pith. sign in

arxiv: 1410.1339 · v1 · pith:UKLUUCNUnew · submitted 2014-10-06 · ⚛️ physics.ins-det

Analysis of boundary point (break point) in Linear Delay Model for nanoscale VLSI standard cell library characterization at PVT corners

classification ⚛️ physics.ins-det
keywords delaylinearpointwillbreakmodelregiontime
0
0 comments X
read the original abstract

In VLSI chip design flow, Static Timing Analysis (STA) is used for fast and accurate analysis of data-path delay. This process is fast because delay is picked from Look Up Tables (LUT) rather than conventional SPICE simulations. But accuracy of this method depends upon the underlying delay model with which LUT was characterized. Non Linear Delay Model (NLDM) based LUTs are quite common in industries. These LUT requires huge amount to time during characterization because of huge number of SPICE simulations done at arbitrary points. To improve this people proposed various other delay models like alpha-power and piecewise linear delay models. Bulusu et al proposed Linear Delay Model(LDM) which reduces LUT generation time to 50 percent. LDM divides delay curve w.r.t input rise time(trin) into two different region one is linear and other is non-linear. This boundary point between linear and non- linear region was called break point (trb). Linear region will be done if we simulate at only two points. This advantage will be possible by having knowledge of this break point at various PVT corners. In this paper, We will analyze this break point and will give a formula to find out this at various PVT corners. Knowledge about (trb) will restrict LUT simulations only in non-linear region and will help us in saving huge amount of time during LUT characterization.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.