large increase in area for ff compared to buf for upsize

[maliberty]

@precisionmoon noticed this large jump in area when upsizing a sequential compared to a buffer. Is this expected?

PDK	BUFX1	BUFX2	BUFX1 -> BUFX2 area increase	DFFX1	DFFX2	DFFX1 -> DFFX2 area increase	DFF1->2 /BFF1->2
ihp	sg13g2_buf_1	sg13g2_buf_2		sg13g2_dfrbp_1	sg13g2_dfrbp_2
	7.2576	9.072	1.8144	47.1744	54.432	7.2576	4

[KrzysztofHerman]

@maliberty thank you for reporting this issue. At a first glance it should not be the case since as shown in the table (I also checked it in LEF and LIB files) the areas of the cells does not increase much as you change the cell type. I have also checked the spice netlist of all 4 circuits and they have identically sized transistors at the output, what should result in equal drive strengths for buffers and FF with the same postfix. I suppose that the increase in the area produces while placing the cells. Could you please mention the most critical parameters which are driving placer/resizer ?

[maliberty]

Drive strength, pin capacitance, and area are the most critical. In other PDKs we observe an smaller ratio of increase. To me this suggests some inefficiency in the cell layout.