Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

DC synthesis of sync D-flip-flop maps to unnexpected flop...

Status
Not open for further replies.

dude123_400

Newbie level 2
Joined
Oct 1, 2009
Messages
2
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,308
Re: DC synthesis of sync D-flip-flop maps to unnexpected flo

The problem I'm having is reproducible w/ the class.db technology library distributed along w/ synopsys design compiler.

I have a very simple synchronous, active-high reset D-flipflop, which I expect to have it mapped
to the tech-mapped flop FD2S (class.db). However, DC synthesize it to the D-flipflop D1
w/ some extra logic fanning-in the 'reset' input pin CD. The mapping is correct, but I was hoping
it would not have the extra logic. I have tried different tricks w/ DC to make the mapping what I
expect, but I have not been successful.

Is there anybody out there who would know to tell DC to use the 'right' flop, tying its 'clear' input pin to the top-level reset pin?

Here's the information for reproducibility

Verilog source code

module a( b, c, d, e);
input b, c;
input [10] d, e;
reg [10] e;
//synopsys sync_set_reset c
always@(posedge b)
if (c)
e <= 2'd0;
else
begin
e <= d;
end
endmodule

Here's my synthesis script syn.tcl (remember to set link and target library to class.db prior to sourcing this script)
analyze -library work -format verilog test2.v
elaborate -library work a
compile
write -format verilog -output a_mapped.v

Here's the a_mapped.v
module a ( b, c, d, e );
input [10] d;
output [10] e;
input b, c;
wire n3, n6, n7, n8;

FD1 \e_reg[1] ( .D(n6), .CP(b), .Q(e[1]) ); // <<<<<<<<---- Not resettable flop
FD1 \e_reg[0] ( .D(n3), .CP(b), .Q(e[0]) ); // <<<<<<<<---- Not resettable flop
NR2 U7 ( .A(n7), .B(c), .Z(n6) );
IV U8 ( .A(d[1]), .Z(n7) );
NR2 U9 ( .A(n8), .B(c), .Z(n3) );
IV U10 ( .A(d[0]), .Z(n8) );
endmodule

Added after 1 hours 36 minutes:

For those of you who have acess to Synopsys documentation:

The answer lies in the following doc:
Design Compiler Optimization Reference Manual, version C-2009.06, Chapter 8 "Sequential Mapping".
 

jaydip

Member level 3
Joined
Jun 7, 2008
Messages
64
Helped
6
Reputation
12
Reaction score
1
Trophy points
1,288
Activity points
1,665
Re: DC synthesis of sync D-flip-flop maps to unnexpected flo

See comment in-line with the code :


always@(posedge b)
if (c)
e <= 2'd0; // Why did you use 2'b0? width of e is 10 bits, is it not?
else
begin
e <= d;
end
 

dude123_400

Newbie level 2
Joined
Oct 1, 2009
Messages
2
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,308
Re: DC synthesis of sync D-flip-flop maps to unnexpected flo

Sorry, my mistake while copying and pasting the code. Not sure if it had to do w/ the emoticons or something...

Anyway, the definition of 'd' and 'e' should be read as in :

input [1 : 0] d, e;
reg [1 : 0] e;

It turns out DC budgets the synthesis in a way that it will always add extra logic in front of the D-Flip-Flop Data pin unless you specify exactly what flop you want to use.

Fortunately, all these details are well documented in:
Design Compiler Optimization Reference Manual, version C-2009.06, Chapter 8 "Sequential Mapping".
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top