Definition Multiply and Division in VHDL

[Adnan86]

Hi ,
I want execute this line in vhdl and i have no error but after compile it with Orcad and define the signal .. after RUN i have this error : Run time error accure at time 0 ns :
a , b, is real array matrix ,
c,d just real

a(0,0) <= b(0,0) * (c/d)
and if you have any idea for improve it , plz tell tanks

plz help :?:

 

[TrickyDicky]

Yes.
1. Dont use real type - its not supported for synthesis
2. Dont use OrCad for VHDL.

 

[Adnan86]

Yes.
1. Dont use real type - its not supported for synthesis
2. Dont use OrCad for VHDL.

IF i don't use real type so use what ?
I used integer but answer not float so i can't use it and if i use std_logic i have error for doing "/" and " * " .
instead ORCAD i use Modelsim but i have problem too ..

- - - Updated - - -

Yes.
1. Dont use real type - its not supported for synthesis
2. Dont use OrCad for VHDL.

IF i don't use real type so use what ?
I used integer but answer not float so i can't use it and if i use std_logic i have error for doing "/" and " * " .
instead ORCAD i use Modelsim too but i have problem ..

 

[TrickyDicky]

for floating point you need to use the floating point cores provided by whatever vendor you're using. but the main question there is - do you really need float? fixed point is probably waht you need (because thats basically an integer). FPGAs are not very good for floating point but perfect for fixed point.

 

[Adnan86]

THANKS for answer
I know , but I want to do this just as a project not for FPGA .
What do you recommend for this ... this line : a(0,0) <= b(0,0) * (c/d) that i said befor ,it's part of program : Matrix inversion . and now my question is : What's your recommend to do this . inevers (A) = 1 / determinant * (Adjoint (A)) ;
Any way I appreciate for your help .

 

[TrickyDicky]

well, if you dont care about synthesising it, then your code should work fine in modelsim.

 

[Adnan86]

I'll try it again in modelsim and we'll see...
any way Thanks again ...

 

[Adnan86]

still have problem in output answer ;
Any one here to recommend for VHDL code for matrix inversion use REAL type .
or give help for algorithm of multiplier and diversion for REAL type .

 

[TrickyDicky]

real can be multiplied and divided with the * and / operators.
what problems are you having?

 

[Adnan86]

for 1/x x it's real or integer answer always is zero .. and for (-y)/x answer always it's zero .. i'm confused ... for real type don't work
here my CODE
LIBRARY IEEE ;
USE IEEE.std_logic_1164.all ;
USE IEEE.numeric_std.all ;
USE IEEE.STD_LOGIC_UNSIGNED.ALL;
--use IEEE.MATH_REAL.ALL;
USE work.matrix_2by2.all ;


ENTITY matrix_inversion IS
PORT ( mat_in : IN matrix_2x2 ;
mat_invers : OUT matrix_2x2 );
END ;

ARCHITECTURE behav OF matrix_inversion IS



SIGNAL mat_det : integer ;
SIGNAL mat_adj : matrix_2x2 ;

BEGIN

-- PROCESS (mat_det)
-- BEGIN
-- IF mat_det /= 0.0 THEN
( mat_invers (0,0)) <= (mat_adj(0,0))/(mat_det) ;
( mat_invers (0,1)) <= ((mat_adj(0,1))/(mat_det)) ;
( mat_invers (1,0)) <= ((mat_adj(1,0))/(mat_det)) ;
( mat_invers (1,1)) <= (mat_adj(1,1))/(mat_det) ;
-- ELSE
-- mat_invers <= mat_in ;
-- END IF ;
-- END PROCESS ;
END ;

- - - Updated - - -

here determinant code

LIBRARY IEEE ;
USE IEEE.std_logic_1164.all ;
USE IEEE.numeric_std.all ;
-- use IEEE.MATH_REAL.ALL;
USE IEEE.STD_LOGIC_UNSIGNED.ALL;



USE work.matrix_2by2.all ;


ENTITY determinant IS
PORT ( mat_in : IN matrix_2x2 ;
mat_det : OUT integer );

END determinant;

ARCHITECTURE behav OF determinant IS
BEGIN
mat_det <= ( mat_in(0,0) * mat_in(1,1) ) - ( mat_in(0,1) * mat_in(1,0) ) ;
END behav;

 

[TrickyDicky]

1 is an integer. 1/real will give an integer result, which means answer will always be zero. You need to do 1.0/x for a real result (x must be an integer)

I see all of your matrix are integers. YOu need to cast these to real before you can get a real result. But in your code your output is an integer, so you will always get an integer value. Integer values always floor the result.

 

[Adnan86]

I tried it with real 1.0 / x x: integer
mat_invers (0,0) <= (mat_adj(0,0))* (1.0 /mat_det) ; mat_det : integer
but give me this error : No feasible entres for infix operator "/" if use real ( x ) ihaven't erro but
my output answer it's wrong ang give me this 4.656661e+298

 

[TrickyDicky]

you need to cast the integer to a real. / is not defined for real/integer:

1.0 / real(x)

 

[Adnan86]

I tried it with real 1.0 / x x: integer
mat_invers (0,0) <= (mat_adj(0,0))* (1.0 /mat_det) ; mat_det : integer
but give me this error : No feasible entres for infix operator "/" if use real ( x ) ihaven't erro but
my output answer it's wrong ang give me this 4.656661e+298

- - - Updated - - -

if use real ( x ) ihaven't erro but
my output answer for all matrix it's wrong and give me this 4.656661e+298

 

[FvM]

but give me this error : No feasible entres for infix operator "/" if use real ( x ) ihaven't erro but
my output answer it's wrong ang give me this 4.656661e+298

1. Other than e.g. C language, VHDL doesn't provide automatic type conversion, arithmetic operator can be only applied to the data types they are defined for.
2. The most plausible reason for getting 4.656661e+298 result is zero value of x. In the post #10 code, mat_det is never assigned a value.

 

[Adnan86]

thanks you Fvm and TrickyDicky , my problem almost solved .... I really appreciate for your attention .