DSP programming languages are assembly and C
Let's start with the compilation basics, which are several common operations
MOVE Transfer, ADD addition, sub subtraction, Mpym multiplication, MACM multiply (MACM a,b,c = "C=a*b+c")
Shift: Shift-out bit discard, vacancy 0, left n is equal to n times 2, right shift n is equal to 2 of the n-th square
For the program, mainly consists of data and code, the data is divided into initialized data and uninitialized data, such as you int A; this is uninitialized, int a = 1; this is initialization.
assembly language to segment the program space, storage data and code, note that only the variables may not be initialized, the code must be initialized
Why fragmentation, for management convenience, such as uninitialized variables will be assigned a uniform value 0
Segment not initialized:
. BSS: Holds uninitialized global variables reserved by the system, such as Int,long, where the number of bits is determined by the system
. Usect: Store user-defined, uninitialized global variables, such as defining an ABC type set to 100-bit length
Initialize segment:
. Data: A global variable that holds the initialization of the system reservation
. Sect: A global variable that holds user-defined initialization
. Text: Store Your program code
Like. BSS h,3
is to assemble a variable called H-bit length 3 into a. BSS segment
Let's look at a piece of code
. Mmregs
def start
. Text
start: MOV #0X000A, AC0
Amov #0X018001, XAR2
Amov #0X018002, XAR3
MOV #0X0005, *ar2
MOV #0X0002, *ar3
MACM *ar2,*ar3,ac0 here
: B here;
. End
The. Text above is the code snippet that tells the assembler to assemble the following code into the program space
Note Several rules for registers
The description with X is the secondary register
MACM is a multiplicative, the above code is compiled after the AC0 value is
A+2*5=20
20 in 16 notation, full 10 in 1 is 14
Then AC0 is a 40-bit register, so 0x0000000014,1 that is high, 4 is low, low is full of high-level into 1, converted into a value is 1*16+4=20
If you encounter decimal multiplication, see whether it is fixed-point DSP or floating-point DSP, Low-point, not allowed, fast, power consumption, floating point opposite
Fixed-point calculation formula, Xq = x*2 Q Power Q is the calibration number
Like Q=8, writing 0.3*5 programs.
When writing a program, the 0.3 needs to be converted to 0.3*2 8 power, or 0.3*256, and then with the converted value of the operation