Converting a fixed point Matlab code to Verilog - arrays

Question

I have a fixed point Matlab code and it needs to be converted to Verilog. Below is the Matlab code. yfftshift is 5000x0 and y2shape 100x50.
rows=100;
colms=50;
r=1;
for m=0:colms-1
for n=0:rows-1
y2shape(n+1,m+1)=yfftshift(r,1);
r=r+1;
end
end
How can I create memories in Verilog and call them inside the for loop?

Answer 1

The easiest way to handle fixed precision in Verilog is to introduce a scale factor and allocate sufficiently large registers to hold the maximum value. For example, if you know that the maximum value of your numbers will be 40, and three digits of precision to the right of the decimal place are OK, a scaling factor of 1000 can be used with 16-bit registers. Verilog treats unsigned numbers, so if values can be negative, it's necessary to add "signed" to the declarations. The Verilog could be:
`define NUMBER_ROWS 100
`define NUMBER_COLS 50
`define MAX_ROW (`NUMBER_ROWS-1)
`define MAX_COL (`NUMBER_COLS-1)
module moveMemory();
reg clk;
reg [15:0] y2shape [`MAX_ROW:0][`MAX_COL:0];
reg [15:0] yfftshift [`NUMBER_ROWS * `NUMBER_COLS:0];
integer rowNumber, colNumber;
always #(posedge clk)
begin
for (rowNumber = 0; rowNumber < `NUMBER_ROWS; rowNumber = rowNumber + 1)
for (colNumber = 0; colNumber < `NUMBER_COLS; colNumber = colNumber + 1)
y2shape[rowNumber][colNumber] <= yfftshift[rowNumber * `NUMBER_COLS + colNumber];
end
endmodule
This is OK for an FPGA or simulation project, but for full custom work, an SRAM macro would be used to avoid the die area associated with 16,000 registers. For an FPGA implementation, you've probably already paid for the 16K registers, or you may be able to do some extra work get the synthesizer to map the registers to an on-chip SRAM.
The test bench:
// Testing code
integer loadCount, rowShowNumber, colShowNumber;
initial
begin
// Initialize array with some data
for (loadCount=0; loadCount < (NUMBER_ROWS *NUMBER_COLS); loadCount = loadCount + 1)
yfftshift[loadCount] <= loadCount;
clk <= 0;
// Clock the block
#1
clk <= 1;
// Display the results
#1
$display("Y2SHAPE has these values at time ", $time);
for (rowShowNumber = 0; rowShowNumber < `NUMBER_ROWS; rowShowNumber = rowShowNumber + 1)
for (colShowNumber = 0; colShowNumber < `NUMBER_COLS; colShowNumber = colShowNumber + 1)
$display("y2shape[%0d][%0d] is %d ", rowShowNumber, colShowNumber, y2shape[rowShowNumber][colShowNumber]);
end
The simulation results for NUMBER_ROWS=10, NUMBER_COLS=5
Y2SHAPE has these values at time 2
y2shape[0][0] is 0
y2shape[0][1] is 1
y2shape[0][2] is 2
y2shape[0][3] is 3
y2shape[0][4] is 4
.
.
.
y2shape[9][2] is 47
y2shape[9][3] is 48
y2shape[9][4] is 49