US2018341623A1PendingUtilityA1

Matrix circuits

Assignee: HEWLETT PACKARD ENTPR DEV LPPriority: Apr 28, 2017Filed: Aug 1, 2018Published: Nov 29, 2018
Est. expiryApr 28, 2037(~10.7 yrs left)· nominal 20-yr term from priority
G11C 13/0069G06F 17/16G11C 13/004G11C 11/4096G11C 11/24G11C 8/16G11C 2213/74G11C 11/4087G11C 7/1006G11C 13/0023G11C 11/4093G11C 2213/79G11C 13/003
45
PatentIndex Score
0
Cited by
0
References
0
Claims

Abstract

A circuit is provided. In an example, the circuit includes a memory array that includes a plurality of memory cells to store a matrix and a plurality of data lines coupled to the plurality of memory cells to provide a first set of values of the matrix. The circuit includes a multiplier coupled to the plurality of data lines to multiply the first set of values by a second set of values to produce a third set of values. A summing unit is included that is coupled to the multiplier to sum the third set of values to produce a sum. The circuit includes a shifting unit coupled to the summing unit to shift the sum and to add the shifted sum to a running total.

Claims

exact text as granted — not AI-modified
What is claimed is: 
     
         1 . A circuit comprising:
 a memory array including:
 a plurality of memory cells to store a matrix; and 
 a plurality of data lines coupled to the plurality of memory cells to provide a first set of values of the matrix; 
   a multiplier coupled to the plurality of data lines to multiply the first set of values by a second set of values to produce a third set of values;   a summing unit coupled to the multiplier to sum the third set of values to produce a sum; and   a shifting unit coupled to the summing unit to shift the sum and to add the shifted sum to a running total.   
     
     
         2 . The circuit of  claim 1 , wherein each of the plurality of memory cells includes:
 a capacitor;   an access transistor coupled to the capacitor to program the capacitor; and   a pass transistor having a gate coupled to the capacitor to read a value of the matrix that is stored in the capacitor, wherein the value is provided to a corresponding data line of the plurality of data lines.   
     
     
         3 . The circuit of  claim 2 , wherein the access transistor of each of the plurality of memory cells is coupled between the capacitor and a column write line to program the memory array a row at a time. 
     
     
         4 . The circuit of  claim 3 , wherein the pass transistor of each of the plurality of memory cells is coupled to a row read line to read the memory array a column at a time. 
     
     
         5 . The circuit of  claim 1 , wherein each of the plurality of memory cells includes:
 a resistive memory device;   an access transistor coupled to the resistive memory device to program the resistive memory device; and   a pass transistor coupled to the resistive memory device to read a value of the matrix that is encoded in the resistive memory device, wherein the value is provided to a corresponding data line of the plurality of data lines.   
     
     
         6 . The circuit of  claim 1 , wherein the memory array, the multiplier, the summing unit, and the shifting unit are to multiply the matrix by a vector that includes the second set of values. 
     
     
         7 . The circuit of  claim 1 , wherein the matrix includes a plurality of multi-bit values, and wherein the first set of values of the matrix have a common place value. 
     
     
         8 . The circuit of  claim 1 , wherein the shifting unit is to shift the sum based on a place value of the first set of values or a place value of the second set of values. 
     
     
         9 . A circuit comprising:
 a memory array to store a matrix that includes a plurality of multi-bit values, wherein the plurality of multi-bit values includes a plurality of matrix sets, and wherein each set of the plurality of matrix sets contains values having a common place value; and   a matrix processing unit coupled to the memory array and including logic to, for each set of the plurality of matrix sets:
 receive the respective set; 
 multiply the respective set by a vector set; 
 sum each element of the product of the respective set and the vector set to produce a total; 
 shift the total; and 
 add the shifted total to a running total. 
   
     
     
         10 . The circuit of  claim 9 , wherein the memory array stores the matrix in a plurality of memory cells, and wherein each memory cell includes:
 a capacitor;   an access transistor coupled to the capacitor to program the capacitor; and   a pass transistor having a gate coupled to the capacitor to read a value of the matrix that is stored in the capacitor.   
     
     
         11 . The circuit of  claim 10 , wherein the access transistor of each of the plurality of memory cells is coupled between the capacitor and a column line to program the memory array a row at a time. 
     
     
         12 . The circuit of  claim 11 , wherein the pass transistor of each of the plurality of memory cells is coupled to a row line to read the memory array a column at a time. 
     
     
         13 . The circuit of  claim 9 , wherein the memory array stores the matrix in a plurality of memory cells, and wherein each of the plurality of memory cells includes:
 a resistive memory device;   an access transistor coupled to the resistive memory device to program the resistive memory device; and   a pass transistor coupled to the resistive memory device to read a value of the matrix that is encoded in the resistive memory device.   
     
     
         14 . The circuit of  claim 9 , wherein the vector multiplication unit is to shift the total based on a place value of the respective set of the plurality of matrix sets or a place value of the vector set. 
     
     
         15 . A method comprising:
 reading a first set of values of a matrix from a memory array;   multiplying the first set of values of the matrix by a second set of values of a vector to provide a third set of values;   summing the third set of values to produce a sum;   shifting the sum based on a place value of the first set of values or a place value of the second set of values; and   adding the shifted sum to a running total.   
     
     
         16 . The method of  claim 15  comprising writing the matrix to the memory array a row at a time. 
     
     
         17 . The method of  claim 16 , wherein the reading of the first set of values reads from the memory array a column at a time. 
     
     
         18 . The method of  claim 15 , wherein the matrix includes a plurality of multi-bit values, and wherein the first set of values of the matrix have a common place value. 
     
     
         19 . The method of  claim 15 , wherein the memory array includes a plurality of capacitors that store the matrix, and wherein the reading of the first set of values retains the first set of values within the plurality of capacitors. 
     
     
         20 . The method of  claim 15 , wherein the memory array includes a plurality of resistive memory devices that store the matrix.

Join the waitlist — get patent alerts

Track US2018341623A1 — get alerts on status changes and closely related new filings.

We store only your email — no account needed. See our privacy policy.