Double runs data coding method

Abstract

The invention discloses a double-run data encoding method. The invention is characterized in that the code length of the data encoder consists of three parts: a mark bit, a domain bit and a parity bit. The invention has the advantages that the appearance times of 1 or 0 in original binary data are not taken into account, and 1 and 0 can be processed at the same time for the coding of a testing vector; ideal compressibility and higher coding rate can be obtained in all situations; the increment of the compressibility is faster than that of other data coding methods, and the negative compression phenomenon of the testing data can also be avoided. Therefore, the invention is extremely suitable for the compression of the testing vector of a system chip and provided with strong practical value.

Description

Technical field: [0001] The invention relates to a double-run data encoding method in the field of microelectronics. Background technique: [0002] Current IP manufacturers will provide corresponding test vectors when providing IP cores. These test vectors are not only used in design simulation on workstations, but also used in wafer silicon testing. For a complex system chip, its test vector is very large. For example, in the case of a system chip integrated with ARM7TDMI, the test vectors provided by ARM have nearly 1.1 million cycles of test vectors. Even if parallel testing is used, it can only reduce a certain number of test cycles, and there are 150,000 clock cycles. But at the same time, more test channels are required. For a silicon wafer test process, its cost is directly proportional to the required test time and the number of test channels. The best way to solve the problem of expensive test equipment is to use built-in self-test. However, not all modules can ...

AI Technical Summary

Problems solved by technology

[0009] It can be seen from Table 1 that the shortcomings and limitations of the Golomb data encoding method are only suitable for the case where the number of 0 or 1 in the test vector has an absolute advantage, such as the case where 0 accounts for 80% or 1 accounts for 80%.

In the Golomb data encoding method, only when there are more 0s or 1s in the original data than the other, a better encoding rate will appear, and when the two are almost the same, there will be a phenomenon of negative growth in the encoding rate

Images