I am working on a new C++ design for an embedded system. It uses an ARM
Cortex-A9 running at around 600 MHz and an RTOS. The legacy code makes use
of 32-bit floats in most places but it sometimes uses fixed-point numbers.
For example, a 28-bit fixed-point number is used to represent the power set
point of a device (that has hardware than can make use of that resolution).
Power for the microprocessor is not a concern and we have plenty of margin
on bandwidth so I’m tempted to leverage MPFR for a more consistent use of
types.
Generally speaking our math is not that intense. We do some linear and
quadratic mapping of numbers to do things like convert ADC values to
calibrated physical values. We convert between linear units and dB for gain
and power.
Any thoughts or caveats? Gotta run, I can hear a mob of my peers forming J.
Thanks,
Kenny
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Everyone,
I am working on a new C++ design for an embedded system. It uses an ARM
Cortex-A9 running at around 600 MHz and an RTOS. The legacy code makes use
of 32-bit floats in most places but it sometimes uses fixed-point numbers.
For example, a 28-bit fixed-point number is used to represent the power set
point of a device (that has hardware than can make use of that resolution).
Power for the microprocessor is not a concern and we have plenty of margin
on bandwidth so I’m tempted to leverage MPFR for a more consistent use of
types.
Generally speaking our math is not that intense. We do some linear and
quadratic mapping of numbers to do things like convert ADC values to
calibrated physical values. We convert between linear units and dB for gain
and power.
Any thoughts or caveats? Gotta run, I can hear a mob of my peers forming J.
Thanks,
Kenny