How to avoid the division operator from hdlcoder?
2 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Dear All,
I've got the below verilog code when I generated with hdlcoder.
fx = srcX / dstX;
fy = srcY / dstY;
always @(srcX, srcY, dstX, dstY, recur, in, AP) begin
if (dstX == 3'b000) begin
test_function1_fixpt_tmp = 39'h7FFFFFFFFF;
end
else begin
test_function1_fixpt_cast = {srcX, 24'b000000000000000000000000};
if (dstX == 31'sd0) begin
test_function1_fixpt_div_temp = 36'hFFFFFFFFF;
end
else begin
test_function1_fixpt_div_temp = test_function1_fixpt_cast / dstX;
end
test_function1_fixpt_tmp = {3'b0, test_function1_fixpt_div_temp};
end
test_function1_fixpt_fx = test_function1_fixpt_tmp[33:24];
//fx = (float)srcX / (float)dstX; ;
As you can see that used a division operator " /".
But I'm confused quite a bit the output of hdlcoder.
Because I don't want to use "/" a division operator in RTL code, because utilization problem and Design Compiler and Xilinx VIVADO can not synthesize divider problem.
So I want to resolve this problem with another different way instead of "/".
What am I supposed to do to resolve this problem?
# update
I came across some relative posting as the below when I googling, But In my 2017B matlab version can't find any options. Would you please help and let me know how to resolve this problem?
"The RecipNewton implementation is a HDL Coder option on the Product block. You can reach the HDL Block Options on any HDL-supported block by right-clicking on the block in your model and choosing "HDL Code->HDL Block Properties...". On the Product block, you can then change the 'Architecture' field from the default 'Linear' to 'RecipNewton'. The RecipNewton method is only valid when the block is in division mode; HDL Coder will error out should you choose this architecture when the block is configured for multiplication.
The architecture field describes how the HDL code will be generated for the block. It does not affect Simulink simulation. If the implementation introduces latency, as the RecipNewton architecture can, HDL Coder will notify you of the additional latency that has been inserted. The generated HDL testbench will be automatically adjusted to reflect the change in latencies."
0 Kommentare
Antworten (1)
Bharath Venkataraman
am 31 Jul. 2018
2 Kommentare
Bharath Venkataraman
am 22 Aug. 2018
This is a Simulink block. If you drag the block into a Simulink model, and feed it dstY, it will return (1/dstY) at the output.
Siehe auch
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!