LP-MSPM0G3507: MATHACL data format, arctan(ATAN2) operating

Hi GR,

GR said:

Q1:

In TRM Table 5-1. MATHACL data formats.

SQm.n range is (-2^31)/2^n to (2^31)/2^n.

But in 5.2.4 Signed 32-bit numbers, integer bit is defined as m.

So I think SQm.n range is (-2^m)/2^n to (2^m)/2^n.

Is this correct?

Wikipedia has a good explanation on the range for Q number formats:https://en.wikipedia.org/wiki/Q_(number_format)#Characteristics

Here it is just simplified because we know m + n = 32 (using Arm Notation), so 2^(32-1)/2^n = 2^(32-n-1) = 2^(m-1).

GR said:

Is there any informatin to caliccurate  bus clocks that need for arctan(ATAN2) operation.

We do have this benchmarked as part of our IQMATH library: (https://dev.ti.com/tirex/content/mspm0_sdk_1_20_01_06/docs/english/middleware/iqmath/doc_guide/doc_guide-srcs/Users_Guide.html#benchmarks). That is with a set number of iterations though as far as I know. I will see if we have anything that denotes clock cycles for a single iteration. 

GR said:

Is there any infornation to error/accuracy of ATAN2 operation?

Checking on this for you. 

Best Regards,
Brandon Fisher

Hi Brandon,

Thanks for your information!

Brandon Fisher94 said:

Wikipedia has a good explanation on the range for Q number formats:https://en.wikipedia.org/wiki/Q_(number_format)#Characteristics

Here it is just simplified because we know m + n = 32 (using Arm Notation), so 2^(32-1)/2^n = 2^(32-n-1) = 2^(m-1).

I understand.

Brandon Fisher94 said:

Checking on this for you. 

Thanks for checking.

I will wait for answer.

Best regards,

GR

Hi Brandon,

Is there any update?

If there is, I am glad if you let me know.

Best regards,

GR