Automatic inference from general VHDL code in Quartus II

[Thomas Thorsen]

I have been trying to create a 20bit adder/subtractor in VHDL, but no matter
how i define it, Quartus II won't infer the appropriate megafunction
lpm_add_sub. Even if i use the code that can be downloaded from:
http://www.altera.com/support/examples/vhdl/vhd-add-sub.html

it still does not infer the lpm_add_sub as it should, according to the
description. The reason i want this, is that my adder takes up 41 logic
cells. If i directly instantiate the lpm_add_sub it only takes up 21 logic
cells (the 2's complement negation is merged into the same logic cells).
However, if i directly instantiate the lpm_add_sub the code is not general
VHDL anymore and i need that to (for simulation in ModelSim that does not
know about the megafunctions). So question is, how do i tell Quartus II that
my adder should be implemented as an lpm_add_sub (automatic inference)?

Any helpfull comments are greatly appreciated.

 

Thomas Thorsen wrote:

I have been trying to create a 20bit adder/subtractor in VHDL, but no matter
how i define it, Quartus II won't infer the appropriate megafunction
lpm_add_sub. Even if i use the code that can be downloaded from:
http://www.altera.com/support/examples/vhdl/vhd-add-sub.html

it still does not infer the lpm_add_sub as it should, according to the
description. The reason i want this, is that my adder takes up 41 logic
cells. If i directly instantiate the lpm_add_sub it only takes up 21 logic
cells (the 2's complement negation is merged into the same logic cells).
However, if i directly instantiate the lpm_add_sub the code is not general
VHDL anymore and i need that to (for simulation in ModelSim that does not
know about the megafunctions). So question is, how do i tell Quartus II that
my adder should be implemented as an lpm_add_sub (automatic inference)?

Any helpfull comments are greatly appreciated.

The code from their web site does infer an lpm_add_sub; at least it
does for me.

Verify that it does for you, and then you can just instantiate that
entity.

Modelsim does know about megafunctions. You just need to add the
correct library.

 

[Mike Treseler]

Thomas Thorsen wrote:

I have been trying to create a 20bit adder/subtractor in VHDL, but no matter
how i define it, Quartus II won't infer the appropriate megafunction

see pg 6-7:
http://www.altera.com/literature/hb/qts/qts_qii51007.pdf

-- Mike Treseler