公开(公告)号：US10930365B2

公开(公告)日：2021-02-23

申请号：US16281559

申请日：2019-02-21

Applicant: Intel Corporation

Inventor： Pavel Poliakov ,  Andrey Kudryavtsev ,  Shekoufeh Qawami ,  Amirali Khatib Zadeh ,  Monte Klinkenborg

IPC: G11C29/00 ,  G11C29/38 ,  G11C29/44 ,  G06N3/08 ,  G06N3/04 ,  G11C11/54 ,  G11C16/34 ,  G11C29/56 ,  G11C29/04

Abstract: In embodiments, a memory controller (MC) includes an output interface, and an execution engine (EE) to identify, based on field test results of a die coupled to the MC, initial test results of the die using an artificial neural network (ANN) trained to identify the die from a set of NVM dies based on initial test results of the set of NVM dies obtained at a time of manufacture of the set of dies. The initial test results include a first useful life prediction and the field test results include a second useful life prediction, and the initial test results are regenerated by the ANN to protect their confidentiality. In embodiments, the MC is further to compare the second useful life prediction with the first useful life prediction, to determine a deviation between the two, and output, via the output interface, the deviation to a user.

公开(公告)号：US20190189236A1

公开(公告)日：2019-06-20

申请号：US16281559

申请日：2019-02-21

Applicant: Intel Corporation

Inventor： Pavel Poliakov ,  Andrey Kudryavtsev ,  Shekoufeh Qawami ,  Amirali Khatib Zadeh ,  Monte Klinkenborg

IPC: G11C29/38 ,  G06N3/04 ,  G06N3/08 ,  G11C29/44

CPC classification number: G11C29/38 ,  G06N3/04 ,  G06N3/08 ,  G11C29/44

Abstract: In embodiments, a memory controller (MC) includes an output interface, and an execution engine (EE) to identify, based on field test results of a die coupled to the MC, initial test results of the die using an artificial neural network (ANN) trained to identify the die from a set of NVM dies based on initial test results of the set of NVM dies obtained at a time of manufacture of the set of dies. The initial test results include a first useful life prediction and the field test results include a second useful life prediction, and the initial test results are regenerated by the ANN to protect their confidentiality. In embodiments, the MC is further to compare the second useful life prediction with the first useful life prediction, to determine a deviation between the two, and output, via the output interface, the deviation to a user.