Notes

The amount of pattern counts acquired for D2 and D4 architectures whilst using low power controllers are presented in Fig. 1. Results include fork scheme presented by the blue bars and two P-fork schemes - with two (red bars) and four (green bars) - configurations. It is worth pointing out that for each setup the fault coverage persists at the same level. Several types of registers, starting at the 6-bit and peaking at the 20-bit one, as well as toggling ratios ranging from 15% to 25%, were used during the research.
P-fork is superior compared to the conventional fork scheme regardless of its configuration type. This statement is clearly proven on the basis of pattern counts required for the consecutive register-toggle compositions. The highest level of efficiency is visible at diagram for the 12-bit register. The conventional fork scheme requires almost 6000 patterns whereas P-fork with two configurations is three times more efficient and P-fork scheme with four configurations is nearly six times profitable compared to the standard configuration. The inequality between conventional approach and the P-fork designs becomes less noticeable as the amount of bit registers grows, but the latter is always more preferable.
V Additional Findings
The research provides us with positive conclusions. Fig. 5 demonstrates the benefit of D2 (red bars) and D4 (green bars) designs - compared to the traditional scheme (blue bars) they require less amount of control register size to perform in an adequate manner.
--tutaj cos czego nie rozumiem
The P-fork scheme is a perfect solution for registers with low amount of bits. Fig. 6 shows compromise between the acquired test coverage and the amount of test patterns.