Predicting data correlation using multivalued logical outputs in static random access memory (SRAM) storage cells

• US 10,418,094 B2
• Filed: 02/08/2018
• Issued: 09/17/2019
• Est. Priority Date: 02/21/2017
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Erickson, Karl R., Paone, Phil C., Paulsen, David P., Sheets, II, John E., Uhlmann, Gregory J.
• Primary Examiner(s): Huang, Min
• Application Number:
  US15/892,037
• Publication Number:
  US 20180240512A1
• Time in Patent Office: 586 Days
• Field of Search: 365207
• CPC Class Codes: G11C 11/418 Address circuits
• CPC Class Codes: G11C 11/419 Read-write [R-W] circuits
• CPC Class Codes: G11C 2029/5004 Voltage
• CPC Class Codes: G11C 29/50 Marginal testing, e.g. race, voltage or current testing
• CPC Class Codes: G11C 7/12 Bit line control circuits, e.g. drivers, boosters, pull-up circuits, pull-down circuits, precharging circuits, equalising circuits, for bit lines

Abstract

Predicting data correlation using multivalued logical outputs in SRAM storage cells including generating a plurality of logical outputs for each of a plurality of variable sets, wherein each variable in each variable set is a data point, and wherein each logical output is a binary indication of a relationship between the data points; writing, into storage cells, each logical output of the plurality of logical outputs for each of the plurality of variable sets; and for each group of corresponding logical outputs of the plurality of logical outputs: activating a fight port for the storage cells storing corresponding logical outputs, wherein activating the fight port causes each corresponding logical output to adjust a resulting voltage based on the logical output stored in each storage cell; and measuring the resulting voltage on a bitline of the activated fight port to determine a correlation probability for the corresponding logical outputs.

Claims

1. A method of predicting data correlation using multivalued logical outputs in modified static random access memory (SRAM) storage cells comprising:

writing, into modified SRAM storage cells, a plurality of logical outputs for each of a plurality of variable sets, wherein each variable in each variable set is a data point, and wherein each logical output is a binary indication of a relationship between the data points of each variable set; and

for each group of corresponding logical outputs of the plurality of logical outputs;

activating a fight port for the modified SRAM storage cells storing corresponding logical outputs, wherein activating the fight port causes each corresponding logical output to adjust a resulting voltage based on the logical output stored in each modified SRAM storage cell, wherein the resulting voltage on a bitline of the activated fight port determines a correlation probability for the corresponding logical outputs.

2. The method of claim 1, wherein measuring the resulting voltage on the bitline of the activated fight port to determine a correlation probability for the corresponding logical outputs comprises:

transferring the resulting voltage to a sense amp; and

generating, by the sense amp, a correlation probability signal by comparing the resulting voltage to a voltage threshold.

3. The method of claim 2, wherein the correlation probability signal indicates a confidence level for the correlation probability.

4. The method of claim 1, wherein the each of the logical outputs are generated by applying a different logic operation to each of the plurality of variable sets.

5. The method of claim 1, wherein the fight port activates a plurality of storage cells on a bitline of the modified SRAM, and wherein the resulting voltage is applied to the bitline of the fight port.

6. The method of claim 1, wherein writing, into modified SRAM cells, each logical output of the plurality of logical outputs for each of the plurality of variable sets comprises:

writing each of the plurality of logical outputs for each variable set on a different modified SRAM storage cell along the same wordline of the modified SRAM.

7. The method of claim 1, wherein the fight port is operatively coupled to a group of modified SRAM storage cells on different wordlines.

8. An apparatus for predicting data correlation using multivalued logical outputs in modified static random access memory (SRAM) storage cells, the apparatus configured to carry out the steps of:

wherein each variable in each variable set is a data point, and wherein each logical output is a binary indication of a relationship between the data points of each variable set;

writing, into modified SRAM storage cells, a plurality of logical outputs for each of a plurality of variable sets, wherein each variable in each variable set is a data point, and wherein each logical output is a binary indication of a relationship between the data points of each variable set; and

for each group of corresponding logical outputs of the plurality of logical outputs;

activating a fight port for the modified SRAM storage cells storing corresponding logical outputs, wherein activating the fight port causes each corresponding logical output to adjust a resulting voltage based on the logical output stored in each modified SRAM storage cell, wherein the resulting voltage on a bitline of the activated fight port determines a correlation probability for the corresponding logical outputs.

9. The apparatus of claim 8, wherein measuring the resulting voltage on a bitline of the activated fight port to determine a correlation probability for the corresponding logical outputs comprises:

transferring the resulting voltage to a sense amp; and

generating, by the sense amp, a correlation probability signal by comparing the resulting voltage to a voltage threshold.

10. The apparatus of claim 9, wherein the correlation probability signal indicates a confidence level for the correlation probability.

11. The apparatus of claim 8, wherein the each of the logical outputs are generated by applying a different logic operation to each of the plurality of variable sets.

12. The apparatus of claim 8, wherein the fight port activates a plurality of storage cells on a bitline of the modified SRAM, and wherein the resulting voltage is applied to the bitline of the fight port.

13. The apparatus of claim 8, wherein writing, into modified SRAM cells, each logical output of the plurality of logical outputs for each of the plurality of variable sets comprises:

writing each of the plurality of logical outputs for each variable set on a different modified SRAM storage cell along the same wordline of the modified SRAM.

14. The apparatus of claim 8, wherein the fight port is operatively coupled to a group of modified SRAM storage cells on different wordlines.

15. A computer program product for predicting data correlation using multivalued logical outputs in modified static random access memory (SRAM) storage cells, the computer program product disposed upon a computer readable medium, the computer program product comprising computer program instructions that, when executed, cause a computer to carry out the steps of:

writing, into modified SRAM storage cells, a plurality of logical outputs for each of a plurality of variable sets, wherein each variable in each variable set is a data point, and wherein each logical output is a binary indication of a relationship between the data points of each variable set; and

for each group of corresponding logical outputs of the plurality of logical outputs;

activating a fight port for the modified SRAM storage cells storing corresponding logical outputs, wherein activating the fight port causes each corresponding logical output to adjust a resulting voltage based on the logical output stored in each modified SRAM storage cell, wherein the resulting voltage on a bitline of the activated fight port determines a correlation probability for the corresponding logical outputs.

16. The computer program product of claim 15, wherein measuring the resulting voltage on a bitline of the activated fight port to determine a correlation probability for the corresponding logical outputs comprises:

transferring the resulting voltage to a sense amp; and

generating, by the sense amp, a correlation probability signal by comparing the resulting voltage to a voltage threshold.

17. The computer program product of claim 16, wherein the correlation probability signal indicates a confidence level for the correlation probability.

18. The computer program product of claim 15, wherein the each of the logical outputs are generated by applying a different logic operation to each of the plurality of variable sets.

19. The computer program product of claim 15, wherein the fight port activates a plurality of storage cells on a bitline of the modified SRAM, and wherein the resulting voltage is applied to the bitline of the fight port.

20. The computer program product of claim 15, wherein writing, into modified SRAM cells, each logical output of the plurality of logical outputs for each of the plurality of variable sets comprises:

writing each of the plurality of logical outputs for each variable set on a different modified SRAM storage cell along the same wordline of the modified SRAM.