C Libraries: Difference between revisions

From QB64 Phoenix Edition Wiki
Jump to navigation Jump to search
(Created page with "'''QB64''' uses C++ to compile a BAS program into an executable program. The following is a list of the '''C FUNCTIONs''' that can be used. '''Note: C++ Header files should be placed in the QB64 folder and are not required after a program is compiled.''' {| align="Right" | __TOC__ |} {{WhiteStart}} '''C++ Variable Types''' '''C Name Description Size QB64 Type''' char Character o...")
 
m (Protected "C Libraries" ([Edit=Allow only autoconfirmed users] (indefinite) [Move=Allow only autoconfirmed users] (indefinite)))
(No difference)

Revision as of 22:24, 27 April 2022

QB64 uses C++ to compile a BAS program into an executable program. The following is a list of the C FUNCTIONs that can be used.

Note: C++ Header files should be placed in the QB64 folder and are not required after a program is compiled.



Template:WhiteStart C++ Variable Types 

   C Name        Description                   Size      QB64 Type
   char         Character or small integer.   1 byte      _BYTE
   short int    Short Integer(Word)           2 byte      INTEGER
   int          Integer(Dword)                4 byte      LONG
   long int     Int32, Long integer or Long   4 byte      LONG
   long long    Long long (Qword)             8 byte      _INTEGER64               
   bool         Boolean value true or false.  1 byte      _BYTE 
   float        Floating point number         4 byte      SINGLE
   double       Double precision floating.    8 byte      DOUBLE
   long double  Long double precision float  10 byte      _FLOAT
   wchar_t      Wide character(Unicode)       2 or 4   
   PVOID        void pointer(void *)           ANY        _OFFSET

Template:WhiteEnd


C Functions and Subs

  
DECLARE LIBRARY 
                           'ctime.h
FUNCTION clock () 'arithmetic type elapsed processor representing time.
FUNCTION difftime# (BYVAL time2 AS _UNSIGNED LONG, BYVAL time1 AS _UNSIGNED LONG)
                                      'seconds between time2 and time1
                           'ctype.h
FUNCTION isalnum% (BYVAL c AS INTEGER) 'is an alphabet letter(isalpha(c) or isdigit(c))
FUNCTION isalpha% (BYVAL c AS INTEGER) 'is letter (isupper(c) or islower(c))
FUNCTION isdigit% (BYVAL c AS INTEGER) 'is a decimal digit
FUNCTION isgraph% (BYVAL c AS INTEGER) 'is a printing character other than space
FUNCTION islower% (BYVAL c AS INTEGER) 'is a lower-case letter
FUNCTION isprint% (BYVAL c AS INTEGER) 'is printing character. ASCII: &H20 (" ") to &H7E (~)
FUNCTION ispunct% (BYVAL c AS INTEGER) 'is printing character other than space, letter, digit
FUNCTION isspace% (BYVAL c AS INTEGER) 'is space, formfeed, newline, return, tab, vertical tab 
FUNCTION isupper% (BYVAL c AS INTEGER) 'is upper-case letter
FUNCTION isxdigit% (BYVAL c AS INTEGER)'is a hexdecimal digit character(0 thru 9 or A thru F)
FUNCTION tolower% (BYVAL c AS INTEGER) 'return lower-case equivalent
FUNCTION toupper% (BYVAL c AS INTEGER) 'return upper-case equivalent

                           'math.h  
FUNCTION acos# (BYVAL x AS DOUBLE)  
FUNCTION asin# (BYVAL x AS DOUBLE)
FUNCTION atan# (BYVAL x AS DOUBLE)  'arc tangent of x does not designate the quadrant
FUNCTION atan2# (BYVAL y AS DOUBLE, BYVAL x AS DOUBLE) 'arc-tangent of y/x designates quadrant
FUNCTION cosh# (BYVAL x AS DOUBLE)
FUNCTION ldexp# (BYVAL base AS DOUBLE, BYVAL exponent AS INTEGER) 'base times 2 to exponent
FUNCTION pow# (BYVAL base AS DOUBLE, BYVAL exponent AS DOUBLE) 'base number raised to exponent
FUNCTION sinh# (BYVAL x AS DOUBLE)
FUNCTION tanh# (BYVAL x AS DOUBLE)

                           'stdio.h   return non-zero INTEGER on failure.
FUNCTION remove% (filename AS STRING)    'removes specified file
FUNCTION rename% (oldname AS STRING, newname AS STRING) 'renames file oldname to newname. 

                           'stdlib.h
FUNCTION atol& (str AS STRING)  'convert string to Long (errno not necessarily set)
FUNCTION atoi% (str AS STRING)  'convert string to Integer (errno not necessarily set)
FUNCTION atof# (str AS STRING)  'convert string to Double (errno not necessarily set)
FUNCTION rand& ()               'random number
SUB srand (BYVAL seed AS _UNSIGNED LONG) 'random seeded number

                           'string.h
FUNCTION memchr& (BasePtr, value AS STRING, BYVAL Bytes AS LONG) ' returns pointer to first 
 'occurance of string in a set number of bytes at the memory block pointer designated.

FUNCTION memcmp (pointer1, pointer2, BYVAL Bytes AS _UNSIGNED LONG ) 
 'compares pointer bytes. Returns 0 if match, positive if ptr1>ptr2, negative if ptr1<ptr2

SUB memcpy (DestPtr, SourcePtr, BYVAL Bytes AS _UNSIGNED LONG) 
         'copies the number of source bytes from source pointer to destination pointer

SUB memmove (DestPtr AS var_TYPE, SourcePtr AS var_TYPE, BYVAL bytesize AS LONG)
         'moves a variable pointer value to destination from a source as a type byte size.

SUB memset (pointer AS var_TYPE, BYVAL value AS _UNSIGNED LONG, BYVAL nbytes AS _UNSIGNED LONG)          
'The var_TYPE MUST match the type of the argument passed! Fills a block of memory. Sets the  
'first bytes of the block of memory at pointer to the specified value as a character.

FUNCTION strcspn% (str1 AS STRING, str2 AS STRING) 
  'length of prefix of str1 consisting of characters not in str2.

FUNCTION strcmp% (str1 AS STRING, str2 AS STRING) 'compares str1 with str2, negative value if
 'str1<str2, zero if str1=str2, positive if str1>str2

FUNCTION strncmp% (str1 AS STRING, str2 AS STRING, BYVAL Bytes AS INTEGER)'compares first byte 
 'characters of str1 and str2, negative if str1<str2, 0 if str1=str2, positive if str1>str2

SUB strncpy (dest AS STRING, source AS STRING, BYVAL Bytes AS INTEGER) 'Copies first
 'bytes of source to destination. If source ends before number of bytes, dest padded with 0's
END DECLARE


Example 1: How to use the SUB memmove to transfer TYPE data when using Libraries. 

  
TYPE a   'Note: the TYPE must be placed before the DECLARE LIBRARY if used in it!
  b AS DOUBLE
  c AS LONG
END TYPE

DECLARE LIBRARY
  SUB memmove (Dest AS a, Source AS a, BYVAL bytesize AS LONG)
END DECLARE


DIM d(10) AS a
d(0).b = 1.5
d(0).c = 99

memmove d(10), d(0), LEN(d()) 'LEN gives the total byte size of the TYPE

PRINT d(10).b
PRINT d(10).c  

1.5
99
Explanation: When a TYPE variable is moved to another variable or array index, all TYPE dot values are moved with it.


Example 2: Creating different memset functions for each variable type to be used. 

  
DECLARE LIBRARY
FUNCTION memsetB& ALIAS memset (p AS _BYTE, BYVAL c AS _UNSIGNED LONG, BYVAL n AS _UNSIGNED LONG)
FUNCTION memsetI& ALIAS memset (p AS INTEGER, BYVAL c AS _UNSIGNED LONG, BYVAL n AS _UNSIGNED LONG)
FUNCTION memsetL& ALIAS memset (p AS LONG, BYVAL c AS _UNSIGNED LONG, BYVAL n AS _UNSIGNED LONG)
END DECLARE

DIM Barray(1 TO 10) AS _BYTE
res& = memsetB(Barray(1), 65, 5)

FOR i = 1 TO 10        
  PRINT Barray(i);                       'display BYTE array decimal values
NEXT 
PRINT: PRINT

DIM Larray(1 TO 10) AS LONG
res& = memsetL(Larray(1), 65, 5)

FOR i = 1 TO 10
  PRINT Larray(i);                       'displays LONG array decimal values
NEXT i 
PRINT
FOR i = 1 TO 10
  PRINT " " + HEX$(Larray(i));           'displays each byte value &H41 = 65
NEXT i
Code example by stylin
Explanation: When 5 bytes are put into a 4 byte LONG array value, the fifth byte goes into the next array element.


(Return to Table of Contents)


Bit Casting

Header file: Cast.h 

float bitcast(int t)
{
return *(float*)&t;
}
int swap_endian(unsigned int k)
{
    return (k>>24)|((k<<8) & 0x00FF0000)|((k>>8) & 0x0000FF00)|(k<<24);
}
Courtesy of Darth Who
  
DECLARE LIBRARY "Cast"
    FUNCTION bitcast## (BYVAL t AS LONG)
    FUNCTION swap_endian (BYVAL k AS _UNSIGNED LONG)
END DECLARE
Value& = &HE7750340
PRINT HEX$(Value&)
PRINT bitcast##(Value&)
PRINT bitcast##(swap_endian(Value&))


(Return to Table of Contents)


Fast Math

Fastmath.h header file. Library to speed up program calculations. Use with DECLARE LIBRARY "Fastmath" 

using namespace std;
unsigned long xrander=123456789, yrander=362436069, zrander = 521288629;
unsigned long trander;
double compbase = 2.3025850929940456840179914546844d;
double basecomp = 1.0d;
int shift1 = 1;
int shift2 = 5;
int shift3 = 16;

float Fast_Sqrt(float val) //log2(n) approximation  //this is not used in the LN approximation 
 {
  //float chk
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = val;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  //vals.tmp = (1<<29) + (vals.tmp >> 1) - (1<<22) + 0x4C000;  // I am working on a way to improve this value
  // that represents (e/2)-64 but want e/2
  return vals.val;
 }

float Fast_InvSqrt( float number ) // originally by Silicon Graphics slightly higher accuracy variant below.
 {
  long i;
  float x2, y;
  const float threehalfs = 1.5F;
  x2 = number * 0.5F;
  y  = number;
  i  = * ( long * ) &y;                       // evil floating point bit level hacking [sic]
  i  = 0x5f375a86 - ( i >> 1 );               // what the fuck? [sic]   original silicon graphics constant: 0x5f3759df
  y  = * ( float * ) &i;
  y  = y * ( threehalfs - ( x2 * y * y ) );   // 1st iteration
//y  = y * ( threehalfs - ( x2 * y * y ) );   // 2nd iteration, this can be removed
  return y;
 }

double Fast_Pow(double a, double b) //fastpower originally developed by Martin Ankerl
 {
  int tmp = (*(1 + (int *)&a));
  int tmp2 = (int)(b * (tmp - 1072632447) + 1072632447);
  double p = 0.0;
  *(1 + (int * )&p) = tmp2;
  //p = p * a / 2.71828F ; failed attempt to auto correct the accuracy
  return tmp;
 }

double Fast_Exp(double y) //2.87921
 {
  double d; 
  //*((int*)(&d) + 0) = 0;
  *((int*)(&d) + 1) = (int)(1512775 * y + 1072632447); 
  return d; 
 } 


double Fast_XLnX(double x) //Borchardt's algorithm only accurate close to the origin
 {
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return 6.0F * x * (x - 1.0F) / (x + 4.0F * vals.val ++);
 }

double Fast_LnX(double x)
 {
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return 6.0F * x * (x - 1.0F) / (x * (x + 4.0F * vals.val ++));
 }

double Fast_Log10 (double x)
 {
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return 6.0F * x * (x - 1.0F) / (x * 2.302585092994F * (x + 4.0F * vals.val ++));
 }


double Fast_LogPi (double x)
 {
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return 6.0F * x * (x - 1.0F) / (x * 1.1447298858494F * (x + 4.0F * vals.val ++));
 }

double Fast_Sin(double x) //currently only supports range between -3 * pi and 3 * pi
 {
  double sine;
  //always wrap input angle to -PI..PI I know it is abit of a compicated algorithm but hey
  int Piintdiv = x * 0.15915494309189F;
  double Pix = x - 6.2831853071796F * Piintdiv;
  if (Pix < -3.14159265F)
   Pix += 6.28318531F;
  else
   if (Pix >  3.14159265F)
    Pix -= 6.28318531F;
  if (Pix < 0)
   sine = (0.405284735F * Pix + 1.27323954F) * Pix;
   //sine = 1.27323954 * x + .405284735 * x * x; // I do so love horner form
  else
   sine = (1.27323954 - 0.405284735 * Pix) * Pix;
  return sine;
 }


double Fast_Cos(double x)
 {
  double Cosine;
  int Piintdiv = x * 0.15915494309189F;
  double Pix = x - 6.2831853071796F * Piintdiv;
  Pix += 1.57079632;
  if (Pix < -3.14159265F)
   Pix += 6.28318531F;
  else
   if (Pix >  3.14159265F)
    Pix -= 6.28318531F;
  if (Pix < 0)
    Cosine = (1.27323954 + 0.405284735 * Pix) * Pix;
  else
    Cosine = (1.27323954 - 0.405284735 * Pix) * Pix;
  return Cosine;
 }

double Fast_Tan(double x)
 {
  double Cosine;
  double sine;
  //always wrap input angle to -PI..PI
  if (x < -3.14159265F)
   x += 6.28318531F;
  else
   if (x >  3.14159265F)
    x -= 6.28318531F;
  //compute sine
  if (x < 0)
   sine = (0.405284735F * x + 1.27323954F) * x; // I do so love horner form the alernative on the nex line is slower
   //sine = 1.27323954 * x + .405284735 * x * x; 
  else
   sine = (1.27323954 - 0.405284735 * x) * x;
  x += 1.57079632;
  if (x < 0)
    Cosine = (1.27323954 + 0.405284735 * x) * x;
  else
    Cosine = (1.27323954 - 0.405284735 * x) * x;
  return  sine / Cosine;
 }

double Fast_CoTan(double x)
 {
  double Cosine;
  double sine;
  //always wrap input angle to -PI..PI
  if (x < -3.14159265F)
   x += 6.28318531F;
  else
   if (x >  3.14159265F)
    x -= 6.28318531F;
  //compute sine
  if (x < 0)
   sine = (0.405284735F * x + 1.27323954F) * x; // I do so love horner form the alernative on the nex line is slower
   //sine = 1.27323954 * x + .405284735 * x * x; 
  else
   sine = (1.27323954 - 0.405284735 * x) * x;
  x += 1.57079632;
  if (x < 0)
    Cosine = (1.27323954 + 0.405284735 * x) * x;
  else
    Cosine = (1.27323954 - 0.405284735 * x) * x;
  return  Cosine / sine;
 }

double Fast_Sec(double x)
 {
  double Cosine;
  x += 1.57079632;
  if (x >  3.14159265)
    x -= 6.28318531;
  if (x < 0)
    Cosine = (1.27323954 + 0.405284735 * x) * x;
  else
    Cosine = (1.27323954 - 0.405284735 * x) * x;
  return 1 / Cosine;
 }

double Fast_Csc(double x)
 {
  double sine;
  //always wrap input angle to -PI..PI
  if (x < -3.14159265F)
   x += 6.28318531F;
  else
   if (x >  3.14159265F)
    x -= 6.28318531F;
  //compute sine
  if (x < 0)
   sine = (0.405284735F * x + 1.27323954F) * x;
   //sine = 1.27323954 * x + .405284735 * x * x; // I do so love horner form
  else
   sine = (1.27323954 - 0.405284735 * x) * x;
  return 1 / sine;
 }


// |error| < 0.005
float Fast_Atan2(float y, float x)
 {
  float PIBY2_FLOAT = 1.5707963F;
  float PI_FLOAT = 3.14159265F;
  if (x == 0.0f)
   {
    if (y > 0.0f) return PIBY2_FLOAT;
    if (y == 0.0f) return 0.0f;
    return 0 - PIBY2_FLOAT;
   }
  float atan;
  float z = y/x;
  if ( fabsf( z ) < 1.0f )
   {
    atan = z/(1.0f + 0.28f*z*z);
    if ( x < 0.0f )
     {
      if ( y < 0.0f ) return atan - PI_FLOAT;
      return atan + PI_FLOAT;
     }
   }
   else
   {
    atan = PIBY2_FLOAT - z/(z*z + 0.28f);
    if ( y < 0.0f ) 
     return atan - PI_FLOAT;
   }
   return atan;
 }

float Fast_Atan(float y)
 {
  float x = 1.0F;
  float PIBY2_FLOAT = 1.5707963F;
  float PI_FLOAT = 3.14159265F;
  float atan;
  float z = y/x;
  if (fabsf( z ) < 1.0f )
   {
    atan = z/(1.0f + 0.28f*z*z);
    if ( x < 0.0f )
     {
      if ( y < 0.0f ) return atan - PI_FLOAT;
      return atan + PI_FLOAT;
     }
   }
   else
   {
    atan = PIBY2_FLOAT - z/(z*z + 0.28f);
    if ( y < 0.0f ) 
     return atan - PI_FLOAT;
   }
   return atan;
 }

double Fast_ACos(double x) 
 {
  float retval = (-0.69813170079773212F * x * x - 0.87266462599716477F) * x + 1.5707963267948966F;
  return retval;
 }

double Fast_ASin(double x) 
 {
  float retval = (0.69813170079773212F * x * x + 0.87266462599716477F) * x;
  return retval;
 }

double Fast_SinH(double x)
 {
  double y = -x;
  double k;
  double d; 
  //*((int*)(&d) + 0) = 0;
  *((int*)(&d) + 1) = (int)(1512775 * y + 1072632447); 
  //*((int*)(&k) + 0) = 0;
  *((int*)(&k) + 1) = (int)(1512775 * x + 1072632447);
  return (k - d) / 2; 
 }

double Fast_CosH(double x)
 {
  double y = -x;
  double k;
  double d; 
  //*((int*)(&d) + 0) = 0;
  *((int*)(&d) + 1) = (int)(1512775 * y + 1072632447); 
  //*((int*)(&k) + 0) = 0;
  *((int*)(&k) + 1) = (int)(1512775 * x + 1072632447);
  return (k + d) / 2; 
 }

double Fast_TanH(double x) // from http://www.musicdsp.org/showone.php?id=238
 {
  //double xt;
  if (x < -3)
  {
   return -1;
  }
  else if (x > 3)
  {
   return 1;
  }
  else
  {
   double xz = x * x;
   return x * (27 + xz) / (27 + 9 * xz);
  }
  //return xt
 }

double Fast_ATanH(double x)
 {
  double mlnx = 1-x;
  double plnx = 1+x;
  union
  {
    int tmp;
    float val;
  } vals;
  union
  {
    int tmp;
    float val;
  } vals2;
  vals.val = plnx;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  vals2.val = mlnx;
  vals2.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals2.tmp >>= 1; // divide by 2 
  vals2.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return (6.0F * mlnx * (mlnx - 1.0F) / (mlnx * (mlnx + 4.0F * vals.val ++)) - 6.0F * plnx * (plnx - 1.0F) / (plnx * (plnx + 4.0F * vals2.val ++))) * 0.5f;
 }

double Fast_ACosH(double val)
 {
  double zsqrz;
  //float chk
  union
  {
    int tmp;
    float val;
  } vals;
  union
  {
    int tmp;
    float val;
  } val2s;
  vals.val = val + 1;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  val2s.val = val - 1;
  val2s.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  val2s.tmp >>= 1; // divide by 2 
  val2s.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  zsqrz = val + vals.val * val2s.val;
  return log(zsqrz);
 }

double Fast_ASinH(double x)
 {
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x * x + 1;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return log(x + vals.val);
 }

double Fast_ASecH(double x)
 {
  double zsqrz;
  //float chk
  union
  {
    int tmp;
    float val;
  } vals;
  union
  {
    int tmp;
    float val;
  } val2s;
  vals.val = 1 / x + 1;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  val2s.val = 1 / x - 1;
  val2s.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  val2s.tmp >>= 1; // divide by 2 
  val2s.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  zsqrz = 1 / x + vals.val * val2s.val;
  return log(zsqrz);
 }

double Fast_ACscH(double val)
 {
  double x = 1 / val;
  union
  {
    int tmp;
    float val;
  } vals;
  vals.val = x * x + 1;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0 
  // tmp is now an approximation to logbase2(val) 
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return log(x + vals.val);
 }

double Fast_ACotH(double valts)
 {
  double x = 1 / valts; 
  double mlnx = 1-x;
  double plnx = 1+x;
  union
  {
    int tmp;
    float val;
  } vals;
  union
  {
    int tmp;
    float val;
  } vals2;
  vals.val = plnx;
  vals.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals.tmp >>= 1; // divide by 2 
  vals.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  vals2.val = mlnx;
  vals2.tmp -= 1<<23; // Remove last bit so 1.0 gives 1.0  
  vals2.tmp >>= 1; // divide by 2 
  vals2.tmp += 1<<29; // add 64 to exponent: (e+127)/2 =(e/2)+63;
  return (6.0F * mlnx * (mlnx - 1.0F) / (mlnx * (mlnx + 4.0F * vals.val ++)) - 6.0F * plnx * (plnx - 1.0F) / (plnx * (plnx + 4.0F * vals2.val ++))) / 2;
 }

void Fast_RandInit(float x,int i,int j, int k)
 {
   float ftl;
   float ftl2;
   float ftl3;
   ftl = 10000 * sin(x*7);
   xrander = * ( long * ) &ftl;
   ftl2 = 10000 * cos(x*3);
   yrander = * ( long * ) &ftl2; 
   ftl3 = 10000 * sin(x*2);
   zrander = * ( long * ) &ftl3; 
   if ((i == 0)||(j == 0)||( k == 0 )) 
    {
     shift1 = 1;
     shift2 = 5;
     shift3 = 16;
    }
   else
    {
     shift1 = i; //8 * sin(x) + 9;
     shift3 = k; //8 * cos(x) + 9;
     shift2 = j; //(shift1 + shift3) / 2;
    }
 }

inline float Fast_Rand(void) //based off of George Marsaglia's XORSHIFT algorithms
 {     //
  xrander ^= xrander << shift1; 
  xrander ^= xrander >> shift2;
  xrander ^= xrander << shift3;
  trander = xrander;
  xrander = yrander;
  yrander = zrander;
  zrander = trander ^ xrander ^ yrander;
  return float (zrander)/4294967295;
 }

inline float Fast_Sign(float f)
{
 float r = 1.0f;
 (int&)r |= ((int&)f & 0x80000000); 
 return r;
}

double High_ATanH(double x)
 {
  return (log(1 + x) - log(1 - x)) / 2;
 }

double High_ACosH(double x)
 {
  return log(x + sqrt(x + 1) * sqrt(x - 1));
 }

double High_ASinH(double x)
 {
  return log(x + sqrt(x * x + 1));
 }

double High_ASecH(double x)
 {
  double xz = 1 / x;
  return log(sqrt(xz - 1) * sqrt(xz + 1) + xz);
 }

double High_ACscH(double x)
 {
  return log(sqrt(1 + 1 / (x * x)) + 1 / x);
 }

double High_ACotH(double x)
 {
  double xz = 1 / x;
  return (log(1 + xz) - log(1 - xz)) / 2;
 }

inline float High_Sign(float f)
{
  //this may or may not be used in QB64 in the future
  if (((int&)f & 0x7FFFFFFF)==0)
    {
     return 0.0f; 
    }
    else 
    {
     float r = 1.0f;
     (int&)r |= ((int&)f & 0x80000000);    
     return r;
    }
}
int Misc_TrailZCount(unsigned int v)
 {
  //unsigned int v;     // 32-bit word input to count zero bits on right
  unsigned int c;     // c will be the number of zero bits on the right,
                    // so if v is 1101000 (base 2), then c will be 3
 // NOTE: if 0 == v, then c = 31.
  if (v & 0x1) 
  {
   // special case for odd v (assumed to happen half of the time)
   c = 0;
  }
  else
  {
   c = 1;
   if ((v & 0xffff) == 0) 
   {  
    v >>= 16;  
    c += 16;
   }
   if ((v & 0xff) == 0) 
   {  
    v >>= 8;  
    c += 8;
   }
   if ((v & 0xf) == 0) 
   {  
    v >>= 4;
    c += 4;
   }
   if ((v & 0x3) == 0) 
   {  
    v >>= 2;
    c += 2;
   }
   c -= v & 0x1;
  }	
  return c;
 }

int Misc_Parity(int v)// is the number of 1 bits odd? true if so
 {
  //unsigned int v; // 32-bit word
  v ^= v >> 1;
  v ^= v >> 2;
  v = (v & 0x11111111U) * 0x11111111U;
  return (v >> 28) & 1;
 }

int Misc_BitSet(int v)
//u//nsigned int v; // count the number of bits set in v
 {
  unsigned int c; // c accumulates the total bits set in v
  for (c = 0; v; c++)
  {
   v &= v - 1; // clear the least significant bit set
  }
  return c;
 }

long Misc_FloatToLong(float y) //1132462080
 {
  return * ( long * ) &y;
 }

float Misc_LongToFloat(long y)
 {
  return * ( float * ) &y;
 }

long Misc_UnSetRMBit(long x)
 {
  return x & (x - 1);
 }
Library created by DarthWho


Latest Fast Math Update


Example 1: Using trailzcount to speed up the process of finding the prime factors of a number. 

  
DECLARE LIBRARY "fastmath"
    FUNCTION TrailZCount% ALIAS Misc_TrailZCount (BYVAL vals AS LONG)
END DECLARE
INPUT "enter a number: ", xt&
tabs = TrailZCount%(xt&)
PRINT "the prime factors of"; xt&; " are:";
IF tabs THEN
    FOR i = 1 TO tabs
        PRINT 2;
    NEXT i
END IF
x& = xt& / (2 ^ tabs)
c& = 3
WHILE c& <= x&
    WHILE x& MOD c& = 0
        x& = x& / c&
        PRINT c&;
    WEND
    c& = c& + 1
WEND
PRINT


Example 2: Using parity to detect single bit errors simulated using unsetrmbit. 

'parity checking code which may be used in order to detect a download error:
'will also use the bitset function to simulate a download error .bitset unsets the lowest set bit
'parity checking code which may be used in order to detect a download error:
'will also use the bitset function to simulate a download error .bitset unsets the lowest set bit
DECLARE LIBRARY "fastmath"
    FUNCTION parity& ALIAS Misc_Parity (BYVAL val AS LONG)
    FUNCTION unsetrmbit& ALIAS Misc_UnSetRMBit (BYVAL vals AS LONG)
END DECLARE
TYPE DLDATA
    par AS _BIT
    dataa AS LONG
END TYPE
DIM values(1) 'here is where the data is created
g& = 8
h& = 9
values.par(0) = parity&(g&)
values.dataa(0) = g&
values.par(1) = parity&(h&)
values.dataa(1) = h& 'transfer and error in values.dataa(1)
values.dataa(1) = unsetrmbit&(values.dataa(1)) 'checking for single bit errors in download
FOR i = 0 TO 1
    IF values.par(i) = parity&(values.dataa(i)) THEN
        PRINT "no single bit errors detected in in data packet"; i
    ELSE
        PRINT "error found in data packet"; i
    END IF
NEXT


(Return to Table of Contents)


See also:


Navigation:
Main Page with Articles and Tutorials
Keyword Reference - Alphabetical
Keyword Reference - By usage
Report a broken link
Retrieved from "https://qb64phoenix.com/qb64wiki/index.php?title=C_Libraries&oldid=1241"