Convert 32 bit HEX to signed floating point

Question

antlhem 2017-11-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/364485-convert-32-bit-hex-to-signed-floating-point

评论： Walter Roberson 2022-11-30

So far I use the function hexsingle2num to convert a double type number (i.e. 41c1999a = 24.200000762939453) but when I want to use the same to convert ffffffff it gives me NaN, then I have to manually use nhex2dec('ffffffff',32) = -1 (correct answer),same with another example C0728F5C = -3.79. What I am looking for is a way to automatise this process since I have thousands of values. Is there a function in matlab that includes the conversion of HEX to SIGNED floating point?

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Maitreyee Mordekar 2017-11-13

6
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/364485-convert-32-bit-hex-to-signed-floating-point#answer_290818

编辑：KSSV 2019-12-10

在 MATLAB Online 中打开

Hi,

By hexsingle2num function I assume that you are referring to the following File Exchange submission: https://www.mathworks.com/matlabcentral/fileexchange/6927-hexsingle2num

This is one of the several submissions in MATLAB File Exchange on MATLAB Central which is a forum for our product users to interact, exchange information and knowledge, without MathWorks' involvement. You may contact the author of this submission directly for specific questions about the implementation.

However, you can obtain the same results using the following function:

typecast(uint32(hex2dec('C0728F5C')),'single')

Also, if you still need a way to convert hexstring to double precision number then you can consider using hex2num: https://www.mathworks.com/help/matlab/ref/hex2num.html

Hope that helps!

5 个评论
显示 3更早的评论隐藏 3更早的评论

Walter Roberson 2021-2-4

Note:

If you have a stream of bytes intended to represent a single precision number, then you need to know the order the bytes are in.

The international standard for network transmission that does not specifically specify otherwise, is that the most significant byte should be first (that is the byte that includes the sign and exponent) and the least significant byte should be last. This is known as "big-Endian" order, or more formally as "network byte order"

It is not, however, the order used by Intel x86 or x64 architecture, which puts the least significant byte first.

dec2hex and num2hex output in big-Endian, most significant byte first. But that is not the memory order of Intel.

This matters because when you typecast() bytes to single precision, the order used is memory order of the host system. '01020304' as output from num2hex would have 01 in the most significant byte, but typecast(uint8([1,2,3,4]),'single') would have the 01 in the least significant byte. typecast does not automatically adjust between big-Endian and little-Endian.

Because of these factors, you may need to use swapbytes() on the single() after the typecast.

If your source of data is a program designed for Windows, then a lot of the time they never bothered to think about byte order and just emit binary that is little-Endian and that often does not need to swapbytes for use on Intel architecture.

By the way, matlab .mat files contain a header that tells matlab which order the file was written in, so matlab automatically adjusts data read from mat files.

Badr Sahib 2022-11-30