Cannot Convert RAM Megafunctions with Max 10

[n-slipstream]

Synthesis Error:

Info (276014): Found 1 instances of uninferred RAM logic
	Info (276013): RAM logic "Datapath:Datapath|RAM_AD9914_0:RAM_AD9914_0|ram_instance" is uninferred because MIF is not supported for the selected family

Error (276003): Cannot convert all sets of registers into RAM megafunctions when creating nodes. The resulting number of registers remaining in design exceeds the number of registers in the device or the number specified by the assignment max_number_of_registers_from_uninferred_rams. This can cause longer compilation time or result in insufficient memory to complete Analysis and Synthesis

If I change the project to use 'Single Compressed Image with Memory Initialisation' it builds fine but I must use 'Dual Compressed Images' as this will be a RSU (remote system upgrade) project.

It looks like this is the cause when the ram is defined:

Is there an alternative way to achieve this

library IEEE;
use IEEE.STD_LOGIC_1164.ALL;
use IEEE.NUMERIC_STD.ALL;
use WORK.COMMON.ALL;

entity RAM_DAC is
	generic ( 
			NUM_OF_REGISTERS : natural := 1023;							-- Number of registers containing data to send on SPI
			ADDR_WIDTH : natural := 10;									
			DATA_SIZE : natural := 12); 								-- 12 bit word sets analog output 
    Port ( 
			CLK : in  STD_LOGIC;
			WriteEn : in  STD_LOGIC;
			DataIn : in  STD_LOGIC_VECTOR (11 downto 0);
			Address : in  UNSIGNED (ADDR_WIDTH-1 downto 0);
			WR_ADDR : IN UNSIGNED (ADDR_WIDTH-1 downto 0);
			FRAME_SIZE : in STD_LOGIC_VECTOR (15 downto 0);
			DataOut : out  STD_LOGIC_VECTOR (DATA_SIZE-1 downto 0)
		 );
end RAM_DAC;

architecture Behavioral of RAM_DAC is
	
	-- Declare memory parameters
	type ram_type is array (0 to NUM_OF_REGISTERS) of STD_LOGIC_VECTOR(DATA_SIZE-1 downto 0);
	signal ram_instance: ram_type := (
        0 => "100000000000",         
        1 => "100000000000",  -- bit pattern 1000 0000 0000 is 0V on the DAC output, this is the preferred default output
        2 => "100000000000",       
        3 => "100000000000",    
        4 => "100000000000",       
        5 => "100000000000",       
	    6 => "100000000000",     

[FvM]

Unfortunately, dual image configuration scheme doesn't allow memory initialization. It's clearly stated in the M10 handbook.

Not sure if it's due to limitations of the configuration features implemented in M10 silicon or a software restriction we might get rid of in future Quartus releases, at least for designs that don't utilize 100 percent of configuration memory capacity.

Frank