1 // Auto-generated file. Do not edit! 2 // Template: src/cs16-vsquareabs/scalar.c.in 3 // Generator: tools/xngen 4 // 5 // Copyright 2022 Google LLC 6 // 7 // This source code is licensed under the BSD-style license found in the 8 // LICENSE file in the root directory of this source tree. 9 10 #include <assert.h> 11 #include <stddef.h> 12 #include <stdint.h> 13 14 #include <xnnpack/math.h> 15 #include <xnnpack/vsquareabs.h> 16 17 xnn_cs16_vsquareabs_ukernel__scalar_x2(size_t batch,const int16_t * input,uint32_t * output)18void xnn_cs16_vsquareabs_ukernel__scalar_x2( 19 size_t batch, 20 const int16_t* input, 21 uint32_t* output) { 22 23 assert(batch != 0); 24 assert(input != NULL); 25 assert(output != NULL); 26 27 for (; batch >= 2; batch -= 2) { 28 const int32_t vr0 = (int32_t) input[0]; 29 const int32_t vi0 = (int32_t) input[1]; 30 const int32_t vr1 = (int32_t) input[2]; 31 const int32_t vi1 = (int32_t) input[3]; 32 input += 2 * 2; 33 34 const uint32_t vrsquare0 = (uint32_t) (vr0 * vr0); 35 const uint32_t visquare0 = (uint32_t) (vi0 * vi0); 36 const uint32_t vrsquare1 = (uint32_t) (vr1 * vr1); 37 const uint32_t visquare1 = (uint32_t) (vi1 * vi1); 38 39 const uint32_t vout0 = vrsquare0 + visquare0; 40 const uint32_t vout1 = vrsquare1 + visquare1; 41 42 output[0] = vout0; 43 output[1] = vout1; 44 output += 2; 45 } 46 47 if XNN_UNLIKELY(batch != 0) { 48 do { 49 const int32_t vr = (int32_t) input[0]; 50 const int32_t vi = (int32_t) input[1]; 51 input += 2; 52 53 const uint32_t vrsquare = (uint32_t) (vr * vr); 54 const uint32_t visquare = (uint32_t) (vi * vi); 55 56 const uint32_t vout = vrsquare + visquare; 57 58 *output++ = vout; 59 } while (--batch != 0); 60 } 61 } 62