rom generating code problem

That is a massive rom. Im not surprised it takes a long time...

You cant set Rom values to be Z. It has to be '0' or '1'. 'Z' can only be covered at the tri state buffers in the pins.

Can you post your rom code?

And the error in simualtion is that your address value is coming as 971456 which is clearly out of range of the ROM. You need to sort your addresses out (ie. you need fewer bits in the address bus). 384000 is not a 2^n. When implemented, it will scale up to 2^19.

what is a massive rom..

my rom contains data of an image's pixels

here is its code

entity memo_rom is

generic (pixel : integer := 9 ; img_width_cols : integer := 320; img_height_rows : integer := 240; addr_size : integer := 20);

port(clk : in std_logic;

en : in std_logic;

addr : in std_logic_vector(addr_size-1 downto 0);

data_out: out std_logic_vector(pixel-1 downto 0)

);

end entity;

architecture behavioral of memo_rom is

type ROM_Array is array ((img_width_cols*img_height_rows)-1 downto 0) of std_logic_vector (pixel-1 downto 0);

--image size 240*320 (row*col) pixels

constant rom_data: ROM_Array := (

0 =>"000000000",

1 =>"000000000",

2 =>"000000000",

3 =>"000000000",

4 =>"000010111",

5 =>"010100011",

6 =>"010010001",

......

......

......

76797 =>"000000000",

76798 =>"000000000",

76799 =>"000000000",

OTHERS => "ZZZZZZZZZ"

) ;

begin

data_out <= rom_data(conv_integer(addr)) when(en = '1') else "ZZZZZZZZZ";

end behavioral;

i've changed the "ZZ..ZZ" to "00..00" but nothing changed

there is nothing in my code that reaches this value 971456

Thanks alot,

Niveen

Possible issue 1: you still have ZZs in the "others" case. Avoid having Z and bi-directional signals on you design, except on the I/O pins.

Possible issue 2: That code can't be implemented in a RAM block and thus will be implemented into logic elements, which increases time and design. Try to use a design that can be implemented in a RAM block and matches Quartus ROM inference templates.

An alternative I like for such cases is to LPM_ROM block primitive with .HEX files.

You can also use .MIF instead of .HEX. .MIF have a simpler format, but they're not supported by ModelSIm.

rom : lpm_rom

generic map (

LPM_WIDTH => 8,

LPM_WIDTHAD => 14,

LPM_FILE => "rom.hex",

LPM_ADDRESS_CONTROL => "UNREGISTERED",

LPM_OUTDATA => "REGISTERED"

)

port map(

outclock => Clk,

address => A,

q => D

);

thanks alot rbugalho..

i've already replaced ZZs by 00s but same problem..

for using LPM_ROM..would you plz declare more

about rom.hex file is a file that contains my memo data in hex or wt??

execuse my lack of knowledge.. as i'm new in the field of zero experience..

thanks,

Niveen

Tricky,

Yes, you do. I wasn't sure if the constant array + synchronous read gets inferred but I just tested on Quartus 10 and it does.

So, simply replacing the

data_out <= .... 

with

process(clk) begin
if rising_edge(clk) then
if en = '1' then
	data_out <= rom_data(to_integer(unsigned(addr)));
end if;
end if;
end process;

Niveen,

no, I was talking about an "Intel HEX" file.

http://en.wikipedia.org/wiki/intel_hex

Quartus has a built-in editor which lets you write such a file by hand.

You'll also find a number of tools in the internet to write/convert data into that format.