Warning, /RecoTracker/MkFitCore/src/upParam_kalmanGain_x_propErr.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 {
0005 IntrVec_t a_0 = LD(a, 0);
0006 IntrVec_t b_0 = LD(b, 0);
0007 IntrVec_t c_0 = MUL(a_0, b_0);
0008
0009 IntrVec_t a_1 = LD(a, 1);
0010 IntrVec_t b_1 = LD(b, 1);
0011 c_0 = FMA(a_1, b_1, c_0);
0012
0013 IntrVec_t a_2 = LD(a, 2);
0014 IntrVec_t b_3 = LD(b, 3);
0015 c_0 = FMA(a_2, b_3, c_0);
0016
0017 IntrVec_t a_3 = LD(a, 3);
0018 IntrVec_t c_1 = MUL(a_3, b_0);
0019 IntrVec_t c_2 = MUL(a_3, b_1);
0020
0021 IntrVec_t a_4 = LD(a, 4);
0022 c_1 = FMA(a_4, b_1, c_1);
0023 ST(c, 0, c_0);
0024 IntrVec_t b_2 = LD(b, 2);
0025 c_2 = FMA(a_4, b_2, c_2);
0026
0027 IntrVec_t a_5 = LD(a, 5);
0028 c_1 = FMA(a_5, b_3, c_1);
0029 IntrVec_t b_4 = LD(b, 4);
0030 c_2 = FMA(a_5, b_4, c_2);
0031
0032 IntrVec_t a_6 = LD(a, 6);
0033 IntrVec_t c_3 = MUL(a_6, b_0);
0034 ST(c, 1, c_1);
0035 IntrVec_t c_4 = MUL(a_6, b_1);
0036 ST(c, 2, c_2);
0037 IntrVec_t c_5 = MUL(a_6, b_3);
0038
0039 IntrVec_t a_7 = LD(a, 7);
0040 c_3 = FMA(a_7, b_1, c_3);
0041 c_4 = FMA(a_7, b_2, c_4);
0042 c_5 = FMA(a_7, b_4, c_5);
0043
0044 IntrVec_t a_8 = LD(a, 8);
0045 c_3 = FMA(a_8, b_3, c_3);
0046 c_4 = FMA(a_8, b_4, c_4);
0047 IntrVec_t b_5 = LD(b, 5);
0048 c_5 = FMA(a_8, b_5, c_5);
0049
0050 IntrVec_t a_9 = LD(a, 9);
0051 IntrVec_t c_6 = MUL(a_9, b_0);
0052 ST(c, 3, c_3);
0053 ST(c, 4, c_4);
0054 ST(c, 5, c_5);
0055 IntrVec_t c_7 = MUL(a_9, b_1);
0056 IntrVec_t c_8 = MUL(a_9, b_3);
0057 IntrVec_t b_6 = LD(b, 6);
0058 IntrVec_t c_9 = MUL(a_9, b_6);
0059
0060 IntrVec_t a_10 = LD(a, 10);
0061 c_6 = FMA(a_10, b_1, c_6);
0062 c_7 = FMA(a_10, b_2, c_7);
0063 c_8 = FMA(a_10, b_4, c_8);
0064 IntrVec_t b_7 = LD(b, 7);
0065 c_9 = FMA(a_10, b_7, c_9);
0066
0067 IntrVec_t a_11 = LD(a, 11);
0068 c_6 = FMA(a_11, b_3, c_6);
0069 c_7 = FMA(a_11, b_4, c_7);
0070 c_8 = FMA(a_11, b_5, c_8);
0071 IntrVec_t b_8 = LD(b, 8);
0072 c_9 = FMA(a_11, b_8, c_9);
0073 ST(c, 6, c_6);
0074 ST(c, 7, c_7);
0075 ST(c, 8, c_8);
0076
0077 IntrVec_t a_12 = LD(a, 12);
0078 IntrVec_t c_10 = MUL(a_12, b_0);
0079 ST(c, 9, c_9);
0080 IntrVec_t c_11 = MUL(a_12, b_1);
0081 IntrVec_t c_12 = MUL(a_12, b_3);
0082 IntrVec_t c_13 = MUL(a_12, b_6);
0083 IntrVec_t b_10 = LD(b, 10);
0084 IntrVec_t c_14 = MUL(a_12, b_10);
0085
0086 IntrVec_t a_13 = LD(a, 13);
0087 c_10 = FMA(a_13, b_1, c_10);
0088 c_11 = FMA(a_13, b_2, c_11);
0089 c_12 = FMA(a_13, b_4, c_12);
0090 c_13 = FMA(a_13, b_7, c_13);
0091 IntrVec_t b_11 = LD(b, 11);
0092 c_14 = FMA(a_13, b_11, c_14);
0093
0094 IntrVec_t a_14 = LD(a, 14);
0095 c_10 = FMA(a_14, b_3, c_10);
0096 c_11 = FMA(a_14, b_4, c_11);
0097 c_12 = FMA(a_14, b_5, c_12);
0098 c_13 = FMA(a_14, b_8, c_13);
0099 IntrVec_t b_12 = LD(b, 12);
0100 c_14 = FMA(a_14, b_12, c_14);
0101 ST(c, 10, c_10);
0102 ST(c, 11, c_11);
0103 ST(c, 12, c_12);
0104 ST(c, 13, c_13);
0105 ST(c, 14, c_14);
0106
0107 IntrVec_t a_15 = LD(a, 15);
0108 IntrVec_t c_15 = MUL(a_15, b_0);
0109 IntrVec_t c_16 = MUL(a_15, b_1);
0110 IntrVec_t c_17 = MUL(a_15, b_3);
0111 IntrVec_t c_18 = MUL(a_15, b_6);
0112 IntrVec_t c_19 = MUL(a_15, b_10);
0113 IntrVec_t b_15 = LD(b, 15);
0114 IntrVec_t c_20 = MUL(a_15, b_15);
0115
0116 IntrVec_t a_16 = LD(a, 16);
0117 c_15 = FMA(a_16, b_1, c_15);
0118 c_16 = FMA(a_16, b_2, c_16);
0119 c_17 = FMA(a_16, b_4, c_17);
0120 c_18 = FMA(a_16, b_7, c_18);
0121 c_19 = FMA(a_16, b_11, c_19);
0122 IntrVec_t b_16 = LD(b, 16);
0123 c_20 = FMA(a_16, b_16, c_20);
0124
0125 IntrVec_t a_17 = LD(a, 17);
0126 c_15 = FMA(a_17, b_3, c_15);
0127 c_16 = FMA(a_17, b_4, c_16);
0128 c_17 = FMA(a_17, b_5, c_17);
0129 c_18 = FMA(a_17, b_8, c_18);
0130 c_19 = FMA(a_17, b_12, c_19);
0131 ST(c, 15, c_15);
0132 ST(c, 16, c_16);
0133 ST(c, 17, c_17);
0134 ST(c, 18, c_18);
0135 ST(c, 19, c_19);
0136 IntrVec_t b_17 = LD(b, 17);
0137 c_20 = FMA(a_17, b_17, c_20);
0138 ST(c, 20, c_20);
0139 }
0140
0141 #else
0142
0143 #pragma omp simd
0144 for (int n = 0; n < N; ++n)
0145 {
0146 c[ 0*N+n] = a[ 0*N+n]*b[ 0*N+n] + a[ 1*N+n]*b[ 1*N+n] + a[ 2*N+n]*b[ 3*N+n];
0147 c[ 1*N+n] = a[ 3*N+n]*b[ 0*N+n] + a[ 4*N+n]*b[ 1*N+n] + a[ 5*N+n]*b[ 3*N+n];
0148 c[ 2*N+n] = a[ 3*N+n]*b[ 1*N+n] + a[ 4*N+n]*b[ 2*N+n] + a[ 5*N+n]*b[ 4*N+n];
0149 c[ 3*N+n] = a[ 6*N+n]*b[ 0*N+n] + a[ 7*N+n]*b[ 1*N+n] + a[ 8*N+n]*b[ 3*N+n];
0150 c[ 4*N+n] = a[ 6*N+n]*b[ 1*N+n] + a[ 7*N+n]*b[ 2*N+n] + a[ 8*N+n]*b[ 4*N+n];
0151 c[ 5*N+n] = a[ 6*N+n]*b[ 3*N+n] + a[ 7*N+n]*b[ 4*N+n] + a[ 8*N+n]*b[ 5*N+n];
0152 c[ 6*N+n] = a[ 9*N+n]*b[ 0*N+n] + a[10*N+n]*b[ 1*N+n] + a[11*N+n]*b[ 3*N+n];
0153 c[ 7*N+n] = a[ 9*N+n]*b[ 1*N+n] + a[10*N+n]*b[ 2*N+n] + a[11*N+n]*b[ 4*N+n];
0154 c[ 8*N+n] = a[ 9*N+n]*b[ 3*N+n] + a[10*N+n]*b[ 4*N+n] + a[11*N+n]*b[ 5*N+n];
0155 c[ 9*N+n] = a[ 9*N+n]*b[ 6*N+n] + a[10*N+n]*b[ 7*N+n] + a[11*N+n]*b[ 8*N+n];
0156 c[10*N+n] = a[12*N+n]*b[ 0*N+n] + a[13*N+n]*b[ 1*N+n] + a[14*N+n]*b[ 3*N+n];
0157 c[11*N+n] = a[12*N+n]*b[ 1*N+n] + a[13*N+n]*b[ 2*N+n] + a[14*N+n]*b[ 4*N+n];
0158 c[12*N+n] = a[12*N+n]*b[ 3*N+n] + a[13*N+n]*b[ 4*N+n] + a[14*N+n]*b[ 5*N+n];
0159 c[13*N+n] = a[12*N+n]*b[ 6*N+n] + a[13*N+n]*b[ 7*N+n] + a[14*N+n]*b[ 8*N+n];
0160 c[14*N+n] = a[12*N+n]*b[10*N+n] + a[13*N+n]*b[11*N+n] + a[14*N+n]*b[12*N+n];
0161 c[15*N+n] = a[15*N+n]*b[ 0*N+n] + a[16*N+n]*b[ 1*N+n] + a[17*N+n]*b[ 3*N+n];
0162 c[16*N+n] = a[15*N+n]*b[ 1*N+n] + a[16*N+n]*b[ 2*N+n] + a[17*N+n]*b[ 4*N+n];
0163 c[17*N+n] = a[15*N+n]*b[ 3*N+n] + a[16*N+n]*b[ 4*N+n] + a[17*N+n]*b[ 5*N+n];
0164 c[18*N+n] = a[15*N+n]*b[ 6*N+n] + a[16*N+n]*b[ 7*N+n] + a[17*N+n]*b[ 8*N+n];
0165 c[19*N+n] = a[15*N+n]*b[10*N+n] + a[16*N+n]*b[11*N+n] + a[17*N+n]*b[12*N+n];
0166 c[20*N+n] = a[15*N+n]*b[15*N+n] + a[16*N+n]*b[16*N+n] + a[17*N+n]*b[17*N+n];
0167 }
0168 #endif
