Scribd
Upload
89 views2 pages

Rams Models in VHDL: C. E. Stroud, Dept. of Ece, Auburn Univ. 1 8/06

This document discusses a VHDL model for implementing a RAM using an array of std_logic_vectors. The model is parameterized to allow for different word and address sizes. It initializes the first few RAM locations and the rest to zeros. During reads and writes it converts the address to an integer to access the RAM array. The model is technology independent but inefficient for synthesis as it requires many FPGA slices. For synthesis, a block RAM model specific to the target FPGA should be used instead to gain the benefits of the dedicated block RAM resources.

Uploaded by

Daniel García
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
89 views2 pages

Rams Models in VHDL: C. E. Stroud, Dept. of Ece, Auburn Univ. 1 8/06

This document discusses a VHDL model for implementing a RAM using an array of std_logic_vectors. The model is parameterized to allow for different word and address sizes. It initializes the first few RAM locations and the rest to zeros. During reads and writes it converts the address to an integer to access the RAM array. The model is technology independent but inefficient for synthesis as it requires many FPGA slices. For synthesis, a block RAM model specific to the target FPGA should be used instead to gain the benefits of the dedicated block RAM resources.

Uploaded by

Daniel García
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

RAMs Models in VHDL

A data type definition that is useful is constructing an array of std_logic_vectors. This can be
used as a RAM as illustrated in the example model given below:
library IEEE;
use IEEE.std_logic_1164.all;
use IEEE.std_logic_unsigned.all;
entity RAM is
generic (K: integer:=8;
-- number of bits per word
W: integer:=8);
-- number of address bits
port (WR: in std_logic;
-- active high write enable
ADDR: in std_logic_vector (W-1 downto 0);
-- RAM address
DIN: in std_logic_vector (K-1 downto 0);
-- write data
DOUT: out std_logic_vector (K-1 downto 0));
-- read data
end entity RAM;
architecture RAMBEHAVIOR of RAM is
subtype WORD is std_logic_vector ( K-1 downto 0);
-- define size of WORD
type MEMORY is array (0 to 2**W-1) of WORD;
-- define size of MEMORY
signal RAM256: MEMORY;
-- define RAM256 as signal of type MEMORY
begin
process (WR, DIN, ADDR)
variable RAM_ADDR_IN: integer range 0 to 2**W-1;
-- to translate address to integer
variable STARTUP: boolean :=true;
-- temporary variable for initialization
begin
if (STARTUP = true) then -- for initialization of RAM during start of simulation
RAM256 <= (0 => "00000101",
-- initializes first 5 locations in RAM
1 => "00110100",
-- to specific values
2 => "00000110",
-- all other locations in RAM are
3 => "00011000",
-- initialized to all 0s
4 => "00000011",
others => "00000000");
DOUT <= "XXXXXXXX"; -- force undefined logic values on RAM output
STARTUP :=false; -- now this portion of process will only execute once
else
RAM_ADDR_IN := conv_integer (ADDR);
-- converts address to integer
if (WR='1') then
-- write operation to RAM
RAM256 (RAM_ADDR_IN) <= DIN ;
end if;
DOUT <= RAM256 (RAM_ADDR_IN); -- always does read operation
end if;
end process;
end architecture RAMBEHAVIOR;
The model above is technology independent and good for simulation. It is parameterized such
that it can easily be used to emulate many RAMs of various sizes with similar behavior. It can
now be used as the basis for simulation and verification of other hierarchical functions such as a
First-In-First-Out (FIFO) memory. However, this model is not usually good for synthesis. For
C. E. Stroud, Dept. of ECE, Auburn Univ.

8/06

RAMs Models in VHDL


example, the model will synthesize in ISE but it requires 1736 slices in a Spartan 3 200 FPGA
(see Figure 1) but can easily fit in only one block RAM in the device illustrated in the upper left
hand corner in Figure 1 and only requires 2K of the 18K total memory space in that block RAM.
Therefore, one would want to instantiate the Xilinx specific block RAM model prior to final
synthesis in order to obtain the more efficient (and higher performance) implementation of a
block RAM. While the Xilinx specific block RAM model can be used for simulation as well as
synthesis, it is not technology dependent and requires

Figure 1. RAM model synthesized in flip-flops of FPGA.

C. E. Stroud, Dept. of ECE, Auburn Univ.

8/06

You might also like