---

GLAST/LAT > DAQ and FSW > FSW > Doxygen Index > LDT / V0-2-0

Constituent: encdec     Tag: rad750

---

         

---

BWP.h File Reference

Bit Word interface definitions, packing specific routines. More...

#include "LDT/BW.h"

Include dependency graph for BWP.h:

This graph shows which files directly or indirectly include this file:

Functions
BW_vector	BWP_collapseL (unsigned int val, unsigned int msk)
	Collapses the bit pattern in val considering only those bits that are set in msk. The bit pattern is left justified.
BW_vector	BWP_collapseR (unsigned int val, unsigned int msk)
	Collapses the bit pattern in val considering only those bits that are set in msk. The bit pattern is right justified.
BW_vector	BWP_convertL (unsigned int val, unsigned int nsig, unsigned int nexp)
	Encodes small integer values that are embedded in a large field.
BW_vector	BWP_convertR (unsigned int val, unsigned int nsig, unsigned int nexp)
	Encodes small integer values that are embedded in a large field.

---

Detailed Description

Bit Word interface definitions, packing specific routines.

Author:

JJRussell - russell@slac.stanford.edu

   CVS $Id: BWP.h,v 1.1 2006/09/13 17:09:08 russell Exp $

---

Function Documentation

BW_vector BWP_collapseL (  unsigned int  val, unsigned int  msk )

Collapses the bit pattern in val considering only those bits that are set in msk. The bit pattern is left justified. Returns: The bit pattern and the number of significant bits Parameters: val  The word to extract the bits from msk  The mask of bits to consider Note: There should be a PPC specific version of this. The rotate left and insert under mask is perfect for this routine.

BW_vector BWP_collapseR (  unsigned int  val, unsigned int  msk )

Collapses the bit pattern in val considering only those bits that are set in msk. The bit pattern is right justified. Returns: The bit pattern and the number of significant bits Parameters: val  The word to extract the bits from msk  The mask of bits to consider

BW_vector BWP_convertL (  unsigned int  val, unsigned int  nval, unsigned int  nexp )

Encodes small integer values that are embedded in a large field. Returns: The left justified bit pattern and the number of significant bits. Parameters: val  The value to be encoded nval  The bit field width of val nexp  The number of bits to devote to the exponent encoding See BWP_convertR for a complete description. (This routine just produces a right justified version of the bit pattern.)

__inline BW_vector BWP_convertR (  unsigned int  val, unsigned int  nval, unsigned int  nexp )

Encodes small integer values that are embedded in a large field. Returns: The right justified bit pattern and the number of significant bits. Parameters: val  The interger value to be encoded nval  The bit field width of val nexp  The number of bits to devote to the exponent encoding This encoding is most effective on small integer values that are embedded in a larger field. The val is broken into two pieces. The first piece a field consisting of the number of leading zeros, while the second piece is all the rest, sans the first bit, which, in all but one case must be a 1, so therefore need not be encoded. That special case is when the number of leading zeroes is greater than or equal to (1 << nexp) - 1. Since only that many zeroes can be encoded, one cannot make any assumptions about the value of the next bit. Example Consider the number 0x1fff embedded in a 25-bit field using nexp = 4. The call would be bwv = BWP__convertR (0x1fff, 25, 4); One can encode up to 15 leading zeroes. In this case there are only 8. This gets encoded as EXP = 7 (this is the usual n-1), MANTISSA = 0xfff or 0x7fff, NBITS = 16. Warning: As currently written, this routine can safely accomodate nexp = 5 27-bit integer nexp = 4 28-bit integer nexp = 3 29-bit integer nexp = 2 30-bit integer nexp = 1 31-bit integer Of course, some of these are silly combinations in the sense that the would produce no compression and would more than likely result in an expansion.

---