Problem with SLL: "sll can not have such operands in this context" and bit-testing

[Frank Buss]

I've defined this signals:

signal accu : std_logic_vector(31 downto 0) := (others => '0');
signal data : integer range 0 to 255 := 0;

Within a process, which is triggered with like this:

if clk'event and clk = '0' then

I try to shift the accu (I'm trying to build a CPU) :

accu <= accu sll data;

But WebPACK ISE 8.1, with the service pack 3, says:

"sll can not have such operands in this context"

Even for this line it reports the same error:

accu <= accu sll 1;

How can I rotate the signal?


I have a similiar problem with my bit-test instruction:

z_flag <= accu(x_register);

This produces "Wrong index type for accu.", with the same definition for
x_register like for accu:

signal x_register : std_logic_vector(31 downto 0) := (others => '0');

[Mike Treseler]

Post by Frank Buss
How can I rotate the signal?

I would use numeric_std.rotate_left.

Here's a usage example:
http://home.comcast.net/~mike_treseler/sync_template.vhd

Post by Frank Buss
z_flag <= accu(x_register);
This produces "Wrong index type for accu.", with the same definition for
signal x_register : std_logic_vector(31 downto 0) := (others => '0');

To test a bit, you need a natural index.

-- Mike Treseler

[Frank Buss]

Post by Mike Treseler
I would use numeric_std.rotate_left.
http://home.comcast.net/~mike_treseler/sync_template.vhd

This doesn't work. The full VHDL code:

http://www.frank-buss.de/tmp/displaytest.vhd

ISE WebPack says:

Line 748. rotate_left can not have such operands in this context.

if I use the rotate_left. As a workaround I have used a counter and sliced
vector access to simulate sll and the bit test function, which works, but
which is slow.

It is my first VHDL program, so I would appreciate any other idea how to
improve it as well. It needs about 20,000 gates, which is nearly as much as
the 6809 implementation from OpenCores, so I think there is much to
improve, because my CPU has much less functionality than the 6809 :-)

Post by Mike Treseler
To test a bit, you need a natural index.

What is a natural index and how can I convert a signal to a natural index?

Another problem with my code: Looks like "char <= conv_integer(accu(7
downto 0));" or "accu <= accu + 1;" limits the range to 0..127, because the
charset, which is printed by the assembler program, displays the first 128
characters twice.

[Mike Treseler]

Post by Frank Buss

Post by Mike Treseler
I would use numeric_std.rotate_left.
http://home.comcast.net/~mike_treseler/sync_template.vhd

http://www.frank-buss.de/tmp/displaytest.vhd

It works if you declare unsigned types.

library IEEE;
use IEEE.STD_LOGIC_1164.ALL;
-- use IEEE.STD_LOGIC_ARITH.ALL;
-- use IEEE.STD_LOGIC_UNSIGNED.ALL;
use IEEE.NUMERIC_STD.ALL;

http://ghdl.free.fr/ghdl/IEEE-library-pitfalls.html

Post by Frank Buss
What is a natural index and how can I convert a signal to a natural index?

0, 1, 2 ...

This is a FAQ. Google a bit.

Post by Frank Buss
Another problem with my code: Looks like "char <= conv_integer(accu(7
downto 0));" or "accu <= accu + 1;" limits the range to 0..127, because the
charset, which is printed by the assembler program, displays the first 128
characters twice.

use IEEE.NUMERIC_STD.ALL;

http://www.csee.umbc.edu/help/VHDL/packages/numeric_std.vhd


-- Mike Treseler

[Frank Buss]

Post by Mike Treseler
It works if you declare unsigned types.
library IEEE;
use IEEE.STD_LOGIC_1164.ALL;
-- use IEEE.STD_LOGIC_ARITH.ALL;
-- use IEEE.STD_LOGIC_UNSIGNED.ALL;
use IEEE.NUMERIC_STD.ALL;
http://ghdl.free.fr/ghdl/IEEE-library-pitfalls.html

Thanks, I've corrected it, now rotate_left and rotate_right works.

Post by Mike Treseler

Post by Frank Buss
What is a natural index and how can I convert a signal to a natural index?

0, 1, 2 ...
This is a FAQ. Google a bit.

I've corrected this, too, with a subtype of the form "subtype byte is
natural range 0 to 255;" for all signals which needs this type, now the
program looks more clear and I can access a bit of a signal with a signal
as index.

Post by Mike Treseler

Post by Frank Buss
Another problem with my code: Looks like "char <= conv_integer(accu(7
downto 0));" or "accu <= accu + 1;" limits the range to 0..127, because the
charset, which is printed by the assembler program, displays the first 128
characters twice.

use IEEE.NUMERIC_STD.ALL;
http://www.csee.umbc.edu/help/VHDL/packages/numeric_std.vhd

Thanks, but the problem was, that the vdu module, which I used, has only
128 characters, but the characters above 128 are calculated block graphics,
if the 7th bit in the attribute RAM is set :-)

The new version:

http://www.frank-buss.de/tmp/displaytest2.vhdl

Now the last problem is to optimize it to synthesize not so many gates. Any
ideas?

[Duane Clark]

Post by Frank Buss
z_flag <= accu(x_register);
This produces "Wrong index type for accu.", with the same definition for
signal x_register : std_logic_vector(31 downto 0) := (others => '0');

Assuming you are using the numeric_std package, you could write that as:
z_flag <= accu(to_integer(unsigned(x_register)));
A bit cumbersome perhaps. If you are using x_register in many places for
a similar purpose, create an intermediate integer variable.