Design analysis tool and method for deriving correspondence between storage elements of two memory models
View Patent ↗A method and system automatically generates a bit-cell correspondence between a first memory model and a second memory model of a memory. The method includes receiving data from the first and the second memory model, obtaining true-inverted fan-in cones for words in the memory models to obtain correspondence between sets of words in the two models, writing word binary sequences into the words to obtain a set of bit-cell correspondences, and using inherent structural information in memory designs to generalize bit-cell correspondence obtained on bit-cells of a pair of corresponding words to obtain bit-cell correspondence information for all the bit-cells in the memory models. Correspondence is detected if one of the bit-cell binary sequences written into a bit-cell in the first memory model is equal to or an invert of another of the bit-cell binary sequences written into a bit-cell in the second memory model.
1. A method for automatically generating a bit-cell correspondence between a first memory model and at least a second memory model, the method comprising:
receiving data from the first and second memory model;
receiving primary input and output correspondence for the first and second memory model;
identifying word correspondence between the first and second memory model by comparing the true-inverted fan-in cones of words in the first and second memory models; and
identifying bit-cells in one or more pairs of corresponding words to generate the bit-cell correspondence;
wherein the first and second memory models contain k sections, where k≧1, and further comprising:
if a priori knowledge concerning the value of k is known, then using k to determine bit-cell correspondences for the first and second memory models.
2. The method of claim 1 further comprising generalizing bit-cell correspondence obtained on the bit-cells of the one or more pairs of corresponding words to obtain bit-cell correspondence for bit-cells in the first and the second memory models.
3. The method of claim 1 wherein bit-cell correspondence includes correspondences that are in a true sense and an inverted sense.
4. The method of claim 1 wherein the identifying bit-cells includes:
choosing an arbitrary address associated with the first and second memory models;
writing a plurality of word binary sequences into a first word addressed by the arbitrary address in the first memory model;
writing the plurality of word binary sequences into a second word addressed by the same arbitrary address in the second memory model; and
obtaining a set of bit-cell correspondences for the bit-cells of the first word in the first memory model and the second word in the second memory model.
5. The method of claim 4 wherein the plurality of word binary sequences is a predetermined number of patterns, and wherein the number of patterns is determined using a function including a logarithm to the base 2 of the number of bit-cells in a word.
6. The method of claim 4 wherein a correspondence is detected if one of a plurality of bit-cell binary sequences resulting from the writing of the plurality of word binary sequences written into a bit-cell of the first memory model is either equal to, or is equal to the invert of, another of the plurality of bit-cell binary sequences written into a bit-cell in the second memory model for the corresponding words.
7. The method of claim 4 wherein the plurality of word binary sequences results in bit-cell binary sequences that are unique and are not an invert of another bit-cell binary sequence.
8. The method of claim 1 wherein each of the first and second memory models include one or more of a circuit level memory model, a register transfer language (RTL) memory model, a lowest level memory model, and a highest level memory model.
9. The method of claim 1 further comprising:
if the memory has one section, then determining the bit-cell correspondence for a pair of corresponding words; and
generalizing results obtained to all bit-cells of the memory models.
10. The method of claim 1 further comprising:
if the memory has multiple sections, then for each section finding a bit-cell correspondence for at least one pair of corresponding words; and
generalizing results obtained to all bit-cells of the section.
11. The method of claim 1 wherein if the memory has no data manipulation, then logic bit-cell correspondence is found using structural analysis of the bit-cells of corresponding words.
12. The method of claim 1 further comprising:
determining whether the memory represented by the first and second memory model is divided into different sections such that each individual section has each of the plurality of data lines connected with bit-cells having data lines that do not change with different words across a given memory column.
13. The method of claim 12 wherein the determining includes performing random writes into the first and second memory models to discover whether more than one section in the memory exists.
14. The method of claim 1 wherein identifying word correspondence includes obtaining a plurality of true-inverted fan-in cone sets by making a distinction between each address line and its complement in determining a dependence for each word.
15. The method of claim 14 wherein the distinction between each address line and its complement enables determining an address that corresponds to each word in the two memory models.
16. A method for automatically generating a bit-cell correspondence between first and second memory models which each contain k sections, where k≧1, the method comprising:
receiving data from the first and second memory models;
receiving primary input and output correspondence for the first and second memory models;
identifying word correspondence between the first and second memory models;
identifying bit-cells in one or more pairs of corresponding words to generate the bit-cell correspondence;
if the memory has multiple sections, then for each section finding a bit-cell correspondence for at least one pair of corresponding words; and
generalizing results obtained to all bit-cells of the section.
17. The method of claim 16 wherein, if a priori knowledge concerning the value of k is known, then using k to determine bit-cell correspondences for the first and second memory models.
18. The method of claim 16 wherein bit-cell correspondence includes correspondences that are in a true sense and an inverted sense.
19. The method of claim 16 wherein the identifying bit-cells includes:
choosing an arbitrary address associated with the first and second memory models;
writing a plurality of word binary sequences into a first word addressed by the arbitrary address in the first memory model;
writing the plurality of word binary sequences into a second word addressed by the same arbitrary address in the second memory model; and
obtaining a set of bit-cell correspondences for the bit-cells of the first word in the first memory model and the second word in the second memory model.
20. A method for automatically generating a bit-cell correspondence between first and second memory models, the method comprising:
receiving data from the first and second memory models;
receiving primary input and output correspondence for the first and second memory models;
identifying word correspondence between the first and second memory models; and
identifying bit-cells in one or more pairs of corresponding words by writing a plurality of word binary sequences into the first and second memory models, thereby generating a bit-cell correspondence;
wherein the plurality of word binary sequences is a predetermined number of patterns, and
wherein the number of patterns is determined using a function including a logarithm to the base 2 of the number of bit-cells in a word.
21. The method of claim 20 , wherein the step of identifying bit-cells further includes:
choosing an arbitrary address associated with the first and second memory models;
writing the plurality of word binary sequences into a first word addressed by the arbitrary address in the first memory model;
writing the plurality of word binary sequences into a second word addressed by the same arbitrary address in the second memory model; and
obtaining a set of bit-cell correspondences for the bit-cells of the first word in the first memory model and the second word in the second memory model.
22. The method of claim 20 wherein bit-cell correspondence includes correspondences that are in a true sense and an inverted sense.
23. A method for automatically generating a bit-cell correspondence between first and second memory models, the method comprising:
receiving data from the first and second memory model;
receiving primary input and output correspondence for the first and second memory model;
identifying word correspondence between the first and second memory model;
identifying bit-cells in one or more pairs of corresponding words to generate the bit-cell correspondence; and
determining whether the memory represented by the first and second memory model is divided into different sections such that each individual section has each of the plurality of data lines connected with bit-cells having data lines that do not change with different words across a given memory column.
24. The method of claim 23 , wherein the step of determining includes performing random writes into the first and second memory models to discover whether more than one section in the memory exists.
25. The method of claim 23 wherein bit-cell correspondence includes correspondences that are in a true sense and an inverted sense.
26. The method of claim 23 wherein the identifying bit-cells includes:
choosing an arbitrary address associated with the first and second memory models;
writing a plurality of word binary sequences into a first word addressed by the arbitrary address in the first memory model;
writing the plurality of word binary sequences into a second word addressed by the same arbitrary address in the second memory model; and
obtaining a set of bit-cell correspondences for the bit-cells of the first word in the first memory model and the second word in the second memory model.