macro mapping

[omar97]

hello all,
i have a problem in macro mapping.
i am using icc2 tool (synopsys).
i have a memory in design (SRAM macros).
i generated it from memory compiler then i used generated .lef and .db to get ndm fram view and use it in ic compiler II.
i finished the design and when writing gds i show in gds viewer (Klayout) that macros are not mapped well.
everything is mapped succefully but macros are not.
i see that a wide peace of metal 1 and 2 and 3 are a macros, there is no polysi, no diffusion, only a wide width of metal with no thing belw.
when i write gds in icc2 i give it a memory gds to map layers.
can anyone help me to solve this problem, and tell me is that flow is correct to use memory as a macros?
thanks alot

 

[ThisIsNotSam]

the memory macro should have its own gds, generated from the compiler. when exporting a gds from ICC, you should tell it to merge your design gds with the memory gds. Is this the procedure you followed?

 

[omar97]

Yes, I did exactly as you say
As example
Command : write_gds ....... - merge_files memory.gds

 

[artecoll]

Hello.

i see that a wide peace of metal 1 and 2 and 3 are a macros, there is no polysi, no diffusion, only a wide width of metal with no thing belw.

If i correct understood - inside final gds there is no polysilicon and other layers of macro (except metal)?
Might be stupid question - there is no some errors in job log icc2 in write_gds?
Did you watched gds of macro? May be he initially generated like this (error of compiler).

and tell me is that flow is correct to use memory as a macros?

Yes, this right way.

 

[omar97]

Hello.

If i correct understood - inside final gds there is no polysilicon and other layers of macro (except metal)?
Might be stupid question - there is no some errors in job log icc2 in write_gds?
Did you watched gds of macro? May be he initially generated like this (error of compiler).

Yes, this right way.

Actually, there no errors or warning message during write_gds
Icc2 maps macros as it is a wide sheet of metal (about 75 um wide).
I tried to not give it a gds of memory and wait for any warning that is a missing cell, I found that there is no warning message appeared and it's mapped as it is a wide sheet of metal.

Actually, I don't know what I should I do??!!
There is many floating gates errors related to memories and other errors.

Can you tell me the flow of using macros from the beginning ?

 

[ThisIsNotSam]

I think this sheet of metal you see exported is blockage. It is not real. Now that would be problem #1, and still does not explain why the final gds has no memory macro. Try a more basic experiment, instantiate a design that is only a single memory. Export that gds. see what you get.

 

[oratie]

Still stupid question: did initial macro.gds contain all needed layers (poly, diff...)?

 

[omar97]

Still stupid question: did initial macro.gds contain all needed layers (poly, diff...)?

Yes, it has all layers.
I can see it in gds viewer (klayout)

--- Updated May 21, 2024 ---

I think this sheet of metal you see exported is blockage. It is not real. Now that would be problem #1, and still does not explain why the final gds has no memory macro. Try a more basic experiment, instantiate a design that is only a single memory. Export that gds. see what you get.

Yes it is a blockage in fram view.
But I want to get real shape in writing final gds.

 

[oratie]

Did you try this option "-lib_cell_view layout" ?

 

[omar97]

Did you try this option "-lib_cell_view layout" ?

Yes, it gives me a warning massage: the user specified view layout for lib_cell ....... is not available