double support, even though --float_support is set to fpu32

Robert Wolfe

Hello,

I have a F28375D project, that has --float_support configured to fpu32. I'd like to leave that as a whole for the project. But there is a single case of an array that I need 64 bit resolution on. Is that possible?

Please advise,

Robert

over 2 years ago

0 Omer Amir over 2 years ago

TI__Mastermind 30185 points

Hello Robert,

You can use 32-bit or 64-bit floating-point, the only difference is that the float_support enables the hardware accelerator that's on the device. By not using fpu64, you can still do the same operations through software, but they may take longer than if the accelerator is used.

Keep in mind that the F2837xD family of devices does not have FPU64 support for the floating-point hardware accelerator.

Best regards,

Omer Amir

0 Robert Wolfe over 2 years ago in reply to Omer Amir

Mastermind 7860 points

Hello,

Thank you for the reply. I've declared a variable as double, but it is coming up as float, 2 words - I can see it in the Memory Browser window.

Can you think of any reason that might be?

Robert

0 Omer Amir over 2 years ago in reply to Robert Wolfe

TI__Mastermind 30185 points

Hello Robert,

What are you using to declare the 64-bit floating-point type? Are you using float64_t? Please share your code and show me what you're seeing in the Memory Browser.

Best regards,

Omer Amir

+1 Robert Wolfe over 2 years ago in reply to Omer Amir

Mastermind 7860 points

Hello,

It turns out that I needed to declare the variables as "long double" to get 64 bit floats.

Thanks,

Robert

C2000™︎ microcontrollers

C2000 microcontrollers forum

double support, even though --float_support is set to fpu32