sky130_sram_* modules pins sizes changed to be slightly > 0.24

[ahmednofal]

sky130_sram_1kbyte_1rw1r_32x256_8 pins sizes changed to be slightly > 0.24 sky130_sram_1kbyte_1rw1r_8x1024_8 pins sizes changed to be slightly > 0.24

sky130_sram_2kbyte_1rw1r_32x512_8 pins sizes changed to be slightly > 0.24

sky130_sram_4kbyte_1rw1r_32x1024_8 pins sizes changed to be slightly > 0.24

sky130_sram_8kbyte_1rw1r_32x2048_8_pin_ext pins sizes changed to be slightly > 0.24

[mguthaus]

These macros are generated by OpenRAM. Any changes would need to be made in that project. Why are there pin sizes changed?

[RTimothyEdwards]

@mguthaus : The output pins need to meet the minimum area requirement for the metal that the label is on, or else it will generate an error for any output pin that is left unconnected. This is okay on input pins, since they must not be left unconnected, but output pins can be.

[mguthaus]

Yes, I wasn't intending pins to not be connected. There is actual metal in the GDS so just the LEF pin can be adjusted. I will fix it.

[RTimothyEdwards]

@mguthaus : So the question to you would be: Can you make the pin areas 0.38um x 0.64um instead of 0.38um x 0.38um so that when the LEF view is made and cuts off all the metal outside of the pin, the pin area by itself still meets the minimum metal area rule?

[mguthaus]

Yes, I'll update it.

[RTimothyEdwards]

@mguthaus : I guess normally one wouldn't just not connect the output pins. Our current special case is that we have to use the dual port design whether or not we need two ports, so if we end up using one port of a dual port memory, then the outputs are unconnected. One could make the case that a design should do something with those outputs, especially as this is an interim thing until we have single port blocks to use.

[RTimothyEdwards]

Apologies for the asynchronous out-of-order comments.