00001
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026
00027
00028
00029
00030
00031
00032
00033
00034
00035
00036
00037
00038
00039
00040
00041
00042
00043
00044
00045
00046
00047 #include "preprocessor.h"
00048
00049 #if !defined(FORCE_ALL_GENERICS) && \
00050 !defined(FORCE_GENERIC_VECT32_MIN) && \
00051 defined(TARGET_SPECIFIC_VECT32_MIN)
00052
00053 #if __GNUC__
00054 # define DSP32_MIN_END_KERNEL_X_FCT(x_num, data) __attribute__((__naked__)) DSP32_MIN_END_KERNEL_X_FCT__(x_num, data)
00055 #elif __ICCAVR32__
00056 # define DSP32_MIN_END_KERNEL_X_FCT(x_num, data) DSP32_MIN_END_KERNEL_X_FCT__(x_num, data)
00057 #endif
00058
00059
00060 #if __GNUC__
00061 # define ASM_INSTRUCT_COMPACKED(str) str
00062 # define ASM_INSTRUCT_EXTENDED(str) str
00063 #elif __ICCAVR32__
00064 # define ASM_INSTRUCT_COMPACKED(str) str":C"
00065 # define ASM_INSTRUCT_EXTENDED(str) str":E"
00066 #endif
00067
00068
00069
00070
00071
00072
00073
00074
00075
00076
00077
00078
00079
00080
00081 #define DSP32_MINIMUM_0(r_vect1, min)
00082
00083 #define DSP32_MINIMUM_1(r_vect1, min) \
00084 "ld.w r0, "ASTRINGZ(r_vect1)"[0x0]\n\t" \
00085 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r0\n\t"
00086
00087 #define DSP32_MINIMUM_2(r_vect1, min) \
00088 "ld.d r0, "ASTRINGZ(r_vect1)"[0x0]\n\t" \
00089 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r0\n\t" \
00090 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r1\n\t"
00091
00092 #define DSP32_MINIMUM_3(r_vect1, min) \
00093 "ld.d r0, "ASTRINGZ(r_vect1)"[0x0]\n\t" \
00094 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r0\n\t" \
00095 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r1\n\t" \
00096 "ld.w r0, "ASTRINGZ(r_vect1)"[0x8]\n\t" \
00097 "min "ASTRINGZ(min)", "ASTRINGZ(min)", r0\n\t"
00098
00099
00100
00101
00102
00103
00104
00105 #define DSP32_MIN_END_KERNEL_X_FCT__(x_num, data) \
00106 static dsp32_t TPASTE2(dsp32_vect_min_end_kernel_x, x_num)(dsp32_t *vect1, dsp32_t min) \
00107 { \
00108 __asm__ __volatile__ ( \
00109 "pushm r0-r3, lr\n\t" \
00110 TPASTE2(DSP32_MINIMUM_, x_num)(r12, r11) \
00111 "mov r12, r11\n\t" \
00112 "popm r0-r3, pc\n\t" \
00113 ); \
00114 \
00115 return 0; \
00116 }
00117
00118
00119
00120
00121
00122
00123
00124
00125
00126
00127
00128
00129
00130
00132 #if __GNUC__
00133 __attribute__((__naked__))
00134 __attribute__((__noinline__))
00135 #elif __ICCAVR32__
00136 # pragma shadow_registers=full
00137 # pragma optimize=none no_inline
00138 #endif
00139 static int dsp32_vect_min_kernel_ext(dsp32_t *vect1, int size, dsp32_t *min)
00140 {
00141 __asm__ __volatile__ (
00142 "pushm r0-r7, lr\n\t"
00143
00144 "ld.w r2, r10[0x00]\n\t"
00145
00146 "sub lr, r11, 4\n\t"
00147 ASM_INSTRUCT_COMPACKED("brlt __dsp32_min_ext_end_loop")"\n"
00148
00149 "__dsp32_min_ext_loop:\n\t"
00150
00151 "ld.d r0, r12[0x00]\n\t"
00152 "min r2, r2, r0\n\t"
00153 "min r2, r2, r1\n\t"
00154
00155 "ld.d r0, r12[0x08]\n\t"
00156 "min r2, r2, r0\n\t"
00157 "min r2, r2, r1\n\t"
00158
00159 "sub r12, -16\n\t"
00160
00161 "sub lr, 4\n\t"
00162 ASM_INSTRUCT_COMPACKED("brge __dsp32_min_ext_loop")"\n"
00163
00164 "__dsp32_min_ext_end_loop:\n\t"
00165
00166 "st.w r10[0x00], r2\n\t"
00167
00168 "sub lr, -4\n\t"
00169 "sub r12, r11, lr\n\t"
00170 "popm r0-r7, pc\n\t"
00171 );
00172
00173 return 0;
00174 }
00175
00176
00177 DSP32_MIN_END_KERNEL_X_FCT(0, "")
00178 DSP32_MIN_END_KERNEL_X_FCT(1, "")
00179 DSP32_MIN_END_KERNEL_X_FCT(2, "")
00180 DSP32_MIN_END_KERNEL_X_FCT(3, "")
00181
00182 dsp32_t dsp32_vect_min(dsp32_t *vect1, int size)
00183 {
00184 typedef dsp32_t (*min_end_kernel_opti_t)(dsp32_t *, dsp32_t);
00185 static const min_end_kernel_opti_t min_end_kernel_opti[4] = {
00186 dsp32_vect_min_end_kernel_x0,
00187 dsp32_vect_min_end_kernel_x1,
00188 dsp32_vect_min_end_kernel_x2,
00189 dsp32_vect_min_end_kernel_x3
00190 };
00191 int n;
00192 dsp32_t min = (dsp32_t) DSP_Q_MAX(DSP32_QA, DSP32_QB);
00193
00194 n = dsp32_vect_min_kernel_ext(vect1, size, &min);
00195
00196
00197 return min_end_kernel_opti[size&0x3](&vect1[n], min);
00198 }
00199
00200 #endif