DPU architecture

1. A dynamic random access memory (DRAM) processing unit (DPU), comprising:

at least one computing cell array comprising a plurality of DRAM-based computing cells arranged in an array having at least a first predetermined number of columns and a second predetermined number of rows in which the first predetermined number is greater than or equal to three and the second predetermined number is greater than or equal to three, each column being configured to provide a DPU operation comprising a logic function that digitally operates on a first and a second row of the column and being configured to store a result of the logic function in a third row of the column, the at least one computing cell array further comprising a third predetermined number of shift lines in which the third predetermined number is two times the first predetermined number, each shift line being coupled to a column of computing cells through at least one corresponding first transistor, the shift lines and the corresponding first transistors being configured to shift contents of two rows of computing cells of a selected column at least two columns in a right direction or in a left direction in the at least one computing cell array; and

a controller coupled to the at least one computing cell array to configure the at least one computing cell array to perform the DPU operation.

2. The DPU of claim 1 , wherein the controller receives instructions for the DPU operation over an address bus.

3. The DPU of claim 1 , wherein the DRAM-based computing cells of at least one column each comprising a three transistor, one capacitor (3T1C) DRAM memory cell.

4. The DPU of claim 3 , wherein the DRAM-based computing cells of at least one column provide a NOR logic function.

5. The DPU of claim 1 , wherein the DRAM-based computing cells of at least one column each comprising a one transistor, one capacitor (1T1C) DRAM memory cell.

6. The DPU of claim 5 , wherein each DRAM-based computing cell further comprising an arithmetic logic unit (ALU) coupled to a bit line of the DRAM-based computing cell, the ALU providing the logic function.

7. The DPU of claim 6 , wherein the ALU provides a NOR logic function.

8. The DPU of claim 1 , further comprising:

at least one memory cell array comprising at least one DRAM-based memory cell arranged in the first predetermined number of columns, each column of DRAM-based memory cells of at least one data cell array corresponding to a column of a corresponding computing cell array; and

a sense amplifier coupled to each column of computing cells, each sense amplifier comprising an input that is electrically coupled to a read bit line of the computing cells of the column and an output that is electrically coupled to a write bit line of the computing cells of the column,

wherein a controller is further coupled to the at least one memory cell array to configure the at least one memory cell array to perform a memory operation, and

wherein the controller further receives instructions for the memory operation over an address bus.

9. The DPU of claim 1 , further comprising:

at least one stochastic computing cell array comprising a plurality of DRAM-based stochastic computing cells arranged in the first predetermined number of columns and the second predetermined number of rows, each column of DRAM-based stochastic computing cells of at least one data cell array corresponding to a column of a corresponding computing cell array, each column being configured to provide a stochastic logic function that operates on a first stream of data received by a first row and a second stream of data received by a second row of the column and configured to store a stream of data resulting from the stochastic logic function in a third row of the column,

wherein a controller is further coupled to the at least one stochastic computing cell array to configure the at least one stochastic computing cell array to perform a stochastic logic operation corresponding to the stochastic logic function, and

wherein the controller further receives instructions for the stochastic logic operation over an address bus.

10. A dynamic random access memory (DRAM) processing unit (DPU), comprising:

at least one computing cell array comprising a plurality of DRAM-based computing cells arranged in an array having at least a first predetermined number of columns and a second predetermined number of rows in which the first predetermined number is greater than or equal to three and the second predetermined number is greater than or equal to three, each column being configured to provide a logic function that digitally operates on a first row and a second row of the column and configured to store a result of the logic function in a third row of the column, the at least one computing cell array further comprising a third predetermined number of shift lines in which the third predetermined number is two times the first predetermined number, each shift line being coupled to a column of computing cells through at least one corresponding first transistor, the shift lines and the corresponding first transistors being configured to shift contents of two rows of computing cells of a selected column at least two columns in a right direction or in a left direction in the at least one computing cell array;

at least one data cell array comprising at least one DRAM-based memory cell arranged in the first predetermined number of columns, each column of DRAM-based memory cells of at least one data cell array corresponding to a column of a corresponding computing cell array; and

a controller coupled to the at least one computing cell array to configure the at least one computing cell array to perform a DPU operation and coupled to the at least one data cell array to perform a memory operation.

11. The DPU of claim 10 , wherein the controller receives instructions for the DPU operation over an address bus.

12. The DPU of claim 10 , wherein the DRAM-based computing cells of at least one column each comprising a three transistor, one capacitor (3T1C) DRAM memory cell, and

wherein the DRAM-based computing cells of at least one column provide a NOR logic function.

13. The DPU of claim 10 , wherein the DRAM-based computing cells of at least one column each comprising a one transistor, one capacitor (1T1C) DRAM memory cell, and

wherein each DRAM-based computing cell further comprising an arithmetic logic unit (ALU) coupled to a bit line of the DRAM-based computing cell, the ALU providing the logic function.

14. The DPU of claim 13 , wherein the ALU provides a NOR logic function.

15. The DPU of claim 10 , further comprising:

at least one stochastic computing cell array comprising a plurality of DRAM-based stochastic computing cells arranged in the first predetermined number of columns and the second predetermined number of rows, each column of DRAM-based stochastic computing cells of at least one data cell array corresponding to a column of a corresponding computing cell array, each column being configured to provide a stochastic logic function that operates on a first stream of data received by a first row and a second stream of data received by a second row of the column and configured to store a stream of data resulting from of the stochastic logic function in a third row of the column,

wherein a controller is further coupled to the at least one stochastic computing cell array to configure the at least one stochastic computing cell array to perform a stochastic logic operation, and

wherein the controller further receives instructions for the stochastic logic operation over an address bus.

16. The DPU of claim 15 , wherein the DRAM-based stochastic computing cells of at least one column each comprise a three transistor, one capacitor (3T1C) DRAM memory cell, or a one transistor, one capacitor (1T1C) DRAM memory cell.

17. A dynamic random access memory (DRAM) processing unit (DPU), comprising:

at least one computing cell array comprising a plurality of DRAM-based computing cells arranged in an array having at least a first predetermined number of columns and a second predetermined number of rows in which the first predetermined number is greater than or equal to three and the second predetermined number is greater than or equal to three, each column being configured to provide a logic function that digitally operates on a first row and a second row of the column and configured to store a result of the logic function in a third row of the column, the at least one computing cell array further comprising a third predetermined number of shift lines in which the third predetermined number is two times the first predetermined number, each shift line being coupled to a column of computing cells through at least one corresponding first transistor, the shift lines and the corresponding first transistors being configured to shift contents of two rows of computing cells of a selected column at least two columns in a right direction or in a left direction in the at least one computing cell array;

at least one stochastic computing cell array comprising a plurality of DRAM-based stochastic computing cells arranged in the first predetermined number of columns and the second predetermined number of rows, each column of DRAM-based stochastic computing cells of at least one data cell array corresponding to a column of a corresponding computing cell array, each column being configured to provide a stochastic logic function that operates on a first stream of data received by a first row and second stream of data received by a second row of the column and configured to store a stream of data resulting from the stochastic logic function in a third row of the column; and

a controller coupled to the at least one computing cell array to configure the at least one computing cell array to perform a DPU operation and coupled to the at least one stochastic computing cell array to perform a stochastic logic operation corresponding to the stochastic logic function.

18. The DPU of claim 17 , wherein the controller receives instructions for the DPU operation over an address bus.

19. The DPU of claim 18 , wherein the DRAM-based computing cells of at least one column each comprising a three transistor, one capacitor (3T1C) DRAM memory cell, and

wherein the DRAM-based computing cells of at least one column provide a NOR logic function.

20. The DPU of claim 17 , wherein the DRAM-based stochastic computing cells of at least one column each comprise a three transistor, one capacitor (3T1C) DRAM memory cell.