⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 atl_ncmmijk.c

📁 基于Blas CLapck的.用过的人知道是干啥的
💻 C
📖 第 1 页 / 共 2 页
字号:
                 const SCALAR beta, TYPE *C, const int ldc);   void (*mm_b1)(const int M, const int N, const int K, const SCALAR alpha,                 const TYPE *A, const int lda, const TYPE *B, const int ldb,                 const SCALAR beta, TYPE *C, const int ldc);   void (*mmcu) (const int M, const int N, const int K, const SCALAR alpha,                 const TYPE *A, const int lda, const TYPE *B, const int ldb,                 const SCALAR beta, TYPE *C, const int ldc);   void (*mm_fixedKcu)(const int M, const int N, const int K,                       const SCALAR alpha, const TYPE *A, const int lda,                       const TYPE *B, const int ldb, const                       SCALAR beta, TYPE *C, const int ldc);   if (TA == AtlasNoTrans)   {      if (TB == AtlasNoTrans)      {         mm_fixedKcu=Mjoin(Mjoin(Mjoin(NCmm00,Mjoin(0x0x,KB)),NN),0x0x0_aX_bX);         mmcu = Mjoin(Mjoin(Mjoin(NCmm00,0x0x0),NN),0x0x0_aX_bX);      }      else      {         mm_fixedKcu=Mjoin(Mjoin(Mjoin(NCmm00,Mjoin(0x0x,KB)),NT),0x0x0_aX_bX);         mmcu = Mjoin(Mjoin(Mjoin(NCmm00,0x0x0),NT),0x0x0_aX_bX);      }      incAk = lda * KB;      incAn = -Kb * incAk;      incAm = MB;   }   else   {      if (TB == AtlasNoTrans)      {         mm_fixedKcu=Mjoin(Mjoin(Mjoin(NCmm00,Mjoin(0x0x,KB)),TN),0x0x0_aX_bX);         mmcu = Mjoin(Mjoin(Mjoin(NCmm00,0x0x0),TN),0x0x0_aX_bX);      }      else      {         mm_fixedKcu=Mjoin(Mjoin(Mjoin(NCmm00,Mjoin(0x0x,KB)),TT),0x0x0_aX_bX);         mmcu = Mjoin(Mjoin(Mjoin(NCmm00,0x0x0),TT),0x0x0_aX_bX);      }      incAk = KB;      incAn = -Kb * incAk;      incAm = MB * lda;   }   if (TB == AtlasNoTrans)   {      incBk = KB;      incBn = ldb*NB - K + kr;      incBm = -Nb * ldb * NB;   }   else   {      incBk = KB*ldb;      incBn = NB - Kb*incBk;      incBm = -Nb*NB;   }   if (alpha == ATL_rone)   {      if (TA == AtlasNoTrans)      {         if (TB == AtlasNoTrans)         {            mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,NN),0x0x0),_a1_b1);            if (beta == ATL_rone) mm_bX = mm_b1;            else if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NN),0x0x0),_a1_b0);            else mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NN),0x0x0),_a1_bX);         }         else         {            mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,NT),0x0x0),_a1_b1);            if (beta == ATL_rone) mm_bX = mm_b1;            else if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NT),0x0x0),_a1_b0);            else mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NT),0x0x0),_a1_bX);         }      }      else      {         if (TB == AtlasNoTrans)         {            mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,TN),0x0x0),_a1_b1);            if (beta == ATL_rone) mm_bX = mm_b1;            else if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TN),0x0x0),_a1_b0);            else mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TN),0x0x0),_a1_bX);         }         else         {            mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,TT),0x0x0),_a1_b1);            if (beta == ATL_rone) mm_bX = mm_b1;            else if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TT),0x0x0),_a1_b0);            else mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TT),0x0x0),_a1_bX);         }      }   }   else  /* non-one alpha */   {      btmp = Mabs(beta);      if (btmp < ATL_rone) btmp = 1.0;/* *    If needed, call version that uses temp C to handle alpha & beta safely */      if (Kb >= ATL_MaxMMalpha || Mabs(alpha) < btmp)         return(Mjoin(PATL,NCmmIJK_c)(TA, TB, M, N, K, alpha, A, lda, B, ldb,                                      beta, C, ldc));      if (TA == AtlasNoTrans)      {         if (TB == AtlasNoTrans)         {            mm_bX = mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,NN),0x0x0),_aX_bX);            if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NN),0x0x0),_aX_b0);         }         else         {            mm_bX = mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,NT),0x0x0),_aX_bX);            if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,NT),0x0x0),_aX_b0);         }      }      else      {         if (TB == AtlasNoTrans)         {            mm_bX = mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,TN),0x0x0),_aX_bX);            if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TN),0x0x0),_aX_b0);         }         else         {            mm_bX = mm_b1 = Mjoin(Mjoin(Mjoin(NCmm0,TT),0x0x0),_aX_bX);            if (beta == ATL_rzero)               mm_bX = Mjoin(Mjoin(Mjoin(NCmm0,TT),0x0x0),_aX_b0);         }      }   }   for (i=Mb; i; i--, a += incAm, b += incBm, c += incCm)   {      for (j=Nb; j; j--, a += incAn, b += incBn, c += incCn)      {         if (Kb)         {            mm_bX(MB, NB, KB, alpha, a, lda, b, ldb, beta, c, ldc);            a += incAk;  b += incBk;            for (k=Kb-1; k; k--, a += incAk, b += incBk)               mm_b1(MB, NB, KB, alpha, a, lda, b, ldb, ATL_rone, c, ldc);            if (kr)               mmcu(MB, NB, kr, alpha, a, lda, b, ldb, ATL_rone, c, ldc);         }         else if (kr)         {            if (BetaIsZero) Mjoin(PATL,gezero)(MB, NB, c, ldc);            mmcu(MB, NB, kr, alpha, a, lda, b, ldb, beta, c, ldc);         }      }   }   if (mr)  /* M-loop remainder */   {      for (j=Nb; j; j--, a += incAn, b += incBn, c += incCn)      {         if (BetaIsZero) Mjoin(PATL,gezero)(mr, NB, c, ldc);         if (Kb)         {            mm_fixedKcu(mr, NB, KB, alpha, a, lda, b, ldb, beta,                        c, ldc);            a += incAk;  b += incBk;            for (k=Kb-1; k; k--, a += incAk, b += incBk)               mm_fixedKcu(mr, NB, KB, alpha, a, lda, b, ldb, ATL_rone,                           c, ldc);            if (kr)               mmcu(mr, NB, kr, alpha, a, lda, b, ldb, ATL_rone, c, ldc);         }         else if (kr)            mmcu(mr, NB, kr, alpha, a, lda, b, ldb, beta, c, ldc);      }   }   if (nr)      ATL_assert(Mjoin(PATL,NCmmJIK)(TA, TB, M, nr, K, alpha, A, lda,                                     B+Nb*(incBn+Kb*incBk), ldb,                                     beta, C+Nb*NB*ldc, ldc) == 0);   return(0);}

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -