Intel® Quartus® Prime Standard Edition User Guide: Third-party Synthesis

ID 683796
Date 9/24/2018
Public
Document Table of Contents

2.9.6.3. Setting the dedicated_mult Attribute

To control the implementation of a multiplier in your HDL code, use the dedicated_mult attribute with the appropriate value as shown in the examples below.

Setting the dedicated_mult Attribute in Verilog HDL

//synthesis attribute <signal name> dedicated_mult <value>

Setting the dedicated_mult Attribute in VHDL

ATTRIBUTE dedicated_mult: STRING;
ATTRIBUTE dedicated_mult OF <signal name>: SIGNAL IS <value>;

The dedicated_mult attribute can be applied to signals and wires; it does not work when applied to a register. This attribute can be applied only to simple multiplier code, such as a = b * c.

Some signals for which the dedicated_mult attribute is set can be removed during synthesis by the Precision Synthesis software for design optimization. In such cases, if you want to force the implementation, you should preserve the signal by setting the preserve_signal attribute to TRUE.

Setting the preserve_signal Attribute in Verilog HDL

//synthesis attribute <signal name> preserve_signal TRUE

Setting the preserve_signal Attribute in VHDL

ATTRIBUTE preserve_signal: BOOLEAN;
ATTRIBUTE preserve_signal OF <signal name>: SIGNAL IS TRUE;

Verilog HDL Multiplier Implemented in Logic

module unsigned_mult (result, a, b);
   output [15:0] result;
   input [7:0] a;
   input [7:0} b;
   assign result = a * b;
   //synthesis attribute result dedicated_mult OFF
endmodule

VHDL Multiplier Implemented in Logic

LIBRARY ieee;
USE ieee.std_logic_1164.ALL;
USE ieee.std_logic_arith.ALL;
USE ieee.std_logic_unsigned.ALL;

ENTITY unsigned_mult IS
   PORT(
      a: IN std_logic_vector (7 DOWNTO 0);
      b: IN std_logic_vector (7 DOWNTO 0);
      result: OUT std_logic_vector (15 DOWNTO 0));
ATTRIBUTE dedicated_mult: STRING;
END unsigned_mult;

ARCHITECTURE rtl OF unsigned_mult IS
   SIGNAL a_int, b_int: UNSIGNED (7 downto 0);
   SIGNAL pdt_int: UNSIGNED (15 downto 0);
ATTRIBUTE dedicated_mult OF pdt_int: SIGNAL IS "OFF;
BEGIN
   a_int <= UNSIGNED (a);
   b_int <= UNSIGNED (b);
   pdt_int <= a_int * b_int;
   result <= std_logic_vector(pdt_int);
END rtl;