Auto-calibrating crossbar-based apparatuses

1. A method, comprising:

obtaining output data of a crossbar-based apparatus comprising a plurality of cross-point devices with tunable conductance, wherein the output data of the crossbar-based apparatus represents computing results of at least one operation performed by the crossbar-based apparatus, and wherein the output data corresponding to a plurality of settings of a plurality of analog components of the crossbar-based apparatus;

obtaining, by a processing device, one or more calibration parameters based on the output data of the crossbar-based apparatus, wherein the one or more calibration parameters correspond to one or more errors associated with one or more of the analog components of the crossbar-based apparatus; and

calibrating the crossbar-based apparatus using the one or more calibration parameters to compensate for deviations of the computing results of the at least one operation performed by the crossbar-based apparatus from expected results of the at least one operation.

2. The method of claim 1 , wherein the output data of the crossbar-based apparatus corresponds to a conductance pattern of the cross-point devices, wherein the conductance pattern comprises conductance values of the cross-point devices.

3. The method of claim 1 , wherein obtaining, by the processing device, the one or more calibration parameters comprises:

determining a plurality of error parameters approximating the errors associated with the one or more of the analog components of the crossbar-based apparatus; and

generating the one or more calibration parameters based on the error parameters.

4. The method of claim 3 , wherein determining the plurality of error parameters approximating the errors associated with the one or more of the analog components of the crossbar-based apparatus comprises:

generating simulated output data based on an error model; and

minimizing a difference between the output data and the simulated output data.

5. The method of claim 4 , wherein generating the one or more calibration parameters based on the error parameters further comprises:

calibrating the simulated output data to generate calibrated simulated output data; and

minimizing a difference between the calibrated simulated output data and expected output data representative of the expected results of the at least one operation performed by the crossbar-based apparatus.

6. The method of claim 5 , further comprising:

generating the simulated output data by processing input data using the error model, wherein the input data comprises the plurality of settings of the plurality of analog components of the crossbar-based apparatus.

7. The method of claim 6 , further comprising:

generating the expected output data by processing the input data using a system model.

8. The method of claim 6 , wherein the input data further comprises a calibrated conductance pattern.

9. The method of claim 1 , wherein the at least one operation comprises at least one of a read operation, a VMM operation, or a neural processing operation.

10. The method of claim 1 , wherein the one or more errors comprise at least one of a voltage offset or a gain error.

11. The method of claim 10 , wherein the plurality of analog components of the crossbar-based apparatus comprises at least one of an op-amp, an ADC, a DAC, or a TIA.

12. The method of claim 1 , wherein calibrating the crossbar-based apparatus comprises: storing the calibration parameters in one or more circuit registers of the crossbar-based apparatus.

13. A system comprising:

a memory; and

a processing device operatively coupled to the memory, the processing device to:

obtain output data of a crossbar-based apparatus comprising a plurality of cross-point devices with tunable conductance, wherein the output data of the crossbar-based apparatus represents computing results of at least one operation performed by the crossbar-based apparatus, and wherein the output data corresponding to a plurality of settings of a plurality of analog components of the crossbar-based apparatus;

obtain, by the processing device, one or more calibration parameters based on the output data of the crossbar-based apparatus, wherein the one or more calibration parameters correspond to one or more errors associated with one or more of the analog components of the crossbar-based apparatus; and

calibrate the crossbar-based apparatus using the one or more calibration parameters to compensate for deviations of the computing results of the at least one operation performed by the crossbar-based apparatus from expected results of the at least one operation.

14. The system of claim 13 , wherein the one or more errors comprise at least one of a voltage offset or a gain error, and wherein the plurality of analog components of the crossbar-based apparatus comprises at least one of an op-amp, an ADC, a DAC, or a TIA.

15. The system of claim 13 , wherein, to obtain the one or more calibration parameters, the processing device is further to:

determine a plurality of error parameters approximating the errors associated with the one or more of the analog components of the crossbar-based apparatus; and

generate the one or more calibration parameters based on the error parameters.

16. The system of claim 15 , wherein, to determine the plurality of error parameters approximating the errors associated with the one or more of the analog components of the crossbar-based apparatus, the processing device is further to:

generate simulated output data based on an error model; and

minimize a difference between the output data and the simulated output data.

17. The system of claim 16 , wherein, to generate the one or more calibration parameters based on the error parameters, the processing device is further to:

calibrate the simulated output data to generate calibrated simulated output data; and

minimize a difference between the calibrated simulated output and expected output data representative of the expected results of the at least one operation performed by the crossbar-based apparatus.

18. The system of claim 17 , wherein the processing device is further to:

generate the simulated output data by processing input data using the error model, wherein the input data comprises the plurality of settings of the plurality of analog components of the crossbar-based apparatus.

19. The system of claim 13 , wherein the at least one operation comprises at least one of a read operation, a VMM operation, or a neural processing operation.

20. A non-transitory machine-readable storage medium including instructions that, when accessed by a processing device, cause the processing device to:

obtain output data of a crossbar-based apparatus comprising a plurality of cross-point devices with tunable conductance, wherein the output data of the crossbar-based apparatus represents computing results of at least one operation performed by the crossbar-based apparatus, and wherein the output data corresponding to a plurality of settings of a plurality of analog components of the crossbar-based apparatus;

obtain, by the processing device, one or more calibration parameters based on the output data of the crossbar-based apparatus, wherein the one or more calibration parameters correspond to one or more errors associated with one or more of the analog components of the crossbar-based apparatus; and

calibrate the crossbar-based apparatus using the one or more calibration parameters to compensate for deviations of the computing results of the at least one operation performed by the crossbar-based apparatus from expected results of the at least one operation.