local 160 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_ukernel_5x8__wasmrelaxedsimd_fma_splat() 165 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_ukernel_5x8__wasmrelaxedsimd_fma_splat()
|
D | 4x8-minmax-wasmsimd-x86-splat.c | 134 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat() local 142 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat() 146 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
/external/XNNPACK/src/f32-gemm/gen-inc/ |
D | 4x8inc-minmax-neon-dup-ld128.c | 134 const float32x4_t va3c3 = vdupq_lane_f32(vget_high_f32(va3), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128() local 138 vacc3x0123 = vmlaq_f32(vacc3x0123, va3c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128() 142 vacc3x4567 = vmlaq_f32(vacc3x4567, va3c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neon_dup_ld128()
|
D | 4x8inc-minmax-neonfma-dup-ld128.c | 134 const float32x4_t va3c3 = vdupq_lane_f32(vget_high_f32(va3), 1); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128() local 138 vacc3x0123 = vfmaq_f32(vacc3x0123, va3c3, vb0123c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128() 142 vacc3x4567 = vfmaq_f32(vacc3x4567, va3c3, vb4567c3); in xnn_f32_gemminc_minmax_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8inc-minmax-wasmrelaxedsimd-splat.c | 136 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmrelaxedsimd_splat() local 144 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmrelaxedsimd_splat() 148 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmrelaxedsimd_splat()
|
D | 4x8inc-minmax-relaxedwasmsimd-splat.c | 136 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__relaxedwasmsimd_splat() local 144 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__relaxedwasmsimd_splat() 148 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__relaxedwasmsimd_splat()
|
D | 4x8inc-minmax-wasmsimd-x86-splat.c | 136 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat() local 144 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat() 148 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_x86_splat()
|
D | 4x8inc-minmax-wasmsimd-arm-splat.c | 136 const v128_t va3c3 = wasm_v32x4_shuffle(va3, va3, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat() local 144 vacc3x0123 = wasm_f32x4_add(vacc3x0123, wasm_f32x4_mul(va3c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat() 148 vacc3x4567 = wasm_f32x4_add(vacc3x4567, wasm_f32x4_mul(va3c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_4x8__wasmsimd_arm_splat()
|