32 bit integer, 32bit float and 64 bit doubles - huh?

[YISH313z]

A quick question in regards to these 3.

Scope says that it uses 32 bit integer for its sound and up to 40 bit float for its parameters and algorithms.

32 bit integer is pure mantissa, correct?

32bit float is pure 24bit mantissa and 8 bit exponent, correct?

What is 64 bit(Doubles)? especially in terms of the mantissa?

is it 32 bit integer with 32 bits of exponent, 32 bit float with 32 bits of exponent(making it still 24bit mantissa) or 24 bit with 40 bit exponent?

I am getting back into learning programming and this has me somewhat inquisitive?

[Warp69]

32 bit integer is pure mantissa, correct?

Yes, if you don't make your own fp system.

32bit float is pure 24bit mantissa and 8 bit exponent, correct?

Correct

What is 64 bit(Doubles)? especially in terms of the mantissa?

64bit float have a 53bit mantissa.

Just remember that you actual have sign bits - so encoding is abit different - but overall the above is correct when talking about precission and not encoding.

[YISH313z]

So that 53 bit mantissa is integer I assume?

[jksuperstar]

You should check out the IEEE 754 spec, which is what defines many of the terms and how they are calculated in processors that conform to the standard (such as Intel and Analog Devices). The DSPs also allow for some extended formats that are beyond the IEEE spec (40-bit float).

More info:
http://en.wikipedia.org/wiki/Binary32
http://en.wikipedia.org/wiki/Binary64

[YISH313z]

Cool, thanks for the links