When creating code to read a SEG-Y file, I get the IEEE and IBM Floating Point format data. IBM floating point numbers are represented by one bit for the sign (S), 7 bits for the exponent, and 24 bits for the fraction. The exponent is to the base 16 (not 2), and has a bias of 64. Because of the base 16, the fraction may have up to three leading zeros. In this manner, some of the exponential value may be assumed by the fraction, increasing its range, at the cost of some precision. Normalization to the leading bit is not possible. Floating point numbers have no complement (neither one’s nor two’s complement). They are in sign-magnitude format (the sign being bit 31). You get the more information with the table below :
My name is Toto Sugito. This is my notes when I try something. Maybe, this is NOT THE BEST SOLUTION for your problems. If you have other method or idea that better with my post, please share in this blog. Thank for visiting my site.