Doubts regarding pin sizes and locations

[b224hisl]

Hello, recently I have found something really strange about the pin size and location.

For example, when using the sky130hd, after I place a pin, then I execute

set dbBlock [ord::get_db_block]
set db_bterm [$dbBlock findBTerm clk]

then I can get the size and location of this pin. As you can see, the Bterm's location is consistent with what sta::port_location produces:

The pin sizes in place_pin and Bterm are consistent with the dimensions in the DEF file as well.

However, when it comes to GF180, I also place a clk pin by place_pin -pin_name clk -pin_size {2.56 0.88} -location {4 150} -layer Metal5, the gui image sounds right:

But the dx and dy in the Bterm is twice larger, remember I have set -pin_size {2.56 0.88} when I place the pin:

Also, the location is not consistent with what sta::port_location produces

And here is the def file

it can be summaried as:
def file = Bterm
gui image = place_pin = sta::port_location
However, there is no dispute regarding this in the case of Sky130hd. I'm unsure if this is a bug specific to GF180 or if it's a misunderstanding on my part. I would appreciate some guidance on this matter.

[rovinski]

This is the concept of database units. Database units defined as X/um where X is the database unit scale. X is defined in the technology LEF.

For sky130 the DBU scale is 1000.

OpenROAD-flow-scripts/flow/platforms/sky130hd/lef/sky130_fd_sc_hd.tlef

Line 27 in 5bd5bff

DATABASE MICRONS 1000 ;

In your example, dx = 2500 dbu = 2500/1000 um = 2.5um.

For gf180 the DBU scale is 2000.

OpenROAD-flow-scripts/flow/platforms/gf180/lef/gf180mcu_2LM_1TM_30K_7t_tech.lef

Line 40 in 5bd5bff

DATABASE MICRONS 2000 ;

In your example, dx = 5120 dbu = 5120/2000 um = 2.56um.

[rovinski]

You can use the Tcl function ord::dbu_to_microns to convert from DBU to um if you need to.

[b224hisl]

btw, in gf180mcu_2LM_1TM_30K_7t_tech.lef

Do these PITCH, WIDTH, AREA, RECT, SPCAE, RANGE, etc. use this dbu ? for example, the thickness is 0.99/2000 um ?

[rovinski]

No, those are all in um.

[b224hisl]

thank you very much！