Warning, /RecoTracker/MkFitCore/src/PsErrLocTransp.ah is written in an unsupported language. File is not indexed.
0001 #ifdef MPLEX_INTRINSICS
0002
0003 for (int n = 0; n < N; n += MPLEX_INTRINSICS_WIDTH_BYTES / sizeof(T)) {
0004 IntrVec_t b_3 = LD(b, 3);
0005 IntrVec_t a_3 = LD(a, 3);
0006 IntrVec_t c_0 = MUL(b_3, a_3);
0007
0008 IntrVec_t b_5 = LD(b, 5);
0009 IntrVec_t a_5 = LD(a, 5);
0010 c_0 = FMA(b_5, a_5, c_0);
0011
0012 IntrVec_t b_6 = LD(b, 6);
0013 IntrVec_t a_6 = LD(a, 6);
0014 IntrVec_t c_2 = MUL(b_6, a_6);
0015
0016 IntrVec_t b_7 = LD(b, 7);
0017 IntrVec_t a_7 = LD(a, 7);
0018 c_2 = FMA(b_7, a_7, c_2);
0019 ST(c, 0, c_0);
0020
0021 IntrVec_t b_8 = LD(b, 8);
0022 IntrVec_t a_8 = LD(a, 8);
0023 c_2 = FMA(b_8, a_8, c_2);
0024
0025 IntrVec_t b_9 = LD(b, 9);
0026 IntrVec_t c_1 = MUL(b_9, a_3);
0027
0028 IntrVec_t b_10 = LD(b, 10);
0029 IntrVec_t a_10 = LD(a, 10);
0030 c_2 = FMA(b_10, a_10, c_2);
0031
0032 IntrVec_t b_11 = LD(b, 11);
0033 c_1 = FMA(b_11, a_5, c_1);
0034 IntrVec_t a_11 = LD(a, 11);
0035 c_2 = FMA(b_11, a_11, c_2);
0036
0037 IntrVec_t b_12 = LD(b, 12);
0038 IntrVec_t c_4 = MUL(b_12, a_6);
0039 ST(c, 1, c_1);
0040 IntrVec_t a_12 = LD(a, 12);
0041 IntrVec_t c_5 = MUL(b_12, a_12);
0042 ST(c, 2, c_2);
0043
0044 IntrVec_t b_13 = LD(b, 13);
0045 c_4 = FMA(b_13, a_7, c_4);
0046 IntrVec_t a_13 = LD(a, 13);
0047 c_5 = FMA(b_13, a_13, c_5);
0048
0049 IntrVec_t b_14 = LD(b, 14);
0050 c_4 = FMA(b_14, a_8, c_4);
0051 IntrVec_t a_14 = LD(a, 14);
0052 c_5 = FMA(b_14, a_14, c_5);
0053
0054 IntrVec_t b_15 = LD(b, 15);
0055 IntrVec_t c_3 = MUL(b_15, a_3);
0056
0057 IntrVec_t b_16 = LD(b, 16);
0058 c_4 = FMA(b_16, a_10, c_4);
0059 IntrVec_t a_16 = LD(a, 16);
0060 c_5 = FMA(b_16, a_16, c_5);
0061
0062 IntrVec_t b_17 = LD(b, 17);
0063 c_3 = FMA(b_17, a_5, c_3);
0064 c_4 = FMA(b_17, a_11, c_4);
0065 IntrVec_t a_17 = LD(a, 17);
0066 c_5 = FMA(b_17, a_17, c_5);
0067
0068 IntrVec_t b_18 = LD(b, 18);
0069 IntrVec_t c_7 = MUL(b_18, a_6);
0070 ST(c, 3, c_3);
0071 ST(c, 4, c_4);
0072 ST(c, 5, c_5);
0073 IntrVec_t c_8 = MUL(b_18, a_12);
0074 IntrVec_t a_18 = LD(a, 18);
0075 IntrVec_t c_9 = MUL(b_18, a_18);
0076
0077 IntrVec_t b_19 = LD(b, 19);
0078 c_7 = FMA(b_19, a_7, c_7);
0079 c_8 = FMA(b_19, a_13, c_8);
0080 IntrVec_t a_19 = LD(a, 19);
0081 c_9 = FMA(b_19, a_19, c_9);
0082
0083 IntrVec_t b_20 = LD(b, 20);
0084 c_7 = FMA(b_20, a_8, c_7);
0085 c_8 = FMA(b_20, a_14, c_8);
0086 IntrVec_t a_20 = LD(a, 20);
0087 c_9 = FMA(b_20, a_20, c_9);
0088
0089 IntrVec_t b_21 = LD(b, 21);
0090 IntrVec_t c_6 = MUL(b_21, a_3);
0091
0092 IntrVec_t b_22 = LD(b, 22);
0093 c_7 = FMA(b_22, a_10, c_7);
0094 c_8 = FMA(b_22, a_16, c_8);
0095
0096 IntrVec_t b_23 = LD(b, 23);
0097 c_6 = FMA(b_23, a_5, c_6);
0098 c_7 = FMA(b_23, a_11, c_7);
0099 c_8 = FMA(b_23, a_17, c_8);
0100
0101 IntrVec_t b_24 = LD(b, 24);
0102 IntrVec_t c_11 = MUL(b_24, a_6);
0103 ST(c, 6, c_6);
0104 ST(c, 7, c_7);
0105 ST(c, 8, c_8);
0106 ST(c, 9, c_9);
0107 IntrVec_t c_12 = MUL(b_24, a_12);
0108 IntrVec_t c_13 = MUL(b_24, a_18);
0109 IntrVec_t a_24 = LD(a, 24);
0110 IntrVec_t c_14 = MUL(b_24, a_24);
0111
0112 IntrVec_t b_25 = LD(b, 25);
0113 c_11 = FMA(b_25, a_7, c_11);
0114 c_12 = FMA(b_25, a_13, c_12);
0115 c_13 = FMA(b_25, a_19, c_13);
0116 IntrVec_t a_25 = LD(a, 25);
0117 c_14 = FMA(b_25, a_25, c_14);
0118
0119 IntrVec_t b_26 = LD(b, 26);
0120 c_11 = FMA(b_26, a_8, c_11);
0121 c_12 = FMA(b_26, a_14, c_12);
0122 c_13 = FMA(b_26, a_20, c_13);
0123 IntrVec_t a_26 = LD(a, 26);
0124 c_14 = FMA(b_26, a_26, c_14);
0125
0126 IntrVec_t b_27 = LD(b, 27);
0127 IntrVec_t c_10 = MUL(b_27, a_3);
0128
0129 IntrVec_t b_28 = LD(b, 28);
0130 c_11 = FMA(b_28, a_10, c_11);
0131 c_12 = FMA(b_28, a_16, c_12);
0132
0133 IntrVec_t b_29 = LD(b, 29);
0134 c_10 = FMA(b_29, a_5, c_10);
0135 c_11 = FMA(b_29, a_11, c_11);
0136 c_12 = FMA(b_29, a_17, c_12);
0137 ST(c, 10, c_10);
0138 ST(c, 11, c_11);
0139 ST(c, 12, c_12);
0140 ST(c, 13, c_13);
0141 ST(c, 14, c_14);
0142 }
0143
0144 #else
0145
0146 #pragma omp simd
0147 for (int n = 0; n < N; ++n) {
0148 c[0 * N + n] = b[3 * N + n] * a[3 * N + n] + b[5 * N + n] * a[5 * N + n];
0149 c[1 * N + n] = b[9 * N + n] * a[3 * N + n] + b[11 * N + n] * a[5 * N + n];
0150 c[2 * N + n] = b[6 * N + n] * a[6 * N + n] + b[7 * N + n] * a[7 * N + n] + b[8 * N + n] * a[8 * N + n] +
0151 b[10 * N + n] * a[10 * N + n] + b[11 * N + n] * a[11 * N + n];
0152 c[3 * N + n] = b[15 * N + n] * a[3 * N + n] + b[17 * N + n] * a[5 * N + n];
0153 c[4 * N + n] = b[12 * N + n] * a[6 * N + n] + b[13 * N + n] * a[7 * N + n] + b[14 * N + n] * a[8 * N + n] +
0154 b[16 * N + n] * a[10 * N + n] + b[17 * N + n] * a[11 * N + n];
0155 c[5 * N + n] = b[12 * N + n] * a[12 * N + n] + b[13 * N + n] * a[13 * N + n] + b[14 * N + n] * a[14 * N + n] +
0156 b[16 * N + n] * a[16 * N + n] + b[17 * N + n] * a[17 * N + n];
0157 c[6 * N + n] = b[21 * N + n] * a[3 * N + n] + b[23 * N + n] * a[5 * N + n];
0158 c[7 * N + n] = b[18 * N + n] * a[6 * N + n] + b[19 * N + n] * a[7 * N + n] + b[20 * N + n] * a[8 * N + n] +
0159 b[22 * N + n] * a[10 * N + n] + b[23 * N + n] * a[11 * N + n];
0160 c[8 * N + n] = b[18 * N + n] * a[12 * N + n] + b[19 * N + n] * a[13 * N + n] + b[20 * N + n] * a[14 * N + n] +
0161 b[22 * N + n] * a[16 * N + n] + b[23 * N + n] * a[17 * N + n];
0162 c[9 * N + n] = b[18 * N + n] * a[18 * N + n] + b[19 * N + n] * a[19 * N + n] + b[20 * N + n] * a[20 * N + n];
0163 c[10 * N + n] = b[27 * N + n] * a[3 * N + n] + b[29 * N + n] * a[5 * N + n];
0164 c[11 * N + n] = b[24 * N + n] * a[6 * N + n] + b[25 * N + n] * a[7 * N + n] + b[26 * N + n] * a[8 * N + n] +
0165 b[28 * N + n] * a[10 * N + n] + b[29 * N + n] * a[11 * N + n];
0166 c[12 * N + n] = b[24 * N + n] * a[12 * N + n] + b[25 * N + n] * a[13 * N + n] + b[26 * N + n] * a[14 * N + n] +
0167 b[28 * N + n] * a[16 * N + n] + b[29 * N + n] * a[17 * N + n];
0168 c[13 * N + n] = b[24 * N + n] * a[18 * N + n] + b[25 * N + n] * a[19 * N + n] + b[26 * N + n] * a[20 * N + n];
0169 c[14 * N + n] = b[24 * N + n] * a[24 * N + n] + b[25 * N + n] * a[25 * N + n] + b[26 * N + n] * a[26 * N + n];
0170 }
0171 #endif
