Data range 32-Bit Float

[John Gerhardy]

Hello,

Creating a data set with 30 Million entries at an intervat of 0.25, the information after the decimal point is cut off after 4194303.75. It looks like this:
Data-Nr. Value
16777215 4194303,25
16777216 4194303,5
16777217 4194303,75
16777218 4194304
16777219 4194304
16777220 4194304,5
16777221 4194305

in the documentation the 32-Bit float is described as going form -3.4e+38 to +3.4e+38.

Could someone explain how the bits are divied up, i.e. how many significands and exponands? Or where can I read more about the format in Flexpro?

[John Gerhardy]

Hello,

Creating a data set with 30 Million entries at an intervat of 0.25, the information after the decimal point is cut off after 4194303.75. It looks like this:
Data-Nr. Value
16777215 4194303,25
16777216 4194303,5
16777217 4194303,75
16777218 4194304
16777219 4194304
16777220 4194304,5
16777221 4194305

in the documentation the 32-Bit float is described as going form -3.4e+38 to +3.4e+38.

Could someone explain how the bits are divied up, i.e. how many significands and exponands? Or where can I read more about the format in Flexpro?

[Bernhard Kantz]

The bits for 32-Bit floating point values are divided up like this: 23 bit for the mantissa, 8 bits for the exponent. Floating point values represent a single-precision 32-bit IEEE 754 value. The type has an accuracy of about seven digits.
A detailed description of the floating point type can be found for instance in WikiPedia at http://en.wikipedia.org/wiki/Floating_point in english or at http://de.wikipedia.org/wiki/Gleitkommazahl in german.

Support
support@weisang.com

[John Gerhardy]

At 22 bits (23 bits minus one sign bit) I get 2^22=4194304 which matches to what Flexpro made. Got it. Thanks for the clarification and for the link.

[Author]

Posts