ualgobase.cc

   1 // This file is part of the ustl library, an STL implementation.
   2 //
   3 // Copyright (C) 2005 by Mike Sharov <msharov@users.sourceforge.net>
   4 // This file is free software, distributed under the MIT License.
   5 //
   6 // ualgobase.cc
   7 //
   8 // Copy and fill optimizations are here.
   9 //
  10
  11 #ifndef NDEBUG  // Optimized code here. asserts slow it down, and are checked elsewhere.
  12 #define NDEBUG
  13 #endif
  14
  15 #include "ualgo.h"
  16
  17 namespace ustl {
  18
  19 // Generic version for implementing fill_nX_fast on non-i386 architectures.
  20 template <typename T> static inline void stosv (T*& p, size_t n, T v)
  21     { while (n--) *p++ = v; }
  22
  23 #if defined(__i386__) || defined(__x86_64__)
  24
  25 //----------------------------------------------------------------------
  26 // Copy functions
  27 //----------------------------------------------------------------------
  28
  29 #if __GNUC__ >= 3
  30 static inline void movsb_dir_up (void) __attribute__((always_inline));
  31 static inline void movsb_dir_down (void) __attribute__((always_inline));
  32 static inline void movsb (const void*& src, size_t nBytes, void*& dest) __attribute__((always_inline));
  33 static inline void movsd (const void*& src, size_t nWords, void*& dest) __attribute__((always_inline));
  34 #endif
  35
  36 static inline void movsb_dir_up (void) { asm volatile ("cld"); }
  37 static inline void movsb_dir_down (void) { asm volatile ("std"); }
  38
  39 static inline void movsb (const void*& src, size_t nBytes, void*& dest)
  40 {
  41     asm volatile ("rep;\n\tmovsb"
  42         : "=&S"(src), "=&D"(dest), "=&c"(nBytes)
  43         : "0"(src), "1"(dest), "2"(nBytes)
  44         : "memory");
  45 }
  46
  47 static inline void movsd (const void*& src, size_t nWords, void*& dest)
  48 {
  49     asm volatile ("rep;\n\tmovsl"
  50         : "=&S"(src), "=&D"(dest), "=&c"(nWords)
  51         : "0"(src), "1"(dest), "2"(nWords)
  52         : "memory");
  53 }
  54
  55 template <> inline void stosv (uint8_t*& p, size_t n, uint8_t v)
  56 { asm volatile ("rep;\n\tstosb" : "=&D"(p), "=c"(n) : "0"(p), "1"(n), "a"(v) : "memory"); }
  57 template <> inline void stosv (uint16_t*& p, size_t n, uint16_t v)
  58 { asm volatile ("rep;\n\tstosw" : "=&D"(p), "=c"(n) : "0"(p), "1"(n), "a"(v) : "memory"); }
  59 template <> inline void stosv (uint32_t*& p, size_t n, uint32_t v)
  60 { asm volatile ("rep;\n\tstosl" : "=&D"(p), "=c"(n) : "0"(p), "1"(n), "a"(v) : "memory"); }
  61
  62 #if CPU_HAS_MMX
  63 #define MMX_ALIGN       16U     // Data must be aligned on this grain
  64 #define MMX_BS          32U     // Assembly routines copy data this many bytes at a time.
  65
  66 static inline void simd_block_copy (const void* src, void* dest) __attribute__((always_inline));
  67 static inline void simd_block_store (uint8_t* dest) __attribute__((always_inline));
  68 static inline void simd_block_cleanup (void) __attribute__((always_inline));
  69
  70 static inline void simd_block_copy (const void* src, void* dest)
  71 {
  72     const char* csrc ((const char*) src);
  73     char* cdest ((char*) dest);
  74     #if CPU_HAS_SSE
  75     asm (
  76         "movaps\t%2, %%xmm0     \n\t"
  77         "movaps\t%3, %%xmm1     \n\t"
  78         "movntps\t%%xmm0, %0    \n\t"
  79         "movntps\t%%xmm1, %1"
  80         : "=m"(cdest[0]), "=m"(cdest[16])
  81         : "m"(csrc[0]), "m"(csrc[16])
  82         : "xmm0", "xmm1", "memory");
  83     #else
  84     asm (
  85         "movq   %4, %%mm0       \n\t"
  86         "movq   %5, %%mm1       \n\t"
  87         "movq   %6, %%mm2       \n\t"
  88         "movq   %7, %%mm3       \n\t"
  89         "movq   %%mm0, %0       \n\t"
  90         "movq   %%mm1, %1       \n\t"
  91         "movq   %%mm2, %2       \n\t"
  92         "movq   %%mm3, %3"
  93         : "=m"(cdest[0]), "=m"(cdest[8]), "=m"(cdest[16]), "=m"(cdest[24])
  94         : "m"(csrc[0]), "m"(csrc[8]), "m"(csrc[16]), "m"(csrc[24])
  95         : "mm0", "mm1", "mm2", "mm3", "st", "st(1)", "st(2)", "st(3)", "memory");
  96     #endif
  97 }
  98
  99 static inline void simd_block_store (uint8_t* dest)
 100 {
 101     #if CPU_HAS_SSE
 102     asm volatile (
 103         "movntq %%mm0, %0\n\t"
 104         "movntq %%mm0, %1\n\t"
 105         "movntq %%mm0, %2\n\t"
 106         "movntq %%mm0, %3"
 107         : "=m"(dest[0]), "=m"(dest[8]), "=m"(dest[16]), "=m"(dest[24])
 108         :: "memory");
 109     #else
 110     asm volatile (
 111         "movq %%mm0, %0 \n\t"
 112         "movq %%mm0, %1 \n\t"
 113         "movq %%mm0, %2 \n\t"
 114         "movq %%mm0, %3"
 115         : "=m"(dest[0]), "=m"(dest[8]), "=m"(dest[16]), "=m"(dest[24])
 116         :: "memory");
 117     #endif
 118 }
 119
 120 static inline void simd_block_cleanup (void)
 121 {
 122     #if !CPU_HAS_SSE
 123         simd::reset_mmx();
 124     #endif
 125     asm volatile ("sfence");
 126 }
 127
 128 /// The fastest optimized raw memory copy.
 129 void copy_n_fast (const void* src, size_t nBytes, void* dest) throw()
 130 {
 131     movsb_dir_up();
 132     size_t nHeadBytes = Align(uintptr_t(src), MMX_ALIGN) - uintptr_t(src);
 133     nHeadBytes = min (nHeadBytes, nBytes);
 134     movsb (src, nHeadBytes, dest);
 135     nBytes -= nHeadBytes;
 136     if (!(uintptr_t(dest) % MMX_ALIGN)) {
 137         const size_t nMiddleBlocks = nBytes / MMX_BS;
 138         for (uoff_t i = 0; i < nMiddleBlocks; ++ i) {
 139             prefetch (advance (src, 512), 0, 0);
 140             simd_block_copy (src, dest);
 141             src = advance (src, MMX_BS);
 142             dest = advance (dest, MMX_BS);
 143         }
 144         simd_block_cleanup();
 145         nBytes %= MMX_BS;
 146     }
 147     movsb (src, nBytes, dest);
 148 }
 149 #endif // CPU_HAS_MMX
 150
 151 /// The fastest optimized backwards raw memory copy.
 152 void copy_backward_fast (const void* first, const void* last, void* result) throw()
 153 {
 154     prefetch (first, 0, 0);
 155     prefetch (result, 1, 0);
 156     size_t nBytes (distance (first, last));
 157     movsb_dir_down();
 158     size_t nHeadBytes = uintptr_t(last) % 4;
 159     last = advance (last, -1);
 160     result = advance (result, -1);
 161     movsb (last, nHeadBytes, result);
 162     nBytes -= nHeadBytes;
 163     if (uintptr_t(result) % 4 == 3) {
 164         const size_t nMiddleBlocks = nBytes / 4;
 165         last = advance (last, -3);
 166         result = advance (result, -3);
 167         movsd (last, nMiddleBlocks, result);
 168         nBytes %= 4;
 169     }
 170     movsb (last, nBytes, result);
 171     movsb_dir_up();
 172 }
 173 #endif // __i386__
 174
 175 //----------------------------------------------------------------------
 176 // Fill functions
 177 //----------------------------------------------------------------------
 178
 179 #if CPU_HAS_MMX
 180 template <typename T> static inline void build_block (T) {}
 181 template <> inline void build_block (uint8_t v)
 182 {
 183     asm volatile (
 184         "movd %0, %%mm0\n\tpunpcklbw %%mm0, %%mm0\n\tpshufw $0, %%mm0, %%mm0"
 185         : : "g"(uint32_t(v)) : "mm0");
 186 }
 187 template <> inline void build_block (uint16_t v)
 188 {
 189     asm volatile (
 190         "movd %0, %%mm0\n\tpshufw $0, %%mm0, %%mm0"
 191         : : "g"(uint32_t(v)) : "mm0");
 192 }
 193 template <> inline void build_block (uint32_t v)
 194 {
 195     asm volatile (
 196         "movd %0, %%mm0\n\tpunpckldq %%mm0, %%mm0"
 197         : : "g"(uint32_t(v)) : "mm0");
 198 }
 199
 200 static inline void simd_block_fill_loop (uint8_t*& dest, size_t count)
 201 {
 202     prefetch (advance (dest, 512), 1, 0);
 203     for (uoff_t i = 0; i < count; ++ i, dest += MMX_BS)
 204         simd_block_store (dest);
 205     simd_block_cleanup();
 206     simd::reset_mmx();
 207 }
 208
 209 template <typename T>
 210 static inline void fill_n_fast (T* dest, size_t count, T v)
 211 {
 212     size_t nHead = Align(uintptr_t(dest), MMX_ALIGN) - uintptr_t(dest) / sizeof(T);
 213     nHead = min (nHead, count);
 214     stosv (dest, nHead, v);
 215     count -= nHead;
 216     build_block (v);
 217     uint8_t* bdest = (uint8_t*) dest;
 218     simd_block_fill_loop (bdest, count * sizeof(T) / MMX_BS);
 219     count %= MMX_BS;
 220     dest = (T*) bdest;
 221     stosv (dest, count, v);
 222 }
 223
 224 void fill_n8_fast (uint8_t* dest, size_t count, uint8_t v) throw()
 225     { fill_n_fast (dest, count, v); }
 226 void fill_n16_fast (uint16_t* dest, size_t count, uint16_t v) throw()
 227     { fill_n_fast (dest, count, v); }
 228 void fill_n32_fast (uint32_t* dest, size_t count, uint32_t v) throw()
 229     { fill_n_fast (dest, count, v); }
 230 #else
 231 void fill_n8_fast (uint8_t* dest, size_t count, uint8_t v) throw() { memset (dest, v, count); }
 232 void fill_n16_fast (uint16_t* dest, size_t count, uint16_t v) throw() { stosv (dest, count, v); }
 233 void fill_n32_fast (uint32_t* dest, size_t count, uint32_t v) throw() { stosv (dest, count, v); }
 234 #endif // CPU_HAS_MMX
 235
 236 /// Exchanges ranges [first, middle) and [middle, last)
 237 void rotate_fast (void* first, void* middle, void* last) throw()
 238 {
 239 #ifdef HAVE_ALLOCA_H
 240     const size_t half1 (distance (first, middle)), half2 (distance (middle, last));
 241     const size_t hmin (min (half1, half2));
 242     if (!hmin)
 243         return;
 244     void* buf = alloca (hmin);
 245     if (buf) {
 246         if (half2 < half1) {
 247             copy_n_fast (middle, half2, buf);
 248             copy_backward_fast (first, middle, last);
 249             copy_n_fast (buf, half2, first);
 250         } else {
 251             copy_n_fast (first, half1, buf);
 252             copy_n_fast (middle, half2, first);
 253             copy_n_fast (buf, half1, advance (first, half2));
 254         }
 255     } else
 256 #else
 257     if (first == middle || middle == last)
 258         return;
 259 #endif
 260     {
 261         char* f = (char*) first;
 262         char* m = (char*) middle;
 263         char* l = (char*) last;
 264         reverse (f, m);
 265         reverse (m, l);
 266         while (f != m && m != l)
 267             iter_swap (f++, --l);
 268         reverse (f, (f == m ? l : m));
 269     }
 270 }
 271
 272 #if __GNUC__ < 4
 273 size_t popcount (uint32_t v)
 274 {
 275     const uint32_t w = v - ((v >> 1) & 0x55555555); // Algorithm from AMD optimization guide
 276     const uint32_t x = (w & 0x33333333) + ((w >> 2) & 0x33333333);
 277     return (((x + (x >> 4) & 0x0F0F0F0F) * 0x01010101) >> 24);
 278 }
 279
 280 #if HAVE_INT64_T
 281 /// \brief Returns the number of 1s in \p v in binary.
 282 size_t popcount (uint64_t v)
 283 {
 284     v -= (v >> 1) & UINT64_C(0x5555555555555555);               // Algorithm from Wikipedia
 285     v = (v & UINT64_C(0x3333333333333333)) + ((v >> 2) & UINT64_C(0x3333333333333333));
 286     v = (v + (v >> 4)) & UINT64_C(0x0F0F0F0F0F0F0F0F);
 287     return ((v * UINT64_C(0x0101010101010101)) >> 56);
 288 }
 289 #endif  // HAVE_INT64_T
 290 #endif  // !__GNUC__
 291
 292 //----------------------------------------------------------------------
 293 // Miscellaneous instantiated stuff from headers which don't have enough
 294 // to warrant creation of a separate file.cc
 295 //----------------------------------------------------------------------
 296
 297 // Used in uspecial to print printable characters
 298 const char _FmtPrtChr[2][8]={"'%c'","%d"};
 299
 300 } // namespace ustl
 301